@vargonian 100% agree. There's a lot of prejudice about AI because many people let it do their work or thinking. I predict that people who use AI to learn a skill will have an edge over people who lable it as "evil."
@@YourComputer I actually find it extremely useful for software engineering even without it completing any code for me (which I disable anyway; I find those tools too intrusive). I regularly ask it for advice on how to approach modeling different systems and their relationships, which design patterns apply most appropriately to the situation, etc., and it often introduces me to context-specific concepts that I wasn't previously familiar with. For example, the other day, without going too far into specifics, I was trying to design an efficient pattern-matcher for words and it introduced me to a data structure that makes searches dramatically faster than what I was doing previously, as well as formalizing concepts in my mind like the "most-constraining-variable heuristic". If I were exploring an unfamiliar field like Big Data or something, it would be extremely valuable to learn from. I used it extensively a short while ago to learn Docker beyond the basics, for example. And the real power, in my opinion, is being able to ask it "why" instead of just "what" to really solidify your knowledge.
I've "learned" how to do projects in python and React (and some other tools). At my age there was no possible way I was going to truly learn those languages. But with AI I can now use them at an expert level. I guess one downside is if I say I know how to use Python and someone says "oh great how would you select all the primes up to one million" and without chatgpt I'm like, "uh, import "math" or something" . The reality is I know how to use AI, not the programming tools. But as I noted earlier I was never going to learn them so for me it feels like a big advance. If anything, AI causes me to think more. However, I am not so sure that would be true if I were a kid growing up with AI and just offloaded all my thinking to it.
This whole AI impact on our brains thing is a similar one to the introduction of cars and other vehicles. They help get us to places much quicker than by foot - yet the benefit of walking isn't simply to get us from one place to another, it makes us move our body, our heart pump and our blood flow and many other important bodily processes, that if we were to stop moving, eventually we would develop physical health problems, and perhaps the day we need to walk again we are no longer able to like we could before.
The example is similar but different, in the sense that it is a parallel at the physical-material level. In the past there was something similar, more striking and more subtle that is directly correlated to AI Around 1995-1996 Intellisense was created, a tool that facilitated access to object properties, which turned out to be a tremendously useful tool and was quickly adopted by every self-respecting language. This made developers rely on this tool, leaving behind any conventional editor. What was overlooked is that this tool lowered the effort needed to be able to program. It was no longer necessary to inspect objects, properties, methods, the tool did it for us. Jumping a little further, raise your hand if you did NOT use stack overflow to solve particular (and not so particular) programming problems. It was no longer necessary to solve a problem for us, it was enough to google it. AI is a new level of the same: making our lives 'easier'. What we have already figured out is that this has a cost on our quality of life that we need to control. Part of the video contains a good proposal, although I consider it incomplete. Internal values and discipline are lacking; recognizing that 'making our lives easier' can end up harming us in the long term. Like the car example, it will be necessary to recognize that even if we have the best car in the world, we will have to walk (and very often) to maintain our quality of life.
@@TheNefastor That's sort of the same here. I mean... you can always do simple math with the calculator without knowing math (e.g. adding numbers), but you're not going to recognize it then when you misclick, or you get an overflow, or any other sort of issue. You need to know so that you know when your tool is wrong. And it's the same with AI... it's not dependable, you have to already know. And once you already know... the AI just isn't that valuable, it's more of an annoying liability just like a noob you have to teach, but it doesn't improve whereas the noob can.
I'm a junior dev at the start of my career, and I'm really thankful for hearing this message. I will try my best not to kill my brain and career with AI. Thank you.
Excellent! Aside from improving your job security, that perspective will open doors to more interesting opportunities. You’ll develop the reputation of an out-of-the-box thinker, rather than someone waiting to be told what to do next.
Do not listen to this guy. Learn how to use the tools to speed up your coding. Otherwise you will be replaced by the people who know how to use the tools for coding. Just like this guy will be. Telling people not to use ai is like telling people not to use prebuilt libraries because you should build your own because you wont understand the underlying low level code. Your worth as a developer is not going to be your ability to write code as it is today it's going to be your ability to guide the ai. You going need to break down the problem into smaller parts, keep the ai focused on the issue because it has this tendency to do to much in one go. Don't listen to this guy he is going to lead you down the wrong path. In fact focus on how to build easy to use apps. Easy to use for your users, not you or another coder. And learn how to figure out what the core issue a user needs solved and give them that in app form.
Not only will you be replaced by people who are using the new tools it is imperative to understand the new food chain. At the bottom, you have the consumers they are the ones that use large language, models, and pre-artificial general intelligence. Then you move one level up and you understand there are people using the available open source models, and the API based proprietary ones to build custom Integration solutions. You could call these people the AI integrators. From there, it gets more complex the higher end of the AI integrators are using agent architectures and coming up with novel solutions for limits to context as well as relevancy search using vector databases. Also, at the stratified higher end of the AI integration echelons, you will begin to see developers, fine-tuning models that is to say augmenting the training of an existing model with additional training data to steer that model towards particular specialized domains. Finally, at the very top of the AI food chain are the machine learning researchers that are implementing as well as writing scientific papers in the cutting edge of model architecture and development. If you want to remain relevant, your goal as a software engineer is to recognize this food chain and begin moving up that food chain as fast as possible. If you remain a consumer of the cognitive service of these AI solutions, then you are on the road to replacement.
@@jeffsteyn7174 I'm going to add my two cents as an IT manager. My most productive people are the ones who embrace AI, but have done so AFTER learning the basics, and understanding the technology they're working with. They're the people who could build the solution anyway, but can build it faster with AI. Someone who starts out with AI tools isn't going to have the grounding to know when the AI is wrong, or fix the problems it creates. Also, people should absolutely implement low level libraries from scratch as part of their learning journey. Maybe you don't use your home-grown version in production, but it's very limiting to be completely reliant on high-level tools for everything all the time.
I think this isn’t an AI problem but a lack of learning to think logically and slow .. which is also a lack of reading books, articles, or documentation at all .. also not wanting to feel the discomfort of a hard problem. This effect is very real and something I’ve struggled with as a Sr dev and now manager with Jr and even Mid level devs.
I'm not a dev at all but I've often wondered if people pushing themselves to solve so many problems 'in real time' is a bad idea. And it probably doesn't help that online culture seems to give the impression that everyone is writing code the way they do in movies - moving their fingers nonstop as they talk out the problem in a matter of seconds and when they stop the program is done and works perfectly. I would think that slowing down and actually thinking through the whole process and then once you have a working model, verifying that with tests and line-by-line walk throughs would help in more ways than one. But like I said, I'm not a dev. So I might just be spitting BS at this point.
The biggest thing I try to do and the Leads when teaching how to think and analyze is to have them 1. State the problem clearly 2. Explain at least 2 solutions to the problem Many don’t know to do this and try to just solve stuff without even knowing the problem The next step is then to explain the tradeoffs of the problem, which is tricky and often requires more experienced coders to help .. but if they follow then they’re learning I think in part it’s the google it mindset , some also is a lack of critical thinking and analysis skills. A lot of coding also has become more abstracted so I notice also many developers don’t get the fundamentals ..’which would be like knowing calculus but not really understanding say the trig formulas used or having strong command over algebra. There does seem to be so many more distractions these days and say sitting down for a month and doing the basics seems a big ask. Even if there’s time given at work. This I think is a symptom of our smart phone distraction culture I was a literature major before switching to tech and noticed I couldn’t focus like I used to I realized it was mostly not reading paper books where I can’t get distracted ..so when I need to really learn new concepts even in tech I’ll often get a paper copy of a book so I can focus then I’ll go do hands on stuff So learning how to learn and finding a way to learn challenging concepts versus watching an easy to consume video is part of the challenge too
@@GuitarWithBrett I get that. I find videos just make my eyes glaze over and I learn nothing. But I have loads of old programming books that I learned with. Something about actually sitting in a quiet room and reading actually made my brain work through the entire line of reasoning from the writer and I often found I was getting way ahead of the material as I started to grasp what possibilities were available to me.
@ yeah, it’s hard … really you need the right next level step ideally and a real world problem .. for example I learned to setup background queues as the sr architect asked me to set that up.. I tried , got it wrong , then he explained why was wrong , then I redid .. then after that reading books on queues made total sense because I had the problem and also got stuck I’m also super selective on books I’ll read and I don’t read them straight through So right now I need to remember the graphql pipeline and how the middleware works .. I found a great book that explains that years ago for my language and graphql implementation.. so I’ll reread that part because we need to do stuff with it next week What are you wanting to learn ?
@@GuitarWithBrett Sounds like you were lucky to have a Sr that could take the time to explain something to you. I gather from channels like this that everyone is kinda expected to just come out of the womb fully-formed as a full-stack developer or something these days. As for me, I've never worried about what I want to learn and instead focused on what I wanted to do. The learning seemed to come from that, I guess. Though it also leaves huge gaping holes of knowledge too. I'm currently making a video game that is deliberately going for an early 90s DOOM-like aesthetic. I want to utilize the flat, billboarded, sprites for characters but I don't have time to draw every frame for every character and I want to use more modern tools like motion retargetting and modular character outfits. This lead me down a rabbit hole of creating a system to pre-render 3D characters as sprites in real time. But it's too slow for what I want. I can only get a few hundred on screen at once and I'd liked to aim for 1k+ so I'm going to have to get creative about it. I have some ideas but time's been short lately and I don't think I'll have the opportunity to try them all so I'll have to take a guess at the one that most likely will succeed and go from there.
Recently during a debugging session with a junior dev (us trying to reproduce a bug), I asked them to add another item to an existing array list; and they started typing out a comment for copilot to do it. To 'effin add an item to a simple array list! I was like c'mon! Please don't do this.
I've never met a junior who could do better than ChatGPT. It will lead to a huge shortage of senior devs. Once AI has cannibalized all code and can't get better, we seniors in our 30s and 40s and 50s will be in super demand in the 2030s. Watch.
Soon they wont just destroy junior devs. They will destroy any senior dev too. Now, a person who can't code for nuts, will be able to code just as well as a coder who spent 20 years of his life learning. It's amazing. What a perfect opportunity to get ahead of the curve.
Yes. It is a very dangerous tool if they are allowed to use it. Hell, it can even ruin a senior if he relies too much on it. Thinking is a skill that needs to be constantly trained. And AI is very very tempting. You can spend 4 hours thinking about something or asking AI and not thinking at all. It's a real danger.
@@Ragnar452 using the O1 GPT model or Claude, I almost always find that they cannot solve my problem so i end up having to think. After a certain level of complexity they turn useless
As another commenter put it >Ok, so I am using sand that has been processed for 4-5 months in a remote factory in taiwan >using machines made in amsterdam >shipped 2-3 months via evergreen >stockpiled in some warehouse I never see >delivered to me via a truck that runs on dinosaur bones >in a box made of dead trees >installed in a motherboard that followed a completely different path, made of completely different materials, lithium for the 2032 battery from cancun, the gold for the circuits is mined from zambia, the fan is designed by some random in uruguay that had the credit stolen by him from someone in california that had them manufactured in china >running an OS made by some swedish or norse or danish guy idk >maintained by thousands of people >stolen and re-done by some fruitarian in Palo Alto that died of liver cancer >running on a mix of electricity from uranium mined in Ural Mountains and hydroelectricity from a dam, and geothermal powerplants and wind turbines >getting information to it over long glass wires that reflect light with almost 0 reflectivity loss >displaying it on an array of millions of pixels >and all I do is string together API calls that were written by people long dead before I was born BUT GOD DAMN IT I WILL NOT USE AI CUZ THAT'S ONE ABSTRACTION LAYER TOO MUCH
@@leoym1803And now we become incapable of peeling back any of those abstraction layers, because AI has killed our ability to learn. Your comment is extremely asinine
Maybe the issue is that the devices we use are constantly interrupting us and leverage necessity to do it? A parent gives their kid a phone to keep in contact or so they aren't socially isolated and then every business in the world is trying to interrupt their thought processes in the hopes making a little money. Hard to learn how to complete a thought if the thoughts are never allowed to complete. I notice many of you are real quick to blame the younger generation for systems created by older generations that exploit younger minds. Almost as if you didn't complete the thought...
we have a lost generation of college grads, students who were able to cheat their way through school using AI before schools could get it under control.
@@bullpup1337 generation definition 1.c on Merriam-Webster: a group of individuals having contemporaneously a status (such as that of students in a school) which each one holds only for a limited period
People have always been cheating their ways through college. When I went to college long before AI, other than me, all other hundreds of kids were cheating hard, on every single exam.
Thank you for the awesome video. As a "Junior Dev" I've been getting this nagging feeling that I have been over reliant on AI and your thoughts and the comments on the video really encapsulate how I feel. Im going to try and ensure i use AI as a tool when needed but also actively solve problems and reflect on new things i learn by reading.
If you don't learn the tools you will be left behind. I'm a lead dev and I'm writing code in days that used to take me 3 weeks+ Don't listen to this guy or anyone telling you the opposite. Take your time to learn how to use the tools. It's immensely important.
@@jeffsteyn7174man, he is a junior and as he says and what all this topic is about -- he uses too much of the tools. He already knows how they work. But you're an experienced dev with a lot of knowledge and experience. Now you learn new tools to optimize your knowledge, it's awesome. But let them learn basics and get some knowledge too so they could properly utilize the tools and be able to know when not to.
Back in the 1960s we had a fairly serious problem with university students using Cliffs Notes for everything instead of reading the books. At the same time, there was a movement in primary schools to teach "new maths" and give students a stronger footing on which to build a knowledge of calculus and number theory. When the professors banged on the table and said "read the whole book, it is important" the students laughed and asked why they should bother. And while the primary school students didn't know enough to protest the new maths being taught in schools, their parents raised eight shades of hell about the homework being too difficult. Sixty years later, we have the exact same problem: people are weak and stupid. They LIKE being weak and stupid, and they will fight you to REMAIN weak and stupid. The number of people willing to be strong and smart just keeps going down, because why should they? Weak, stupid people are doing JUST FINE in the modern world. Nothing has actually changed with the invention of generative algorithms, or with the ubiquitous awareness and availability of the garbage-farting machine people are using to do more and more of their work - because they are too weak to do better themselves, and too stupid to know it's garbage. People have always been like this.
I am old school. Was at the airport, flight delayed. I had a phone with internet access in my pocket, but I pulled out my years old walkie man, with tons of music on a micro sd card and started playing 20yo music into my ears. I even danced a bit, lost in the present moment. Nice balance ahead of hours of sitting afterwards. AI helped me to take on a journey with Linux, answering my many questions and troubleshooting, but I don't see doing it all for me, like Windows and Microsoft are promoting it to us. That's why I am running away from this AI organised life. I want to keep control over my devices, not being navigated around by someone's algorithms like a vegetable. I am also back to reading books and watching movies I own. No streaming, no ads.
I think Part of the Problem is that people started taking this career although they don’t enjoy thinking and solving problems. Many people just took this path to earn big money, not knowing that you have to put in a lot of time and ellbow grease into learning
Yep! That's true! I've met more people in this industry who treat coding as just "a job" which makes great money... As opposed to people who are REALLY passionate about coding, and are willing to work their butts off to write the BEST code that they can. Furthermore, the dispassionate money grabbers tend to write bad quality code, because honestly... They can't be asked to refine, refactor and test their code properly! They bang out crap as quickly as possible, do the littlest amount of work to at least get their code building... Normally, with a crap load of warnings... And then they're done! The future of this industry is in SERIOUS jeopardy! 😳
That's most people in all jobs, no? You spend 4 years in school, and in America you gotta pay that loan off. It's fine IMO and not every tech job needs a Depp thinker. Especially if you choose to work in a tech job in a non-tech company (maintaining a website, for example). The real issue IMO is that this extended interview process sure as heck isn't getting companies that do need that proper candidates. They just study to the test and turn off the best candidates who find a job in 1 or 2 stages instead of 7. We can't fiz everyone but we can fix who gets hired.
I had a junior dev who didn't know what a bookmark in a browser was. Couldn't get her to understand a project's directory structure as well. She installed the same project in itself multiple times and thus ran out of disk space on her laptop. It hurt so much. I asked my boss to release me from having this person in my team.
@@frankfahrenheit9537how could she person get hired? What are all those stupid 5 stage interviews over 3 months doing? At that point, you may as well save on recruiting costs and hire a candidate out of a hat.
Curiosly, studying with AI has somehow taught me to catch BS in arguments. When corrected it just says like "oh you are right that is not actually correct". Its really like having a studying mate, you work with it but dont make it think for you.
In the 80s we wrote small cheat sheets to fit on pens and rulers only to end up not needing them at all. The process of processing the information and producing the condensed knowledge was all we needed to learn it properly.
The smartest cheaters will probably get an A without cheating. But on this hyper competitive landscape, that cheating is used to turn a 91% to a 98%. Something that should be negligible, but sadly isn't.
The story of a 22 year old Stanford graduate forgetting simple words because they're so used to having chatgpt complete their thoughts is so farcical that my take away here is that anyone will post anything sensational for the clicks.
I immediately figured that poster was lying. Who the hell tells an interviewer "my brain doesn't work anymore. So, when do I start?" Any idiot who believes that story is as dumb as the fairy tale graduate.
yup, caught that too. What the hell does "finishing thoughts" even mean? As far as i know there is not a single ai that can do that currently, unless he was literally typing out his thoughts and asking chatgpt to finish them, but even then that wouldnt cause you to cause a speak in a "buffered" way. This a another BS story on social media to gain buzz and is faux sophisticated. oh yeah and speaking in a buffered way and forgetting easy words is simply due to a recent lack of socializing. I get like that too when I dont talk conversationally to people for a few days. I agree with everything in this video though, and think it has great advice, but that obviously fake story ticked me off.
@@mosh1987I get the general idea of it. I have these lapses in thoughts too that have me forget words or what I was supposed to be doing in that moment. Sometimes obvious words that I woulda known minutes ago. It happens in interviews, casual conversation, and by myself, so it's not stage fright. Hard to describe it, it's like you feel yourself losing a brain cell in real time and then you're trying to focus back on summoning that brain cell back to get it's info. Or you let it go and just try to find other thoughts to make up for it. Idk if it similar but It sounds like that and I can see it happening (not in an interview though. I'd just give an excuse of nerves). I can mitigate it simply because I've gotten used to mentally retracing my steps and getting back to that topic or word. I can imagine people who don't do that just spacing out.
I speak Russian and English and sometimes I forget the words in both languages. I understand the concept and can explain but just can't remember the word. I can totally believe someone being silly enough to admit having such issues during interview because I actually did it myself once or twice.
I truly do appreciate this. It's reminding me how my day really should be structured vs trying to accomplish all my learning and applying simultaneously for the entirety of my day. I'm just trying to break into Tech. The route I'm starting with is a frontend developer. But the timeframe of how fast I'm trying to achieve is in retrospect obscenely small. But possible. With that I need to be taking priority of how I utilize my day, and one of my notorious weaknesses is not giving myself the proper time or space of dumping my brain for the day, which does not allow my brain to shut the fuck up at night. Hence, the need to writing things down and planning out the day so I'm no longer stressing so much. Progress is progress, not perfection. And I can't do everything at once, so why torture myself by constantly circulating everything in my head? And then the meditation bit really had me thinking. I've always attempted to do it in the morning, but maybe that is not the best time for me.. And a side note, in the midst of me doing projects, and not having things written down or blocked out with like the fill, process, and empty cycle; I'm pretty sure my brain is trying to multi serial task in the middle of me working on my damn project. They are such great ideas, but i don't write them down which I then forget about. And because I trailed off, I forgot where the hell I am in the project! It's exhausting. But lookin back, when I have the defined goals, and the rare moments where I either wrote or planned something out, I was highly more successful with my absorption and production. Now why the hell is that hard to replicate on a daily basis? It's aggravating. I'm capable of so much but my subconscious hinders my capabilities. And continuing on, I think doing this little tidbit proves how much I actually need to formulate and let my brain pour out on paper. To say what I just typed is to the analogy of the tip of the iceberg is a hysterical understatement. Shits work... All-in-all, coming across this video was very helpful for me and I see how it will be for others. Thank you
Damn. That's exactly how I feel. Have the same problem of my brain running like crazy at night. During the night I stress about the day, come up with many solutions and different things I should do and by the time morning comes, I forget about it. Then stress about the stuff I came up with and forgot. I really need to start writing things down in the evening and planning my days better. Thanks for a meaningful insight!
im not employed yet but i found myself in a similar situation, so i bought a thinkpad installed arch, learned vim and started using books and docs instead of videos and chatgpt, it's been hard but i'm getting better slowly
Curious question: Learning something with your approach in 5 years or learning the same thing in 2 years by including gtp in the learning process. Which approach is more streamlined and effective?
@Kyouma. well you'll pick up speed gradually which will decrease the time it takes to get to a certain point, and later you can utilise both in a strategic way like after you have your basics strong, that will increase the speed of learning by a lot. and by any means im not an expert im just trying to improve myself in a way i think will make me better. i just want to fall back in love with computers man
as a senior dev, I don’t use ai in my editor at all. I used to but realized that it made me lazy so i stopped using it. I just use chatgpt occasionally to search a piece of information or a bootstrap code example or something. Thats all
I’m about to finish an app I’ve been making since April, with a few collaborators. And that’s what i find AI to be really useful for. It helps me understand documentation. at this point i rarely need any explanations after building up a vocabulary to better understand documentation
Seniors need to be very wary of relying on ai as well. I have experienced brain rot from using it and had to force myself to stop. One of the smartest people I know also regressed back to the point where he literally can't program without it. Luckily this seems like it's reversible with effort.
Multiple times this year, I've had Junior Devs come to me, expecting that "we have documentation for how to fix/resolve everything." If we had that, why wouldn't we have already fixed those things?? I don't think this is just an AI problem. I believe you touched on "asking Google and getting an answer." I think that was the start of "not having to think as much." I fear AI is just accelerating that. :(
That response of junior devs is also a by-product of how we teach them. I know. I teach computer programming at the post-secondary level. We've been in a building crisis over quality control in our teaching for decades now, and AI has just made it worse. But it's hard to bring about change in educational institutes (it often moves at the speed of tenure). I remember trying to bring in unit testing 20 years ago, and no-one wanted us to have it in our program (mostly because no-one else had tried it, used it, or understood it).
@@danielgilleland8611 Totally agree. Schools teach you to give an already known answer. What it doesn't teach is, that your answers are the most valuable, when nobody knows it yet.
In all fairness, that's part of what makes a difference between a junior dev and a mid level dev. A junior should mostly be focused on handling smaller, discrete tasks. Not necessarily ones with a direct answer, but ones where the approach is rather obvious. After a few years of that to get used to the environment, they could be given more freedom in how to approach and design a solution. Mid level duties.
I'm a mid-level developer (4 years now at a small business) I am just starting to do some blog writing, and I honestly, I've learned a lot just by writing my thoughts out. You think you know something, but then you try and communicate it to others and it feels impossible. I can't imagine what a Junior that has only ever developed with AI feels like. The more I code the more I realize it's quite artistic. Is there a lot of logic in it, sure, but you have to be creative to build good things.
So relatable although as a student it kind of makes me sad as I see my classmates use AI for literally anything: essay writing or even simple tasks given by the teacher. I found this video helpful, telling me what I should avoid.
The forgetting words while speaking or writing is so true. I also feel like I'm getting dummer. I never though it's because of ChatGPT but after watching this video and thinking about it, I think it's true. I wonder how to get over it and become what I was before. Everytime I speak I've to ask my friends what is that words.
As an old timer, starting in a time when RTFM was the default response, not even google. Young people have no idea how things work and cannot think outside of the box.
To be fair though, the current generation is partially responsible for this. It's not often that you can get a fucking decent manual with anything anymore. It's the kids that abstracted away filesystems from their phones, and they weren't around for the analog versions to relate to. So yes, they're massively ignorant. But it's not all their fault.
I find getting enough sleep and exercise helps with my cognitive ability and my desire to pick up a book. But, yes, the internet in general is really addictive and learning to have quiet time is extremely important.
I asked my AI colleague if it could work out why the SQL query was giving the wrong result. AI told me that a comma needed to be reomved. It gave me the corrected version. But the corrected version still had the comma in it. I pointed this out. The AI apologised and then gave me the corrected corrected version which was still the same with the comma that the AI said it had removed. I then copy pasted it, removed the comma & VSC lit-up with syntax errors because the comma had been in the correct place. AI did eventually come up with a solution by adding another alias into the query, but when it explained why the new version worked it made no mention of the added alias. I told it this. The AI apologised for the confusion, and described the code as mine instead of its (blame shifting?) and agreed that I was correct in saying that the alias was the reason. I still don't know why the query overcounted without this extra alias.
All the juniors can do is LeetCode problems since that's how they got through the tech interview. Unfortunately, LeetCode has nothing in common with real world software development.
@@TheMrblaster2012 Hand them a print out of a large stored procedure and ask them what it does and if they see any issues. Alternately hand them the code for a API endpoint that works correctly most of the time but occasionally throws a stack trace. Ask them how they would address this. These are real world problems they will see every day.
Recently I stepped back from using IDE and sophisticated tools to a bare-bone text interface. Confort is great but too much of it kills you. It is quite similar to sport: if you want to evolve, you need to know your limits to better challenge them.
Also, it protects you from being helpless, once the IDE screws you or lacks a feature, or IDE is not available (like some problem on a remote computer). I'm used to get called in these situations, and it's really shocking to see that *devs* might not know anymore how to open a terminal and use it, or even not know how a computer is doing the work in the end, or not know how to use curl, the plain debugger, ... I had this shock on myself long time ago, when I worked complety from inside an IDE, but did not even know how to start the program without clicking the run button inside the IDE, when I got the task to run the program on the server. For me, this was such a shameful experience, that from there I always started to do anything from the rudimentary level and upgrade later to use more powerful tools when I know how to do it without them. Helps me all of the time, and I'm even more efficient (yes: tools are nice, but limited and slow), as I can always switch to the most appropriate way, and I never get stuck, because I have no idea what the tool really does under the hood.
To be fair, I only use Ai to answer questions to thoughts I cannot answer myself (simply because I don't know). It's a game changer. But totally relying on Ai would be learning nothing. But I know that's not what humanity does, we're always looking for efficiency. We both know where consumerism is going, endless, mindless consumption, that's going to be the world soon. Where I in my maths class today, I could just photograph the whiteboard and it can answer it for me.
This video should be played at any new hire onboard presentation in any company. You managed to tackle all the points I’ve seen lately, but not only for young developers. You advices are applicable to any person of any age that is too dependent on social media or AI tools. You nailed it.
Junior dev here, soon to move to mid-level(c++). Have been a hobbyist c++ dev since roughly 2003. I work for a top 100 tech company. We don't get trained on ANYTHING. Design docs are an out-of-date wiki with cryptic notes written by the business people. Senior devs have never been more disconnected, it's like I can consult a google search and get more direct information about a task than what nearly all the senior devs will educate me on. Junior dev coworkers all say the same. They're all in their own information silos and only give a sh1t about what they're currently working on. Getting info from a senior dev feels like pulling teeth. It's not AI causing this, we only very recently got access to a general AI to use. It's because the senior devs suck ass.
The Senior Devs have been put between Scylla and Charybdis. Ageism is rampant in Compter Science careers and Senior Devs know that they have the grim reaper hot on their trails. Training a Jr Dev to do your job, in this environment is freaking suicide. Eventually, you will quit blaming the senior devs and gain a real understanding of the dying career you are in. Mine is gone already.
I think it's an industry problem. IT is a very complex topic. So it's more efficient to have people specialise. Generalists take time to build up, are expensive and rare. This is also made worse by hiring and subseqently teaching practices: imho it doesn't make you a good programmer if you can solve all those little Leetcode Snippets. That works because that's how many companies hire people. But you won't learn programing this way the same way you won't learn chess if you just solve early or endgame problems. You need to play whole games/complete whole projects (to stay in the metaphor). But that takes a lot of time. You say you're a hobbyist since 2003, so you're basically a seasoned professional. But there's also people out there that complete some kind of bootcamp in a few months and then start as junior devs same as you. Now those people will neither have the skill nor the experience you have. They HAVE to take shortcuts in their education to finish this fast. Some things, unfortunately, just take a lot of time to do them right. But that's not incentivized by the modern job market. At least that's my experience.
@@MatthewCleereif you think programming is just writing code and pressing compile, it probably is a dying career for you. For me, programming is thoughts and logic, and architecting and processing data while thinking at scale (something very hard for most people to encapsulate). Your career dies the moment you stop thinking. Software is one of the few areas where a David can compete with a Goliath, despite the obvious disadvantage. If businesses disrespect that, I can still take what really matters and make my own product that people want (be it consumers or those very same businesses).
@raze2012_ Oh give me a break. The "scale" speech. Ancient buzzword nonsense. In reality, 90% of the software out there, and I am talking about retail and ommercial and proprietary, ALL of it, is pure dogshit, and you know it. From internal prohects with air gapped spaghetti worlds full of 20 year old lava, to brand new Android "apps" with spinning toilet drain load "at run time" interfaces full of Murphies and glitches that will NEVER be fixed, because the project manager doesn't have the hours to "refactor". THAT is the real software engineering world, not your dreamy shapeshifting unicorns flying through the cotton candy sky by shooting rainbows out of their "scalable" @$$#@!3$.
Don't entirely trust AI for basic facts, either. Google-searches seem to like to have an AI answer at the beginning. Both the ones I read recently were factually wrong. (One was about whether a particular plant is native to a certain part of North America, and the other was about sources of vitamin B-12 in foods. So, they weren't difficult, complicated topics.)
As someone on a bit of a GPT detox and reading Clean Architecture to understand SOLID for my csharp, this came at the right time. I still remember when we were on a video call and someone asked 'who uses GPT', and one of the mids i respect didn't raise his hand.
Domino !!! Effect!!! there little reward for being knowledgeable.... IN AMERICA TODAY. My grandparents taught me how to farm but also purchased the Tandy TRS-80 later I learned Programming and Electronics
@@crypticsailorThroughout history, most people simply can't think. We think too highly of people. It's the same before and after TikTok. If older programmers could think, they would follow Dijkstra's path to not create so many bugs for the coming generation to fix.
Writing is needed, but even just taking time away from devices in general and taking the opportunity to become bored and thinking about things. Giving the mind space to think is probably a prerequisite for writing.
The thing is though that most companies are fully embracing AI and want their workforce to use it as much as possible. Cynically thinking I believe this is so that in the future they can have less staff, so less direct costs.
It's a cultural problem in how the culture educates children and in how we introduce kids to various technologies. Just based on my own experiences and observations over 50+ years (and conversations with other parents and teachers), proper brain and social development seems to require strongly restricting/curating use of certain tech during a child's first 16 years and possibly completely avoiding certain tech during the first 4 to 6 years of a child's development.
I feel the struggle and anxiety is more to do with programming the logic in a restricted amount of time. If you were given a task that required advanced logical thinking, how do you execute the logic within a given deadline especially in web applications where you aren't allowed to use external libraries. It kind of makes someone anxious when you don't have a team or a co-developer to figure out the logic with you.
It's a self fulfilling disaster. You outsource your cognitive deep thinking and reasoning thus you become weaker in these and fall behind the AI .then people will say I told you so AI is better
Its not just the junior developers changing attitude that worrying. It seems to be having impact across the board. I deploy and customize an off-the-shelf version of our software. We've been pestering a client for their work permitting process to configure it in the work order module. Its a major dependcy, a safety critical procedure and turns out it has been enforced manually at each site but through very different methods. Just before Christmas, the clients PM sends an LLM written response to "How do i make a work permitting procedure?" within an LLM drafted email that trys to make it our fault that the project is behind. People have lost jobs by trying to hide that copy/paste straight off stack overflow or reference docs. Why the hell are they encouraging LLM use when it amounts to the same thing? Yes, we're working to an outcome, but the way we get there is just as important.
The internet is destroying my ability to read books. Early in my younger days I used to read a book every couple of days. Now I struggle to read anything that's not technical, and even that comes through a display of some sort, be it my PC, laptop, tablet or the damn phone (which very negatively impacted my eyes). I was lucky to get my degree in math before the internet took complete control of life. I wonder how the new generation is going to be as skillfull as our generation managed to become. Reading books is essential to maintaining a thinking society.
Very coherent way of thinking! To aim using the right tool, or right ways / processes for each situation, taking into account our own self development. Really, thank you for this insight, Travis!
Very right. I often struggle reviewing Jr Dev code. They must not understand how anything works anymore. The fallacy of AI won't replace you; people using AI will get many.
Excellent advise. It mirrors my experience. I think there is a place for AI as a tool. But we should never ever loose the ability to think things through and form our own opinions. Also as a senior dev with 30+ years of experience I see the great value of seeing the bigger picture and creating real value by questioning and shaping feature and design decisions.
What a brilliant message,Thanks Travis Media. After months of AI anxiety it finally occurred to me that AI should be used as an assistant to help humans learn better. Those who stop learning are doomed. This is really a very important topic. In fact TH-cam is also contributing to the ills as most people dont do research anymone.
I just graduated and I been working as a junior dev in my first rotation for 6 months. I personally have struggled with finding that balance with thinking and when to use chat gpt. For me personally it’s kind of a since of paranoia of am I getting my work done fast enough and is good with all these tech layoffs happening. I’m not trying to have excuses but trying to find the line.
@@TravisMediado you want an honest assessment? 1 year of internship during school, 3 years on a job force. My team lead: “you are a superstar”, I’m busting my ass off, always getting everything thrown at me done. Passing with flying colors all of the eoy assessments and getting raise. March: one of my old projects is resurfaced by a manager. Max, you need to write documentation for it because we plan to release it. April: Max, you will be testing, releasing and collecting all of the signatures May: Max, you are awesome. Good job. June: new manager comes in, I’m being transferred under this new guy. A week later, I am called to HR to find out that my “position is eliminated and I’m terminated without cause effective immediate”. In December I found out from a coworker that it was my first manager that just cut me off. I’ve been looking for a job in Tech sector since July. First I tried purely embedded(4 years of handson experience). Then I tried to include some Enterprise and Web(had some experience with both during my time at that aviation company). Now I’m applying even for testing positions. Entry-level, or the ones with my years of experience. Soon, I’ll try the entry level tech support, and if by the time my savings drain entirely, I don’t find a job, I will just work in a fcking kitchen. Ah, and btw, no one trained me AND I WISH I WAS TRAINED. WHERE IS THE APPRENTICESHIP? WHERE? WHY ALL SENIORS EXPECT YOU TO COME TO WORK KNOWING EVERYTHING? On top of that, the constant pressure of not doing enough or not doing it quick enough, holy shit.
These were kinda my thoughts. I'm not employed yet, but I imagine there is a lot of pressure to be churning stuff out asap to please stakeholders and customers. Seems like a hard rope to walk. I personally don't see how it is any different than googling or browsing stackoverflow. There were memes about being "professional googlers" before, so idk...
Recently, I worked as a staff member to a competitive programming contest and old teacher of mine hosted and felt saddened to see so many submissions made by GPT. The majority of these didn't even remove the comments. We kept denying the submission of what looked like AI-generated code, and they just retried with the same or similar piece of code, sometimes with the comments removed. The sad part is that the contest didn't even have a prize. It was meant just for training and fun. I was really astonished to see it happen.
It not just devs, it all IT roles. Most of our new employees are terrible at basic networking, simple concepts in both windows and linux and just about everything else. Even some of our "mid" level guys cannot seem to fix anything without help or weeks of back and forth (I work remotely) in the end one of our team members almost always has to seem to travel to find a simple mistake that takes hours at worst then discovers 10 other things they didn't even know were broke then work on fixing those.
I think part of the problem is that people are just getting the answer without trying to process how to get to that answer. One of the aspects that I find very fascinating about AI is being able to have an ongoing dialogue back and forth on various given subjects. It's not just a transactional response. It's an in-depth conversation where I'm trying to better understand the root of the problem and I think that's really what we need to do. Ai and chat GPT are more than just a Google search. It's literally like having a person you can discuss ideas with and it's a missed opportunity when Debs are just asking for the code snippet without actually understanding the logic behind the code snippet.
As someone that's been learning software development for a year and a half for in hopes of becoming a junior dev, I've progressively started relying on chatgpt to summarize manuals or give me easy to understand explanations of concepts. It sure did help a lot with progressing through projects! But I've noticed it's detrimental to actual problem-solving for more complex problems. In scenarios where you can't use chatgpt, I'd feel stumped. I caught on to this big issue and started taking a step back to read more and think more without chatgpt. I agree with the video wholeheartedly and it has helped me bring out my issue with chatgpt into spotlight 🙏. Dont use chatgpt as an easy shortcut for learning, take the time to learn concepts to heart.
I've been working on a project lately and I wonder if I've been using AI effectively or not when it comes to learning and growth. I mean I have hardly used any code Ai has provided me other than a refactor. As for solving problems I have used it to walk me through a problem or at least brainstorming, but in the end I have to think and decide if the suggestions and walkthrough fit the bill or I have to think of a difference approach to solve a problem. It's a dilemma
How is it even possible with requirements to get a junior dev position being so high? It sounds like AI basically allowed a bunch of people who have no talent and would have zero chances to get a job in the field before to jump through the hoops.
This was an excellent and a mind blowing video. Sometimes we dragged ourselves by the common sense thought of being knowledgeable without having the merit of thinking. I found myself falling into the abyss of rely on too much on AI tools that promises act and think for me. I'm happy to watch a video that helped me to open my eyes and improve and work on myself rather than let the new AI assist doing it for me. Travis you are an amazing and inspired professional. Best of all and happy new year.
I bet there's a study that shows that there's causation between student math results and their dependency on calculators. That if you do things by hand, you have a deeper and more intuitive understanding of it.
This video is essential not just for jr.Devs, but for everyone using AI. And the "We write for ourselves" is more relevant than ever. And reading human written books/articles is actually not straight forward, you have to read between the lines. Even tutorial books have something to day that they didn't write. Don't avoid reading a fun novel, if you already read all those "Good books", keep the streak up, if you lose the streak, you open your mind to be consumed by AI. Some of jr.devs in my team actually told me that "Indian guy video get me going faster than reading articles." and my response was "Video runs in the teacher's pace, or at most 2x. But, reading allows you to skip straight to the point and make you understand the concept, not just what letters are on screen." Of course, they kept watching tutorials.
just like we have to intentionally go to the gym or go running to give our bodies what machinery took from us, we'll have to intentionally give our brains challenging problems to keep it healthy and active in a world of AI. and just like there are physically fit and unfit people today, we're going to have mentally fit and unfit people who experience life very differently from one another.
11:00 I don't think LLMs can summarize texts. I tried it, and it was just a common blah-blah on the article's topic. The facts from the article weren't there. It makes sense, of course, LLMs use the training data to answer.
The main change is the bar has become extremely low. It used to be that only those enthusiastic about programming could become programmers, and that enthusiasm is the magic sauce needed to become decent at something so difficult. Now a lot of people want to become developers for all sorts of (bad) reasons, but they have no particular interest in it, and they are sort of doomed from the start.
I love your videos they're different from other programming videos. The thinking part is on the spot, people seem to have either no time for thinking or just lazy and wants things done in seconds instead of letting it take some time and actually learn problem solving. I myself learned very quickly when AI came in the picture that my brain became dumber by copy and pasting, if u can't understand the lines you shouldn't use it, because then you can't solve the problems that comes with it.
I'm a fresher from India. And I'm not the one to read. And the problem IS the Internet and our cellphones. I've still manged learn everything except building my own projects because I'm deeply interested in ML. The main issue with my generation is the lack of time and time management due to our addiction of our smartphones. I find myself scrolling garbage on Instagram and watching geopolitics on TH-cam. All because sometimes doing stressful work seems like an active assault on my stressed brain. We also don't have mentors with time and patience to teach us the ropes. Because they themselves are addicted to smartphones. Now all we can do is make AI our mentor. Make it teach us how to do things. Because our own peers are lacking leadership and patience.
@@pythonxz yes everyone at my job right now is a tech lead or senior in the system (/somehow), yet I am shoveling preventable garbage daily -- I am a senior dev as well if going by title.
wow how ironically relatable this is is to me, in this age my attention span is shrinking so much that i literally can't do anything if its haven't nearing the due date
As a university teacher I can say that almost all students have begun to use chatbots with AI actively to complete almost all assignments in all disciplines. Even talented students, who are capable to complete an assignment on their own without any problems and to train their mind, use AI chatbots the same way as weak and mediocre students use them. Why? Simply, because if a weak student using an AI chatbot is able to complete an assignment quickly and pass it then why I will spend much time and efforts to do the same? Whether am I a looser? As a result, the average level of students’ knowledge is rapidly dropping from the level of independent developers to the level of only reading and understanding the generated code, regardless of the talent level of students. And I am not sure that all of them even understand the generated code enough deeply. This is a very serious shift in general psychology of the new young generation. And I think, it is not good trend.
This is the exact reason why I avoid the use of GPS when I can. Jim Kwik named this phenomenon very well as digital dementia, even before the massive popularity of AI nowadays.
As an employee you get paid your worth. If you can do nothing without the chat bot, your value is "anyone willing to tell the chat bot to do it". Well, that doesn't sound well payed.
There are 2 questions regarding this: 1) Surely many junior devs use AI for basically almost everything. And the company and the mid-senior devs might think they're better off using AI anyway. But what impact will it have in the long term? Who will replace those senior devs once they reach the twilight of their career? 2) How to read documentation? There are some frameworks or tools for which the documentation written is simply marvellous (like Django). But even then I can't seem to make too much sense about it sometimes, particularly for complex topics (middle ware, deployment etc.). How should one approach such tough topics while reading their documentation?
As a junior dev i use ai like a much faster and straight to point browser to solve a problem. If i dont understand the answer. I tell him to explain it more till i understand. I dont just copy paste and move on.
I've met more people in this industry who treat coding as just "a job" which makes great money... As opposed to people who are REALLY passionate about coding, and are willing to work their butts off to write the BEST code that they can. Furthermore, the dispassionate money grabbers tend to write bad quality code, because honestly... They can't be asked to refine, refactor and test their code properly! They bang out crap as quickly as possible, do the littlest amount of work to at least get their code building... Normally, with a crap load of warnings... And then their done! The future of this industry is in SERIOUS jeopardy! 😳
I've have such an issue working with the junior developer in my team. The issue to me is not that they're lazy, they actually work really hard, its that they refuse to think for themselves. I ask for a task to be completed and I get back a copy paste chatGPT answer as a PR.... Its actually quite sad and im not sure how i can tackle this issue in a work environment.
Think also initiative seems to be lacking and no desire to meet deadlines. But with a strong and capable tech stack knowledge. Sometimes just blindly following a ticket. Knowing that there will likely be a followup
I treat AI as a digital toolbox. It contains the tools you need to create the task but it is still up to you to complete the project. AI is still pretty dumb even in Software Development. I remember last year I asked ChatGPT how to create and draw a canvas in Java. The AI gave me the template to get started. I was able to get it done but I wondered if ChatGPT could create word for word in the project details. So I asked it to randomly draw a small town. The AI created it and the drawing was so bad that the grass was the sky, the windows were off, the door was somewhere else, everything was off. The AI doesn't know how to draw in Java. That's why it's a tool, not the solution to everything.
6:00 Something's a little off with a video discussing decreased attention spans (in part) interrupting itself with a minute-long ad spot. I understand the necessity, but still. 10:42 Having ChatGPT summarize an article for me is a non-starter; it deprives me of the intimate understanding of the author and their views I can get by reading and processing the whole thing.
I'm not convinced that you can use AI in a way that does not atrophy your skills. In its current form even using it like an advanced Google simplifies problems that may have multiple solutions to just spoon feeding you one way to do something. Sure you can chatgpt facts and that is probably fine, but just searching for facts is not going to give you a productivity boost. I work with people that have surface level knowledge because they rely on one AI or another for even basic things, in the short term that boosts your productivity but in the long term it does not let you build a deep mental model on how systems function. Reading the documentation and finding the why behind the answer does.
What stinks too as a junior is that when I do make something or use new skills by myself, I’ll occasionally get asked “this is good- did you ask chatgpt to do this?” . It was likely because I had consistent descriptive comments to track what I was doing- but still
I could not agree more! The thing is, the social AI-Transformation only gets worse, as the books and articles that are actually valuable and meaningful get burried under an ever growing pile of AI-generated slop! The internet is dying! And without it, our juniors will have it even harder, but perhaps also easier in a different way to fight their AI addiction.
AI is great for explaining complex things, that are impossible to find via google, because (1) it is fixated on selling stuff (sponsored links) and (2) it will always present the most mundane data first, because those are most often used, and rank higher. Finding something specific is hard.
@ruprecht9997 I've used it a lot to not completely rewrite my code but like "I'm using flask and HTML give me an example on how to create a header menu for my website" and it will spit out generic a boilerplate code that I can look at and understand how it works and then apply it to my own code
Normally I use IA like this. "There is this design decision that I have to make what are the possible approaches and what can be the disadvantages" or "Is it recommended to do it like this with these technologies or it's better to use something else?". I commonly do a "How to do it" and adapt the code given, never a copy and paste.
When I have new developers joining the team (junior, intermediate, or senior), I usually give them a task where they don't need any domain knowledge and with very lax requirements - they can do whatever they want to achieve the goal. And I give them plenty time to complete the task. It's meant to test their problem-solving and technical design skills as well as creativity. Three out of four fail this test. They need specific requirements and instructions and they end up developing a bare minimum that is missing common sense functionalities. So what are they going to do with AI?
In my case it is not like that. I use the documentation, I understand the framework and the theory. Even what steps I have to take. And I get to do what I have to do. The problem is that I barely have to write code. At most I correct it. It's weird, because I understand it but I can't start writing it. It's a problem of foundations.In my case, they also completely changed my stack...So I think it's more than normal that it happened like this, although of course it shouldn't be like this, in fact I suffer because of it, because I know that's not how things should be.
What about the corporations who are replacing us with AI to "boost" productivity and save money? imo it will backfire on them, for obvious reasons, but could you also make a video about it? I am studying, re-educating myself, in my 30s, and i am interested in your opinion on the whole subject and possible advice as to how to handle that if it comes up in an interview. Would it even come up as a subject? Thank you!
If you use AI because you don’t want to learn, you’re gonna have a bad time. If you use it because you want to learn, it’s an amazingly powerful tool.
That's the crazy thing the technology is almost designed for people that don't use it.
@vargonian 100% agree. There's a lot of prejudice about AI because many people let it do their work or thinking. I predict that people who use AI to learn a skill will have an edge over people who lable it as "evil."
In the field of programming, though, it's really only useful for autocomplete within certain contexts.
@@YourComputer I actually find it extremely useful for software engineering even without it completing any code for me (which I disable anyway; I find those tools too intrusive). I regularly ask it for advice on how to approach modeling different systems and their relationships, which design patterns apply most appropriately to the situation, etc., and it often introduces me to context-specific concepts that I wasn't previously familiar with. For example, the other day, without going too far into specifics, I was trying to design an efficient pattern-matcher for words and it introduced me to a data structure that makes searches dramatically faster than what I was doing previously, as well as formalizing concepts in my mind like the "most-constraining-variable heuristic".
If I were exploring an unfamiliar field like Big Data or something, it would be extremely valuable to learn from. I used it extensively a short while ago to learn Docker beyond the basics, for example. And the real power, in my opinion, is being able to ask it "why" instead of just "what" to really solidify your knowledge.
I've "learned" how to do projects in python and React (and some other tools). At my age there was no possible way I was going to truly learn those languages. But with AI I can now use them at an expert level. I guess one downside is if I say I know how to use Python and someone says "oh great how would you select all the primes up to one million" and without chatgpt I'm like, "uh, import "math" or something" .
The reality is I know how to use AI, not the programming tools. But as I noted earlier I was never going to learn them so for me it feels like a big advance. If anything, AI causes me to think more. However, I am not so sure that would be true if I were a kid growing up with AI and just offloaded all my thinking to it.
This whole AI impact on our brains thing is a similar one to the introduction of cars and other vehicles. They help get us to places much quicker than by foot - yet the benefit of walking isn't simply to get us from one place to another, it makes us move our body, our heart pump and our blood flow and many other important bodily processes, that if we were to stop moving, eventually we would develop physical health problems, and perhaps the day we need to walk again we are no longer able to like we could before.
@@SaraGarcia_6123 very well put
indeed. Couldn't agree more
right
The example is similar but different, in the sense that it is a parallel at the physical-material level.
In the past there was something similar, more striking and more subtle that is directly correlated to AI
Around 1995-1996 Intellisense was created, a tool that facilitated access to object properties, which turned out to be a tremendously useful tool and was quickly adopted by every self-respecting language. This made developers rely on this tool, leaving behind any conventional editor.
What was overlooked is that this tool lowered the effort needed to be able to program. It was no longer necessary to inspect objects, properties, methods, the tool did it for us.
Jumping a little further, raise your hand if you did NOT use stack overflow to solve particular (and not so particular) programming problems.
It was no longer necessary to solve a problem for us, it was enough to google it.
AI is a new level of the same: making our lives 'easier'. What we have already figured out is that this has a cost on our quality of life that we need to control.
Part of the video contains a good proposal, although I consider it incomplete. Internal values and discipline are lacking; recognizing that 'making our lives easier' can end up harming us in the long term.
Like the car example, it will be necessary to recognize that even if we have the best car in the world, we will have to walk (and very often) to maintain our quality of life.
i have to build so fast i don't even have time to look back
No, teachers were 100% right about calculators. It's not about being able to do calculations, it's about flexing your brain.
Also, you need to know math to use a calculator, not the opposite.
I think he meant teachers were wrong about "not always having a calculator".
@@TheNefastor That's sort of the same here. I mean... you can always do simple math with the calculator without knowing math (e.g. adding numbers), but you're not going to recognize it then when you misclick, or you get an overflow, or any other sort of issue. You need to know so that you know when your tool is wrong. And it's the same with AI... it's not dependable, you have to already know. And once you already know... the AI just isn't that valuable, it's more of an annoying liability just like a noob you have to teach, but it doesn't improve whereas the noob can.
I'm a junior dev at the start of my career, and I'm really thankful for hearing this message. I will try my best not to kill my brain and career with AI. Thank you.
Excellent! Aside from improving your job security, that perspective will open doors to more interesting opportunities. You’ll develop the reputation of an out-of-the-box thinker, rather than someone waiting to be told what to do next.
Do not listen to this guy.
Learn how to use the tools to speed up your coding. Otherwise you will be replaced by the people who know how to use the tools for coding. Just like this guy will be. Telling people not to use ai is like telling people not to use prebuilt libraries because you should build your own because you wont understand the underlying low level code.
Your worth as a developer is not going to be your ability to write code as it is today it's going to be your ability to guide the ai.
You going need to break down the problem into smaller parts, keep the ai focused on the issue because it has this tendency to do to much in one go.
Don't listen to this guy he is going to lead you down the wrong path.
In fact focus on how to build easy to use apps. Easy to use for your users, not you or another coder. And learn how to figure out what the core issue a user needs solved and give them that in app form.
@@jeffsteyn7174 True . At the end of the day they prefer people who can ship code faster
Not only will you be replaced by people who are using the new tools it is imperative to understand the new food chain.
At the bottom, you have the consumers they are the ones that use large language, models, and pre-artificial general intelligence.
Then you move one level up and you understand there are people using the available open source models, and the API based proprietary ones to build custom Integration solutions. You could call these people the AI integrators.
From there, it gets more complex the higher end of the AI integrators are using agent architectures and coming up with novel solutions for limits to context as well as relevancy search using vector databases.
Also, at the stratified higher end of the AI integration echelons, you will begin to see developers, fine-tuning models that is to say augmenting the training of an existing model with additional training data to steer that model towards particular specialized domains.
Finally, at the very top of the AI food chain are the machine learning researchers that are implementing as well as writing scientific papers in the cutting edge of model architecture and development.
If you want to remain relevant, your goal as a software engineer is to recognize this food chain and begin moving up that food chain as fast as possible.
If you remain a consumer of the cognitive service of these AI solutions, then you are on the road to replacement.
@@jeffsteyn7174 I'm going to add my two cents as an IT manager. My most productive people are the ones who embrace AI, but have done so AFTER learning the basics, and understanding the technology they're working with. They're the people who could build the solution anyway, but can build it faster with AI. Someone who starts out with AI tools isn't going to have the grounding to know when the AI is wrong, or fix the problems it creates. Also, people should absolutely implement low level libraries from scratch as part of their learning journey. Maybe you don't use your home-grown version in production, but it's very limiting to be completely reliant on high-level tools for everything all the time.
I think this isn’t an AI problem but a lack of learning to think logically and slow .. which is also a lack of reading books, articles, or documentation at all .. also not wanting to feel the discomfort of a hard problem. This effect is very real and something I’ve struggled with as a Sr dev and now manager with Jr and even Mid level devs.
I'm not a dev at all but I've often wondered if people pushing themselves to solve so many problems 'in real time' is a bad idea. And it probably doesn't help that online culture seems to give the impression that everyone is writing code the way they do in movies - moving their fingers nonstop as they talk out the problem in a matter of seconds and when they stop the program is done and works perfectly. I would think that slowing down and actually thinking through the whole process and then once you have a working model, verifying that with tests and line-by-line walk throughs would help in more ways than one. But like I said, I'm not a dev. So I might just be spitting BS at this point.
The biggest thing I try to do and the Leads when teaching how to think and analyze is to have them
1. State the problem clearly
2. Explain at least 2 solutions to the problem
Many don’t know to do this and try to just solve stuff without even knowing the problem
The next step is then to explain the tradeoffs of the problem, which is tricky and often requires more experienced coders to help .. but if they follow then they’re learning
I think in part it’s the google it mindset , some also is a lack of critical thinking and analysis skills. A lot of coding also has become more abstracted so I notice also many developers don’t get the fundamentals ..’which would be like knowing calculus but not really understanding say the trig formulas used or having strong command over algebra.
There does seem to be so many more distractions these days and say sitting down for a month and doing the basics seems a big ask. Even if there’s time given at work. This I think is a symptom of our smart phone distraction culture
I was a literature major before switching to tech and noticed I couldn’t focus like I used to
I realized it was mostly not reading paper books where I can’t get distracted ..so when I need to really learn new concepts even in tech I’ll often get a paper copy of a book so I can focus then I’ll go do hands on stuff
So learning how to learn and finding a way to learn challenging concepts versus watching an easy to consume video is part of the challenge too
@@GuitarWithBrett I get that. I find videos just make my eyes glaze over and I learn nothing. But I have loads of old programming books that I learned with. Something about actually sitting in a quiet room and reading actually made my brain work through the entire line of reasoning from the writer and I often found I was getting way ahead of the material as I started to grasp what possibilities were available to me.
@ yeah, it’s hard … really you need the right next level step ideally and a real world problem .. for example I learned to setup background queues as the sr architect asked me to set that up.. I tried , got it wrong , then he explained why was wrong , then I redid .. then after that reading books on queues made total sense because I had the problem and also got stuck
I’m also super selective on books I’ll read and I don’t read them straight through
So right now I need to remember the graphql pipeline and how the middleware works .. I found a great book that explains that years ago for my language and graphql implementation.. so I’ll reread that part because we need to do stuff with it next week
What are you wanting to learn ?
@@GuitarWithBrett Sounds like you were lucky to have a Sr that could take the time to explain something to you. I gather from channels like this that everyone is kinda expected to just come out of the womb fully-formed as a full-stack developer or something these days. As for me, I've never worried about what I want to learn and instead focused on what I wanted to do. The learning seemed to come from that, I guess. Though it also leaves huge gaping holes of knowledge too.
I'm currently making a video game that is deliberately going for an early 90s DOOM-like aesthetic. I want to utilize the flat, billboarded, sprites for characters but I don't have time to draw every frame for every character and I want to use more modern tools like motion retargetting and modular character outfits. This lead me down a rabbit hole of creating a system to pre-render 3D characters as sprites in real time. But it's too slow for what I want. I can only get a few hundred on screen at once and I'd liked to aim for 1k+ so I'm going to have to get creative about it. I have some ideas but time's been short lately and I don't think I'll have the opportunity to try them all so I'll have to take a guess at the one that most likely will succeed and go from there.
Recently during a debugging session with a junior dev (us trying to reproduce a bug), I asked them to add another item to an existing array list; and they started typing out a comment for copilot to do it. To 'effin add an item to a simple array list! I was like c'mon! Please don't do this.
@@zshn 😳
Just to clear , use the push method to achieve this ? @zshn
🤣🤣🤣
I've never met a junior who could do better than ChatGPT. It will lead to a huge shortage of senior devs. Once AI has cannibalized all code and can't get better, we seniors in our 30s and 40s and 50s will be in super demand in the 2030s. Watch.
Soon they wont just destroy junior devs. They will destroy any senior dev too. Now, a person who can't code for nuts, will be able to code just as well as a coder who spent 20 years of his life learning. It's amazing. What a perfect opportunity to get ahead of the curve.
AI isn’t destroying Junior Dev jobs- it’s destroying the Junior devs directly 😂
Yes. It is a very dangerous tool if they are allowed to use it. Hell, it can even ruin a senior if he relies too much on it. Thinking is a skill that needs to be constantly trained. And AI is very very tempting. You can spend 4 hours thinking about something or asking AI and not thinking at all. It's a real danger.
@@Ragnar452 using the O1 GPT model or Claude, I almost always find that they cannot solve my problem so i end up having to think. After a certain level of complexity they turn useless
@DavidDamian-s2y Exactly. We should not be too worried about AI
As another commenter put it
>Ok, so I am using sand that has been processed for 4-5 months in a remote factory in taiwan
>using machines made in amsterdam
>shipped 2-3 months via evergreen
>stockpiled in some warehouse I never see
>delivered to me via a truck that runs on dinosaur bones
>in a box made of dead trees
>installed in a motherboard that followed a completely different path, made of completely different materials, lithium for the 2032 battery from cancun, the gold for the circuits is mined from zambia, the fan is designed by some random in uruguay that had the credit stolen by him from someone in california that had them manufactured in china
>running an OS made by some swedish or norse or danish guy idk
>maintained by thousands of people
>stolen and re-done by some fruitarian in Palo Alto that died of liver cancer
>running on a mix of electricity from uranium mined in Ural Mountains and hydroelectricity from a dam, and geothermal powerplants and wind turbines
>getting information to it over long glass wires that reflect light with almost 0 reflectivity loss
>displaying it on an array of millions of pixels
>and all I do is string together API calls that were written by people long dead before I was born
BUT GOD DAMN IT I WILL NOT USE AI CUZ THAT'S ONE ABSTRACTION LAYER TOO MUCH
@@leoym1803And now we become incapable of peeling back any of those abstraction layers, because AI has killed our ability to learn. Your comment is extremely asinine
Maybe the issue is that the devices we use are constantly interrupting us and leverage necessity to do it? A parent gives their kid a phone to keep in contact or so they aren't socially isolated and then every business in the world is trying to interrupt their thought processes in the hopes making a little money. Hard to learn how to complete a thought if the thoughts are never allowed to complete. I notice many of you are real quick to blame the younger generation for systems created by older generations that exploit younger minds. Almost as if you didn't complete the thought...
we have a lost generation of college grads, students who were able to cheat their way through school using AI before schools could get it under control.
A generation? Bro its been like 4 years.
@@bullpup1337 generation definition 1.c on Merriam-Webster: a group of individuals having contemporaneously a status (such as that of students in a school) which each one holds only for a limited period
People have always been cheating their ways through college. When I went to college long before AI, other than me, all other hundreds of kids were cheating hard, on every single exam.
@@bullpup1337 you think we've solved the problem of AI cheating yet?
Thank you for the awesome video. As a "Junior Dev" I've been getting this nagging feeling that I have been over reliant on AI and your thoughts and the comments on the video really encapsulate how I feel. Im going to try and ensure i use AI as a tool when needed but also actively solve problems and reflect on new things i learn by reading.
If you don't learn the tools you will be left behind. I'm a lead dev and I'm writing code in days that used to take me 3 weeks+
Don't listen to this guy or anyone telling you the opposite.
Take your time to learn how to use the tools. It's immensely important.
@@jeffsteyn7174man, he is a junior and as he says and what all this topic is about -- he uses too much of the tools. He already knows how they work. But you're an experienced dev with a lot of knowledge and experience. Now you learn new tools to optimize your knowledge, it's awesome. But let them learn basics and get some knowledge too so they could properly utilize the tools and be able to know when not to.
Back in the 1960s we had a fairly serious problem with university students using Cliffs Notes for everything instead of reading the books. At the same time, there was a movement in primary schools to teach "new maths" and give students a stronger footing on which to build a knowledge of calculus and number theory.
When the professors banged on the table and said "read the whole book, it is important" the students laughed and asked why they should bother. And while the primary school students didn't know enough to protest the new maths being taught in schools, their parents raised eight shades of hell about the homework being too difficult.
Sixty years later, we have the exact same problem: people are weak and stupid. They LIKE being weak and stupid, and they will fight you to REMAIN weak and stupid. The number of people willing to be strong and smart just keeps going down, because why should they? Weak, stupid people are doing JUST FINE in the modern world.
Nothing has actually changed with the invention of generative algorithms, or with the ubiquitous awareness and availability of the garbage-farting machine people are using to do more and more of their work - because they are too weak to do better themselves, and too stupid to know it's garbage. People have always been like this.
They are doing better than fine, they are the president now.
@@technolus5742 And one of them is the richest man in the world
Amazing comment!
@@cdarklock Funny enough it had takne some time for those from opposite side of asile to recognize, that Musk is no genius.
What is the benefit of tryharding?
I am old school. Was at the airport, flight delayed. I had a phone with internet access in my pocket, but I pulled out my years old walkie man, with tons of music on a micro sd card and started playing 20yo music into my ears. I even danced a bit, lost in the present moment. Nice balance ahead of hours of sitting afterwards. AI helped me to take on a journey with Linux, answering my many questions and troubleshooting, but I don't see doing it all for me, like Windows and Microsoft are promoting it to us. That's why I am running away from this AI organised life. I want to keep control over my devices, not being navigated around by someone's algorithms like a vegetable. I am also back to reading books and watching movies I own. No streaming, no ads.
I think Part of the Problem is that people started taking this career although they don’t enjoy thinking and solving problems. Many people just took this path to earn big money, not knowing that you have to put in a lot of time and ellbow grease into learning
Yep! That's true! I've met more people in this industry who treat coding as just "a job" which makes great money... As opposed to people who are REALLY passionate about coding, and are willing to work their butts off to write the BEST code that they can.
Furthermore, the dispassionate money grabbers tend to write bad quality code, because honestly... They can't be asked to refine, refactor and test their code properly! They bang out crap as quickly as possible, do the littlest amount of work to at least get their code building... Normally, with a crap load of warnings... And then they're done!
The future of this industry is in SERIOUS jeopardy! 😳
@ I totally agree! Especially AI tools like CoPilot will produce horrible code bases in the hands of amateurs 😂
That's most people in all jobs, no? You spend 4 years in school, and in America you gotta pay that loan off. It's fine IMO and not every tech job needs a Depp thinker. Especially if you choose to work in a tech job in a non-tech company (maintaining a website, for example).
The real issue IMO is that this extended interview process sure as heck isn't getting companies that do need that proper candidates. They just study to the test and turn off the best candidates who find a job in 1 or 2 stages instead of 7. We can't fiz everyone but we can fix who gets hired.
I had a junior dev who didn't know what a bookmark in a browser was. Couldn't get her to understand a project's directory structure as well. She installed the same project in itself multiple times and thus ran out of disk space on her laptop. It hurt so much. I asked my boss to release me from having this person in my team.
Maybe you should have trained her in those missing steps. 😊
How can you not know what a bookmark is?
Anyway, how could this person make it through college?
@TabletMini If only it was this easy
@@frankfahrenheit9537how could she person get hired? What are all those stupid 5 stage interviews over 3 months doing? At that point, you may as well save on recruiting costs and hire a candidate out of a hat.
How did you hire them without figuring that out first?
Curiosly, studying with AI has somehow taught me to catch BS in arguments. When corrected it just says like "oh you are right that is not actually correct". Its really like having a studying mate, you work with it but dont make it think for you.
In the 80s we wrote small cheat sheets to fit on pens and rulers only to end up not needing them at all. The process of processing the information and producing the condensed knowledge was all we needed to learn it properly.
The smartest cheaters will probably get an A without cheating. But on this hyper competitive landscape, that cheating is used to turn a 91% to a 98%. Something that should be negligible, but sadly isn't.
The story of a 22 year old Stanford graduate forgetting simple words because they're so used to having chatgpt complete their thoughts is so farcical that my take away here is that anyone will post anything sensational for the clicks.
I immediately figured that poster was lying. Who the hell tells an interviewer "my brain doesn't work anymore. So, when do I start?" Any idiot who believes that story is as dumb as the fairy tale graduate.
yup, caught that too. What the hell does "finishing thoughts" even mean? As far as i know there is not a single ai that can do that currently, unless he was literally typing out his thoughts and asking chatgpt to finish them, but even then that wouldnt cause you to cause a speak in a "buffered" way. This a another BS story on social media to gain buzz and is faux sophisticated.
oh yeah and speaking in a buffered way and forgetting easy words is simply due to a recent lack of socializing. I get like that too when I dont talk conversationally to people for a few days. I agree with everything in this video though, and think it has great advice, but that obviously fake story ticked me off.
@@mosh1987I get the general idea of it. I have these lapses in thoughts too that have me forget words or what I was supposed to be doing in that moment. Sometimes obvious words that I woulda known minutes ago. It happens in interviews, casual conversation, and by myself, so it's not stage fright.
Hard to describe it, it's like you feel yourself losing a brain cell in real time and then you're trying to focus back on summoning that brain cell back to get it's info. Or you let it go and just try to find other thoughts to make up for it.
Idk if it similar but It sounds like that and I can see it happening (not in an interview though. I'd just give an excuse of nerves). I can mitigate it simply because I've gotten used to mentally retracing my steps and getting back to that topic or word. I can imagine people who don't do that just spacing out.
I forget random words all the time, it has nothing to do technology it's a condition like dyslexia.
I speak Russian and English and sometimes I forget the words in both languages. I understand the concept and can explain but just can't remember the word.
I can totally believe someone being silly enough to admit having such issues during interview because I actually did it myself once or twice.
I truly do appreciate this. It's reminding me how my day really should be structured vs trying to accomplish all my learning and applying simultaneously for the entirety of my day. I'm just trying to break into Tech. The route I'm starting with is a frontend developer. But the timeframe of how fast I'm trying to achieve is in retrospect obscenely small. But possible. With that I need to be taking priority of how I utilize my day, and one of my notorious weaknesses is not giving myself the proper time or space of dumping my brain for the day, which does not allow my brain to shut the fuck up at night. Hence, the need to writing things down and planning out the day so I'm no longer stressing so much. Progress is progress, not perfection. And I can't do everything at once, so why torture myself by constantly circulating everything in my head? And then the meditation bit really had me thinking. I've always attempted to do it in the morning, but maybe that is not the best time for me..
And a side note, in the midst of me doing projects, and not having things written down or blocked out with like the fill, process, and empty cycle; I'm pretty sure my brain is trying to multi serial task in the middle of me working on my damn project. They are such great ideas, but i don't write them down which I then forget about. And because I trailed off, I forgot where the hell I am in the project! It's exhausting. But lookin back, when I have the defined goals, and the rare moments where I either wrote or planned something out, I was highly more successful with my absorption and production. Now why the hell is that hard to replicate on a daily basis? It's aggravating. I'm capable of so much but my subconscious hinders my capabilities.
And continuing on, I think doing this little tidbit proves how much I actually need to formulate and let my brain pour out on paper. To say what I just typed is to the analogy of the tip of the iceberg is a hysterical understatement.
Shits work...
All-in-all, coming across this video was very helpful for me and I see how it will be for others. Thank you
Damn. That's exactly how I feel. Have the same problem of my brain running like crazy at night. During the night I stress about the day, come up with many solutions and different things I should do and by the time morning comes, I forget about it. Then stress about the stuff I came up with and forgot.
I really need to start writing things down in the evening and planning my days better. Thanks for a meaningful insight!
im not employed yet but i found myself in a similar situation, so i bought a thinkpad installed arch, learned vim and started using books and docs instead of videos and chatgpt, it's been hard but i'm getting better slowly
Curious question: Learning something with your approach in 5 years or learning the same thing in 2 years by including gtp in the learning process. Which approach is more streamlined and effective?
@Kyouma. well you'll pick up speed gradually which will decrease the time it takes to get to a certain point, and later you can utilise both in a strategic way like after you have your basics strong, that will increase the speed of learning by a lot. and by any means im not an expert im just trying to improve myself in a way i think will make me better. i just want to fall back in love with computers man
I saw that you use arch, btw
learning from books is cool, really :d
@@paulsaulpaul arch, btw!
as a senior dev, I don’t use ai in my editor at all. I used to but realized that it made me lazy so i stopped using it. I just use chatgpt occasionally to search a piece of information or a bootstrap code example or something. Thats all
I’m about to finish an app I’ve been making since April, with a few collaborators. And that’s what i find AI to be really useful for. It helps me understand documentation. at this point i rarely need any explanations after building up a vocabulary to better understand documentation
Same here. ChatGPT is the most epic tutor I've ever had, and really teh only one. But I do all of the coding.
Seniors need to be very wary of relying on ai as well. I have experienced brain rot from using it and had to force myself to stop. One of the smartest people I know also regressed back to the point where he literally can't program without it. Luckily this seems like it's reversible with effort.
You will be replaced by devs who use AI. Gl
@Djolesabijacica i have only ever heard this from people that aren’t employed as developers lol
Multiple times this year, I've had Junior Devs come to me, expecting that "we have documentation for how to fix/resolve everything." If we had that, why wouldn't we have already fixed those things??
I don't think this is just an AI problem. I believe you touched on "asking Google and getting an answer." I think that was the start of "not having to think as much." I fear AI is just accelerating that. :(
That response of junior devs is also a by-product of how we teach them. I know. I teach computer programming at the post-secondary level. We've been in a building crisis over quality control in our teaching for decades now, and AI has just made it worse. But it's hard to bring about change in educational institutes (it often moves at the speed of tenure). I remember trying to bring in unit testing 20 years ago, and no-one wanted us to have it in our program (mostly because no-one else had tried it, used it, or understood it).
@@danielgilleland8611 Totally agree. Schools teach you to give an already known answer. What it doesn't teach is, that your answers are the most valuable, when nobody knows it yet.
If you had perfect documentation to solve all problems, you wouldn't need to hire employees at all.
Just make a program.
@@danielgilleland8611 tenure is a problem, they really should retire people instead
In all fairness, that's part of what makes a difference between a junior dev and a mid level dev. A junior should mostly be focused on handling smaller, discrete tasks. Not necessarily ones with a direct answer, but ones where the approach is rather obvious.
After a few years of that to get used to the environment, they could be given more freedom in how to approach and design a solution. Mid level duties.
I'm a mid-level developer (4 years now at a small business) I am just starting to do some blog writing, and I honestly, I've learned a lot just by writing my thoughts out. You think you know something, but then you try and communicate it to others and it feels impossible.
I can't imagine what a Junior that has only ever developed with AI feels like.
The more I code the more I realize it's quite artistic. Is there a lot of logic in it, sure, but you have to be creative to build good things.
Algorithms are the science. Data structures are the art.
So relatable although as a student it kind of makes me sad as I see my classmates use AI for literally anything: essay writing or even simple tasks given by the teacher. I found this video helpful, telling me what I should avoid.
The forgetting words while speaking or writing is so true. I also feel like I'm getting dummer. I never though it's because of ChatGPT but after watching this video and thinking about it, I think it's true.
I wonder how to get over it and become what I was before. Everytime I speak I've to ask my friends what is that words.
As an old timer, starting in a time when RTFM was the default response, not even google. Young people have no idea how things work and cannot think outside of the box.
To be fair though, the current generation is partially responsible for this. It's not often that you can get a fucking decent manual with anything anymore. It's the kids that abstracted away filesystems from their phones, and they weren't around for the analog versions to relate to.
So yes, they're massively ignorant. But it's not all their fault.
I find getting enough sleep and exercise helps with my cognitive ability and my desire to pick up a book.
But, yes, the internet in general is really addictive and learning to have quiet time is extremely important.
I asked my AI colleague if it could work out why the SQL query was giving the wrong result. AI told me that a comma needed to be reomved. It gave me the corrected version. But the corrected version still had the comma in it.
I pointed this out. The AI apologised and then gave me the corrected corrected version which was still the same with the comma that the AI said it had removed. I then copy pasted it, removed the comma & VSC lit-up with syntax errors because the comma had been in the correct place.
AI did eventually come up with a solution by adding another alias into the query, but when it explained why the new version worked it made no mention of the added alias.
I told it this. The AI apologised for the confusion, and described the code as mine instead of its (blame shifting?) and agreed that I was correct in saying that the alias was the reason.
I still don't know why the query overcounted without this extra alias.
All the juniors can do is LeetCode problems since that's how they got through the tech interview. Unfortunately, LeetCode has nothing in common with real world software development.
In this case, is it better to focus on projects rather than coding exercises? Or both but more focused on the former?
@@TheMrblaster2012 Hand them a print out of a large stored procedure and ask them what it does and if they see any issues. Alternately hand them the code for a API endpoint that works correctly most of the time but occasionally throws a stack trace. Ask them how they would address this. These are real world problems they will see every day.
@@BaronBlud thanks
Recently I stepped back from using IDE and sophisticated tools to a bare-bone text interface. Confort is great but too much of it kills you. It is quite similar to sport: if you want to evolve, you need to know your limits to better challenge them.
Also, it protects you from being helpless, once the IDE screws you or lacks a feature, or IDE is not available (like some problem on a remote computer). I'm used to get called in these situations, and it's really shocking to see that *devs* might not know anymore how to open a terminal and use it, or even not know how a computer is doing the work in the end, or not know how to use curl, the plain debugger, ...
I had this shock on myself long time ago, when I worked complety from inside an IDE, but did not even know how to start the program without clicking the run button inside the IDE, when I got the task to run the program on the server. For me, this was such a shameful experience, that from there I always started to do anything from the rudimentary level and upgrade later to use more powerful tools when I know how to do it without them. Helps me all of the time, and I'm even more efficient (yes: tools are nice, but limited and slow), as I can always switch to the most appropriate way, and I never get stuck, because I have no idea what the tool really does under the hood.
To be fair, I only use Ai to answer questions to thoughts I cannot answer myself (simply because I don't know). It's a game changer. But totally relying on Ai would be learning nothing. But I know that's not what humanity does, we're always looking for efficiency. We both know where consumerism is going, endless, mindless consumption, that's going to be the world soon. Where I in my maths class today, I could just photograph the whiteboard and it can answer it for me.
@Exhithronous-y1n The endless pursuit of efficiency is killing the willingness to engage in anything at a deeper level.
This video should be played at any new hire onboard presentation in any company. You managed to tackle all the points I’ve seen lately, but not only for young developers. You advices are applicable to any person of any age that is too dependent on social media or AI tools. You nailed it.
Junior dev here, soon to move to mid-level(c++). Have been a hobbyist c++ dev since roughly 2003.
I work for a top 100 tech company. We don't get trained on ANYTHING. Design docs are an out-of-date wiki with cryptic notes written by the business people. Senior devs have never been more disconnected, it's like I can consult a google search and get more direct information about a task than what nearly all the senior devs will educate me on. Junior dev coworkers all say the same.
They're all in their own information silos and only give a sh1t about what they're currently working on. Getting info from a senior dev feels like pulling teeth. It's not AI causing this, we only very recently got access to a general AI to use. It's because the senior devs suck ass.
The Senior Devs have been put between Scylla and Charybdis. Ageism is rampant in Compter Science careers and Senior Devs know that they have the grim reaper hot on their trails. Training a Jr Dev to do your job, in this environment is freaking suicide. Eventually, you will quit blaming the senior devs and gain a real understanding of the dying career you are in. Mine is gone already.
I think it's an industry problem. IT is a very complex topic. So it's more efficient to have people specialise. Generalists take time to build up, are expensive and rare. This is also made worse by hiring and subseqently teaching practices: imho it doesn't make you a good programmer if you can solve all those little Leetcode Snippets. That works because that's how many companies hire people. But you won't learn programing this way the same way you won't learn chess if you just solve early or endgame problems. You need to play whole games/complete whole projects (to stay in the metaphor). But that takes a lot of time. You say you're a hobbyist since 2003, so you're basically a seasoned professional. But there's also people out there that complete some kind of bootcamp in a few months and then start as junior devs same as you. Now those people will neither have the skill nor the experience you have. They HAVE to take shortcuts in their education to finish this fast. Some things, unfortunately, just take a lot of time to do them right. But that's not incentivized by the modern job market. At least that's my experience.
programming since 2003? junior dev? am I missing context?
@@MatthewCleereif you think programming is just writing code and pressing compile, it probably is a dying career for you.
For me, programming is thoughts and logic, and architecting and processing data while thinking at scale (something very hard for most people to encapsulate). Your career dies the moment you stop thinking.
Software is one of the few areas where a David can compete with a Goliath, despite the obvious disadvantage. If businesses disrespect that, I can still take what really matters and make my own product that people want (be it consumers or those very same businesses).
@raze2012_ Oh give me a break. The "scale" speech. Ancient buzzword nonsense. In reality, 90% of the software out there, and I am talking about retail and ommercial and proprietary, ALL of it, is pure dogshit, and you know it. From internal prohects with air gapped spaghetti worlds full of 20 year old lava, to brand new Android "apps" with spinning toilet drain load "at run time" interfaces full of Murphies and glitches that will NEVER be fixed, because the project manager doesn't have the hours to "refactor". THAT is the real software engineering world, not your dreamy shapeshifting unicorns flying through the cotton candy sky by shooting rainbows out of their "scalable" @$$#@!3$.
Don't entirely trust AI for basic facts, either. Google-searches seem to like to have an AI answer at the beginning. Both the ones I read recently were factually wrong. (One was about whether a particular plant is native to a certain part of North America, and the other was about sources of vitamin B-12 in foods. So, they weren't difficult, complicated topics.)
As someone on a bit of a GPT detox and reading Clean Architecture to understand SOLID for my csharp, this came at the right time. I still remember when we were on a video call and someone asked 'who uses GPT', and one of the mids i respect didn't raise his hand.
It's not a thinking issue, in my opinion. It is learned helplessness. They don't think because they don't have to.
before one can think, one must can read. There r no excusses to not read.
Domino !!! Effect!!! there little reward for being knowledgeable.... IN AMERICA TODAY.
My grandparents taught me how to farm but also purchased the Tandy TRS-80 later I learned Programming and Electronics
They have to and they don't from my experience. It's the tik tok brain melt.
its because chatgpt resolves 50% of tasks
@@crypticsailorThroughout history, most people simply can't think. We think too highly of people. It's the same before and after TikTok. If older programmers could think, they would follow Dijkstra's path to not create so many bugs for the coming generation to fix.
Writing is needed, but even just taking time away from devices in general and taking the opportunity to become bored and thinking about things. Giving the mind space to think is probably a prerequisite for writing.
100%!
Thanks Travis 👍 Great insights covered here. I feel sorry for little kids who grow up with these AIs. Too easy for them to outsource every challenge 😢
Good points and observations! Your content doesn't disappoint, and I'm glad I found your channel.
The thing is though that most companies are fully embracing AI and want their workforce to use it as much as possible. Cynically thinking I believe this is so that in the future they can have less staff, so less direct costs.
Cynically it seems like a great way to train the ai to do the job entirely
Great advice, amazing video! Whats even more amazing is this message isnt just for devs, its for every human being the this modern era
It's a cultural problem in how the culture educates children and in how we introduce kids to various technologies. Just based on my own experiences and observations over 50+ years (and conversations with other parents and teachers), proper brain and social development seems to require strongly restricting/curating use of certain tech during a child's first 16 years and possibly completely avoiding certain tech during the first 4 to 6 years of a child's development.
excellent video. i piped it to a fabric pattern and it says you made some insightful observations
I feel the struggle and anxiety is more to do with programming the logic in a restricted amount of time.
If you were given a task that required advanced logical thinking, how do you execute the logic within a given deadline especially in web applications where you aren't allowed to use external libraries.
It kind of makes someone anxious when you don't have a team or a co-developer to figure out the logic with you.
It's a self fulfilling disaster. You outsource your cognitive deep thinking and reasoning thus you become weaker in these and fall behind the AI .then people will say I told you so AI is better
@@Dynamite3783 that sums it up
Its not just the junior developers changing attitude that worrying. It seems to be having impact across the board. I deploy and customize an off-the-shelf version of our software.
We've been pestering a client for their work permitting process to configure it in the work order module. Its a major dependcy, a safety critical procedure and turns out it has been enforced manually at each site but through very different methods.
Just before Christmas, the clients PM sends an LLM written response to "How do i make a work permitting procedure?" within an LLM drafted email that trys to make it our fault that the project is behind.
People have lost jobs by trying to hide that copy/paste straight off stack overflow or reference docs. Why the hell are they encouraging LLM use when it amounts to the same thing? Yes, we're working to an outcome, but the way we get there is just as important.
I feel the frustration
The internet is destroying my ability to read books.
Early in my younger days I used to read a book every couple of days.
Now I struggle to read anything that's not technical, and even that comes through a display of some sort, be it my PC, laptop, tablet or the damn phone (which very negatively impacted my eyes).
I was lucky to get my degree in math before the internet took complete control of life.
I wonder how the new generation is going to be as skillfull as our generation managed to become.
Reading books is essential to maintaining a thinking society.
Very coherent way of thinking! To aim using the right tool, or right ways / processes for each situation, taking into account our own self development. Really, thank you for this insight, Travis!
Very right. I often struggle reviewing Jr Dev code. They must not understand how anything works anymore. The fallacy of AI won't replace you; people using AI will get many.
Excellent advise. It mirrors my experience. I think there is a place for AI as a tool. But we should never ever loose the ability to think things through and form our own opinions. Also as a senior dev with 30+ years of experience I see the great value of seeing the bigger picture and creating real value by questioning and shaping feature and design decisions.
I got much more here than what I came searching for. I really needed to hear this.
Here we were worried about 1984 becoming true when the real thing to be worried about was Idiocracy.
What a brilliant message,Thanks Travis Media. After months of AI anxiety it finally occurred to me that AI should be used as an assistant to help humans learn better. Those who stop learning are doomed. This is really a very important topic. In fact TH-cam is also contributing to the ills as most people dont do research anymone.
Why would you feel anxious over chatbots?
I just graduated and I been working as a junior dev in my first rotation for 6 months. I personally have struggled with finding that balance with thinking and when to use chat gpt. For me personally it’s kind of a since of paranoia of am I getting my work done fast enough and is good with all these tech layoffs happening. I’m not trying to have excuses but trying to find the line.
@@malachiwhite5955 this is a very honest assessment about the struggle.
@@TravisMediado you want an honest assessment? 1 year of internship during school, 3 years on a job force. My team lead: “you are a superstar”, I’m busting my ass off, always getting everything thrown at me done. Passing with flying colors all of the eoy assessments and getting raise. March: one of my old projects is resurfaced by a manager. Max, you need to write documentation for it because we plan to release it. April: Max, you will be testing, releasing and collecting all of the signatures
May: Max, you are awesome. Good job.
June: new manager comes in, I’m being transferred under this new guy. A week later, I am called to HR to find out that my “position is eliminated and I’m terminated without cause effective immediate”.
In December I found out from a coworker that it was my first manager that just cut me off.
I’ve been looking for a job in Tech sector since July. First I tried purely embedded(4 years of handson experience). Then I tried to include some Enterprise and Web(had some experience with both during my time at that aviation company). Now I’m applying even for testing positions. Entry-level, or the ones with my years of experience. Soon, I’ll try the entry level tech support, and if by the time my savings drain entirely, I don’t find a job, I will just work in a fcking kitchen.
Ah, and btw, no one trained me AND I WISH I WAS TRAINED. WHERE IS THE APPRENTICESHIP? WHERE? WHY ALL SENIORS EXPECT YOU TO COME TO WORK KNOWING EVERYTHING? On top of that, the constant pressure of not doing enough or not doing it quick enough, holy shit.
Don’t worry bro, you’ll be laid off sooner or later too;)
These were kinda my thoughts. I'm not employed yet, but I imagine there is a lot of pressure to be churning stuff out asap to please stakeholders and customers. Seems like a hard rope to walk. I personally don't see how it is any different than googling or browsing stackoverflow. There were memes about being "professional googlers" before, so idk...
Recently, I worked as a staff member to a competitive programming contest and old teacher of mine hosted and felt saddened to see so many submissions made by GPT. The majority of these didn't even remove the comments. We kept denying the submission of what looked like AI-generated code, and they just retried with the same or similar piece of code, sometimes with the comments removed.
The sad part is that the contest didn't even have a prize. It was meant just for training and fun. I was really astonished to see it happen.
It not just devs, it all IT roles. Most of our new employees are terrible at basic networking, simple concepts in both windows and linux and just about everything else. Even some of our "mid" level guys cannot seem to fix anything without help or weeks of back and forth (I work remotely) in the end one of our team members almost always has to seem to travel to find a simple mistake that takes hours at worst then discovers 10 other things they didn't even know were broke then work on fixing those.
why IT? how about doctors, lawyers, teachers?
I think part of the problem is that people are just getting the answer without trying to process how to get to that answer. One of the aspects that I find very fascinating about AI is being able to have an ongoing dialogue back and forth on various given subjects. It's not just a transactional response. It's an in-depth conversation where I'm trying to better understand the root of the problem and I think that's really what we need to do. Ai and chat GPT are more than just a Google search. It's literally like having a person you can discuss ideas with and it's a missed opportunity when Debs are just asking for the code snippet without actually understanding the logic behind the code snippet.
As someone that's been learning software development for a year and a half for in hopes of becoming a junior dev, I've progressively started relying on chatgpt to summarize manuals or give me easy to understand explanations of concepts. It sure did help a lot with progressing through projects! But I've noticed it's detrimental to actual problem-solving for more complex problems. In scenarios where you can't use chatgpt, I'd feel stumped. I caught on to this big issue and started taking a step back to read more and think more without chatgpt. I agree with the video wholeheartedly and it has helped me bring out my issue with chatgpt into spotlight 🙏. Dont use chatgpt as an easy shortcut for learning, take the time to learn concepts to heart.
I've been working on a project lately and I wonder if I've been using AI effectively or not when it comes to learning and growth. I mean I have hardly used any code Ai has provided me other than a refactor. As for solving problems I have used it to walk me through a problem or at least brainstorming, but in the end I have to think and decide if the suggestions and walkthrough fit the bill or I have to think of a difference approach to solve a problem. It's a dilemma
How is it even possible with requirements to get a junior dev position being so high? It sounds like AI basically allowed a bunch of people who have no talent and would have zero chances to get a job in the field before to jump through the hoops.
This was an excellent and a mind blowing video. Sometimes we dragged ourselves by the common sense thought of being knowledgeable without having the merit of thinking. I found myself falling into the abyss of rely on too much on AI tools that promises act and think for me. I'm happy to watch a video that helped me to open my eyes and improve and work on myself rather than let the new AI assist doing it for me. Travis you are an amazing and inspired professional. Best of all and happy new year.
I bet there's a study that shows that there's causation between student math results and their dependency on calculators. That if you do things by hand, you have a deeper and more intuitive understanding of it.
This video is essential not just for jr.Devs, but for everyone using AI. And the "We write for ourselves" is more relevant than ever. And reading human written books/articles is actually not straight forward, you have to read between the lines. Even tutorial books have something to day that they didn't write. Don't avoid reading a fun novel, if you already read all those "Good books", keep the streak up, if you lose the streak, you open your mind to be consumed by AI.
Some of jr.devs in my team actually told me that "Indian guy video get me going faster than reading articles." and my response was "Video runs in the teacher's pace, or at most 2x. But, reading allows you to skip straight to the point and make you understand the concept, not just what letters are on screen." Of course, they kept watching tutorials.
relevant points and info. hope I utilize it
just like we have to intentionally go to the gym or go running to give our bodies what machinery took from us, we'll have to intentionally give our brains challenging problems to keep it healthy and active in a world of AI. and just like there are physically fit and unfit people today, we're going to have mentally fit and unfit people who experience life very differently from one another.
11:00 I don't think LLMs can summarize texts. I tried it, and it was just a common blah-blah on the article's topic. The facts from the article weren't there. It makes sense, of course, LLMs use the training data to answer.
The main change is the bar has become extremely low. It used to be that only those enthusiastic about programming could become programmers, and that enthusiasm is the magic sauce needed to become decent at something so difficult. Now a lot of people want to become developers for all sorts of (bad) reasons, but they have no particular interest in it, and they are sort of doomed from the start.
I love your videos they're different from other programming videos. The thinking part is on the spot, people seem to have either no time for thinking or just lazy and wants things done in seconds instead of letting it take some time and actually learn problem solving. I myself learned very quickly when AI came in the picture that my brain became dumber by copy and pasting, if u can't understand the lines you shouldn't use it, because then you can't solve the problems that comes with it.
I'm a fresher from India. And I'm not the one to read. And the problem IS the Internet and our cellphones. I've still manged learn everything except building my own projects because I'm deeply interested in ML. The main issue with my generation is the lack of time and time management due to our addiction of our smartphones. I find myself scrolling garbage on Instagram and watching geopolitics on TH-cam. All because sometimes doing stressful work seems like an active assault on my stressed brain. We also don't have mentors with time and patience to teach us the ropes. Because they themselves are addicted to smartphones. Now all we can do is make AI our mentor. Make it teach us how to do things. Because our own peers are lacking leadership and patience.
The irony of being stressed because of the crap you read on social networks
@monad_tcp it's not lost on me.
I'm from India too and what you've written hit my bones, that's so true. We were all just one able mentor away from being free of this Gen-z mess.
@@anaybaid Indeed brother. Keep preserving.
Junior dev? All roles want 5+ years of experience
Hmm, when did the AI boom start? Maybe that's why? Just thinking...
TH-cam addiction did that to me long time ago
was recommended this video for a week straight fine ill watch it
December 29th 2024:
Junior developers can't think anymore...
means:
December 29th 2034:
Senior developers can't think anymore...
*December 29th 2025
Mission report: December 16th, 1991.
If my last job was any indication, senior developers already have. It's just an extension of all the clean code nonsense.
@@pythonxz yes everyone at my job right now is a tech lead or senior in the system (/somehow), yet I am shoveling preventable garbage daily -- I am a senior dev as well if going by title.
@@pythonxz can you elaborate?
wow how ironically relatable this is is to me, in this age my attention span is shrinking so much that i literally can't do anything if its haven't nearing the due date
As a university teacher I can say that almost all students have begun to use chatbots with AI actively to complete almost all assignments in all disciplines.
Even talented students, who are capable to complete an assignment on their own without any problems and to train their mind, use AI chatbots the same way as weak and mediocre students use them.
Why? Simply, because if a weak student using an AI chatbot is able to complete an assignment quickly and pass it then why I will spend much time and efforts to do the same? Whether am I a looser?
As a result, the average level of students’ knowledge is rapidly dropping from the level of independent developers to the level of only reading and understanding the generated code, regardless of the talent level of students. And I am not sure that all of them even understand the generated code enough deeply.
This is a very serious shift in general psychology of the new young generation.
And I think, it is not good trend.
This is the exact reason why I avoid the use of GPS when I can. Jim Kwik named this phenomenon very well as digital dementia, even before the massive popularity of AI nowadays.
As an employee you get paid your worth. If you can do nothing without the chat bot, your value is "anyone willing to tell the chat bot to do it".
Well, that doesn't sound well payed.
There are 2 questions regarding this:
1) Surely many junior devs use AI for basically almost everything. And the company and the mid-senior devs might think they're better off using AI anyway. But what impact will it have in the long term? Who will replace those senior devs once they reach the twilight of their career?
2) How to read documentation? There are some frameworks or tools for which the documentation written is simply marvellous (like Django). But even then I can't seem to make too much sense about it sometimes, particularly for complex topics (middle ware, deployment etc.). How should one approach such tough topics while reading their documentation?
As a junior dev i use ai like a much faster and straight to point browser to solve a problem. If i dont understand the answer. I tell him to explain it more till i understand. I dont just copy paste and move on.
I've met more people in this industry who treat coding as just "a job" which makes great money... As opposed to people who are REALLY passionate about coding, and are willing to work their butts off to write the BEST code that they can.
Furthermore, the dispassionate money grabbers tend to write bad quality code, because honestly... They can't be asked to refine, refactor and test their code properly! They bang out crap as quickly as possible, do the littlest amount of work to at least get their code building... Normally, with a crap load of warnings... And then their done!
The future of this industry is in SERIOUS jeopardy! 😳
I've have such an issue working with the junior developer in my team. The issue to me is not that they're lazy, they actually work really hard, its that they refuse to think for themselves. I ask for a task to be completed and I get back a copy paste chatGPT answer as a PR.... Its actually quite sad and im not sure how i can tackle this issue in a work environment.
4:20 I'm calling BS on that story
This could be reduced to 5 seconds short video stating that “junior dev should daily read and write code”. 14 minutes of time lost.
This video was 💯 helpful. Listened to it three times. 🤔 maybe a few more in the future 😊
Yeah, but who is demanding that high productivity?
The generalisations in this are sweeping. We are not all addicted to cellphones. not all juniors are crap. Brain-numbing
Great advice!
Think also initiative seems to be lacking and no desire to meet deadlines. But with a strong and capable tech stack knowledge.
Sometimes just blindly following a ticket. Knowing that there will likely be a followup
I treat AI as a digital toolbox. It contains the tools you need to create the task but it is still up to you to complete the project. AI is still pretty dumb even in Software Development. I remember last year I asked ChatGPT how to create and draw a canvas in Java. The AI gave me the template to get started. I was able to get it done but I wondered if ChatGPT could create word for word in the project details. So I asked it to randomly draw a small town. The AI created it and the drawing was so bad that the grass was the sky, the windows were off, the door was somewhere else, everything was off. The AI doesn't know how to draw in Java. That's why it's a tool, not the solution to everything.
6:00 Something's a little off with a video discussing decreased attention spans (in part) interrupting itself with a minute-long ad spot. I understand the necessity, but still.
10:42 Having ChatGPT summarize an article for me is a non-starter; it deprives me of the intimate understanding of the author and their views I can get by reading and processing the whole thing.
I'm not convinced that you can use AI in a way that does not atrophy your skills. In its current form even using it like an advanced Google simplifies problems that may have multiple solutions to just spoon feeding you one way to do something. Sure you can chatgpt facts and that is probably fine, but just searching for facts is not going to give you a productivity boost. I work with people that have surface level knowledge because they rely on one AI or another for even basic things, in the short term that boosts your productivity but in the long term it does not let you build a deep mental model on how systems function. Reading the documentation and finding the why behind the answer does.
What stinks too as a junior is that when I do make something or use new skills by myself, I’ll occasionally get asked “this is good- did you ask chatgpt to do this?” . It was likely because I had consistent descriptive comments to track what I was doing- but still
I could not agree more! The thing is, the social AI-Transformation only gets worse, as the books and articles that are actually valuable and meaningful get burried under an ever growing pile of AI-generated slop! The internet is dying! And without it, our juniors will have it even harder, but perhaps also easier in a different way to fight their AI addiction.
Great Take on Adapting to Age of AI !
Well said, but this is a bigger issue than just software development. I'm going to share this as best I can.
has been happening increasingly but especially during COVID lockdowns.
To be fair most documentation is cryptic as fuck and has no examples for how to do anything,
AI is great for explaining complex things, that are impossible to find via google, because (1) it is fixated on selling stuff (sponsored links) and (2) it will always present the most mundane data first, because those are most often used, and rank higher. Finding something specific is hard.
@ruprecht9997 I've used it a lot to not completely rewrite my code but like "I'm using flask and HTML give me an example on how to create a header menu for my website" and it will spit out generic a boilerplate code that I can look at and understand how it works and then apply it to my own code
Normally I use IA like this. "There is this design decision that I have to make what are the possible approaches and what can be the disadvantages" or "Is it recommended to do it like this with these technologies or it's better to use something else?". I commonly do a "How to do it" and adapt the code given, never a copy and paste.
thanks, really great video!
Lovely video. Nicely general too, it was barely about coding.
When I have new developers joining the team (junior, intermediate, or senior), I usually give them a task where they don't need any domain knowledge and with very lax requirements - they can do whatever they want to achieve the goal. And I give them plenty time to complete the task. It's meant to test their problem-solving and technical design skills as well as creativity. Three out of four fail this test. They need specific requirements and instructions and they end up developing a bare minimum that is missing common sense functionalities.
So what are they going to do with AI?
In my case it is not like that. I use the documentation, I understand the framework and the theory. Even what steps I have to take. And I get to do what I have to do. The problem is that I barely have to write code. At most I correct it. It's weird, because I understand it but I can't start writing it. It's a problem of foundations.In my case, they also completely changed my stack...So I think it's more than normal that it happened like this, although of course it shouldn't be like this, in fact I suffer because of it, because I know that's not how things should be.
What about the corporations who are replacing us with AI to "boost" productivity and save money? imo it will backfire on them, for obvious reasons, but could you also make a video about it? I am studying, re-educating myself, in my 30s, and i am interested in your opinion on the whole subject and possible advice as to how to handle that if it comes up in an interview. Would it even come up as a subject? Thank you!