Crazy how fast AI is advancing. I didn't think companies will see this as a tool that engineers can use... I think they will view this as a replacement for engineers
i am a dev and i have no idea what the hell i'm doing companies hire devs to have somebody to scapegoat for the CEO's mistakes and mental retardness primarily and mediocre coding output as the secondary let's see how CEOs find a way to blame Devin
If software engineers get replaced, your job will be only one problem. You can expect the vast majority of other office work to go before it. And then what do you do with all those unemployed people?
@@Niblss such a good point! That’s what will make things interesting. But this isn’t a new problem. Innovation and jobs do become obsolete. And the vast majority of jobs today never would have been dreamed of back 70 years ago, ya know?
@@cody_codes_youtube I understand what you're saying but i don't think real general intelligence ("AGI"), if it comes, will be anything like past automation. Once you replace humans "generally" every new task you can come up with (and there certainly are nearly infinite tasks to create) can also be done by the AI. This is something i see a lot of people having trouble wrapping their heads around. The only questions I have about this scenario is: AGI is possible? and if so when? No one knows the answer to this, lots of Sillicon Valley folks seems to believe its imminent, or at least that's what they say, whether it's a true belief or just something to generate hype and buzz i honestly can't tell. I used to believe the LLM/Transformers tech would plateau (this was my reaction to seeing the old BERT models and stuff like GPT-3) and i still would rationally expect them to, but i'd be lying if i didn't say that all my previous expectations were wildly surpassed repeatedly and that I have no idea where we're headed or how fast. I have been working in the field for close to a decade, just so people reading this can get a better perspective of where i'm coming from
Oh I understand. And I was hoping this video would at least show that I’m not worried about this yet. It’s going to be a long time before we don’t need engineering knowledge workers. As long as you’re not pinning your hopes on your only job is being a software coder, and that’s it. At some level, we will all be operating with AI and doing more work than ever. Some random dude in a business suit will not be able to operate an AI agent without knowing the risks and ways they could destroy a business. Us software engineers will have to be the operators, knowing what is the best way to prompt and guide these technologies.
@@XJinfy no worries homie. Knowing the technology, there’s also multiple layers of things that AI is going to have to overcome in 2-5 years that, from a deep technological sense, I don’t know how they can overcome. If it helps, I have kids and I don’t worry about being able to provide for them. I’ll continue to be an engineer
The companies should realise that the entire life cycle of Corporate lies on the shoulders of people buying goods and services. If you snatch the ability to earn then you won't get the money in circulation. The layoffs and cutting down works for now but in the distant future if the people don't get the alternative to earn obviously they can never spend.
I think we always need to look an eys on news as technology is adapting like crazy fast. It's a concern also as the stability of job is unsure in the hand of AI
That’s another thing. It’ll be a long time before trust has been built fully with using these tools. Full autonomy with business data and process? That’s even further away.
@@cody_codes_youtubedo you think I should still pursue a career in technology like cybersecurity, web development, software development, data analytics and pursue a degree in IT and comp science? Because Fck man this is my first year in IT and CODING and idk if I should switch or not. Also a lot of people are saying the market is over saturated
@@Cal827 don’t listen to people, listen to job openings and technology outlooks. Data and not fear mongering. I can’t tell you what to do, but if my kids were going to college this year and they still wanted to get into computer science and coding, I would tell them to go for it. You have to enjoy it too, don’t forget that. It’s up to you, but I’m not scared for new grads. People have been saying that law school is saturated, but, guess what? We still need lawyers. We still need engineers.
Companies will adopt this.. because all they need is a copy of their code base in a sandbox environment and let devin run through it. Once uat is done all they need is to ask developers to go through the whole project and review.. after few successful reviews they can put it on a production server.. and the biggest challenge is that it will fix itself exponentially fast
You have WAAAAY too much optimism for this technology and WAAAAAY to little skepticism about real corporate systems. I don’t know who you’ve worked for, you know some things, but it’s crazy to think “sandbox UAT” is all it needs. Probably 20 out of the last 28 teams I’ve worked on deal with many applications that are integrated, on top of 3rd party sandbox environments that also are not standardized. Companies will definitely adopt this. And things will go south, very fast, and confidence will be lost quickly. So early adopters will burn their hands, resulting in many other Devin AI companies to pivot or crack down
The day we will have an AI able to develop micro saas and small apps that will be a game changer. Will humble developers who are feeling untitled to mess with people’s projects without any integrity and care.
Assuming the AI is also leaning to a more integrity and pragmatic driven purpose. There’s also the world where AI has been trained to make things MORE convoluted, because of the training data
I want to join to conversation, i was stressed out too with all AI buzz, and was panicking, but started to investigate and wanted to understand how it works. 1. It is based on mathematics probabilities, calculus and etc. so it measures and set probability, hallucinations comes from predictions 2. Based on physics and etc. seems mind blowing but realisation for me about this calmed me down a lot :) 3. SORA blew my mind too, but is based on pixels shadowing (bits) how tree looks like and set a lot probabilities again, so it can generate some images and videos based on that pixel like data. When i realised I’m not astonished yet by it, it was expected :) and last thing a ton amount of money is put on it to make that buzz :) it can of course benefit for us for sure in many ways like health and etc., but still where is big money big buzz is happening. We live in interesting future for sure! And as SWE we will need to review what code was written, because it is trained on bad data too.
You get it! Yeah it's definitely something that feels scary but isn't too bad. But also doesn't mean you can't stop following it as it grows. Thanks for the comment!
So what can I learn actually? Front end, backend,devops everything is going to be done by AI it seems, they told AI engineers will be in demand but now AI can create its own training data and tweak its learning algorithms....not only that AI can write songs,generate videos etc so someone who replaced swe with AI,its just a matter of time before his/her profession/business too is taken over by AI? So where do we find moksha/ solace? We are doomed
You still need engineers to be operators. You still need to have engineers to recognize when AI hallucinates and does the wrong thing. You still need engineers to understand the tech for the business (they aren’t watching these videos and have no idea what it all means). You still need engineers to understand and work with the existing architecture that AI might build. We are only doomed if AI (after years of real business use) gains 100% confidence from its users. We are only doomed if our governments agree that this is safe. We are only doomed if AI can satisfy all legal and regulatory requirements for our infrastructure and business practices around commerce, user data, and bulletproof records. I could go on, but you understand where I’m coming from now
@@ThomasTomiczek no one can know for sure what is hype, what is practical, and what will actually happen. So my thoughts are you shouldn’t put your life on hold thinking we are all doomed in 5 short years. Even if coding is 90% gone by then, the knowledge of how computer systems work is a valuable foundational step for any job in the future. I just refuse to be a part of the group where people throw their hands up and say, well it’s all over. Because I don’t think it is
Definitely not there yet, but clearly a full programmer replacement is on the horizon. Senior devs will have jobs for several more years yet, at the very least as the last on staff. I find it really hard to believe that there won't be something equal to the output of a CS degree and 5-ish years experience tool available about a year from now.
You’re not wrong, but you’re also not thinking about the big picture. How else will senior engineers be created? Half of the job of senior engineering is understanding the business. And if junior engineers have the ability to learn to work with Devin agents and learn along the way, then everyone wins.
@@cody_codes_youtube what's the need for senior engineers when AI will do the whole process in a few years, I guess in less than 10 years ? The only way people will be able to hold a job in the field will be by implanting a chip in their head and having GPT running in their brain 24h non stop. Doesnt seem like a nice future.
I think AI will be just like a tool for SWE and also if people should not be worry so much since in the future they will need people who can actually build those AI. Just take the AI positive change just like the internet changed our lives
I feel like suiciding cause I’m Not able to code that good.. I have no option to learn also. I love UI/Ux designing that also getting replaced by AI so how and what can I do.
@@randomfellow1483 thanks for the reply. I’m feeling better now.. I was watching many motivational videos but nothing is helping me.. Financial problems and the job everything is making my heart and mind to die.. I have no one to ask for help..
Devin and whatever AI 'software engineer replacement' they come up with is just the latest trend. Sure it will be able to code, even effectively, but it will always be a niche product used by a small number of companies (and even then, for very specific tasks). Think 3D printing, self driving cars etc.
But that is because Musk was too optimistic. The other self driving companies like Waymo are basically saying it will take a long time for self driving cars to be realised. On the other hand advances in AI like ChatGPT are just staggering. Just look at the progress the first version of ChatGpt to the latest version. And now we have Devin and Sora on the horizon…
@@rchua72 🤷 it’s all hype to me so far. It’ll be interesting when it’s public and VC money has dried up. If it maintains adoption and is cost effective, that’s what’s most interesting to me.
The society’s standard of living will go up massively thanks to AI automation. The prices of most things will go down. It’s like a second industrial revolution.
People also had concerns with cloud when it started. This will mature fast with time. Also, they have said it will assist software development, no where it claims to replace the software engineer role. However, this will have big impact to the number of people needed in next few years. May be divide by 100
No way. After 15 years of software development, when we are able to make it better and faster, we just also pushed the limits of complexity and the number of problems we solve. AI will expedite that trend, and we need more people to do these new crazy things.
@@cody_codes_youtube I'll comment something similar from my previous reply. How are we going to need more people if AI will do the whole thing ? Maybe now it can't do that, but the potential is there. Our only hope is that AI development hits a plateau.
@@jr-yn4lk my theory is that the entry point for customer web solutions will increase. With a huge spike of software applications, the need for “operators” or engineers will increase. AI will never be perfect. I think that’s a fantasy. What happens when it breaks or does something you didn’t want? What if you don’t know how to add to the system because, you’re not familiar with databases, networking, cloud computing, etc? You’ll need some educated and trained to do all that. My prediction is that engineering teams will probably get smaller, but their responsibility and scope will increase.
@@cody_codes_youtube Let's take the case of car manufacturing automation by robots. While types of cars and numbers grew during past 10 years, did net employment in this field grow significantly? When the productivity grows exponentially, development can't catch up with it beyond a certain limit. Yes, some people will be needed but we will have net reduction in employment. Also, will people need web portals and apps at this scale in future if AI can do the job for us on a voice based interaction (virtual customer service representative)?
@@mydreams8092 car automation isn’t a great comparison, because of the product and the audience. Every single business on the planet can use software and the applications are completely limitless. The more we automate in software, the more imagination is engaged and markets jump to tech. Although it can be fun to dream of the technology automating all our life, I think it’s still really far off.
I kind of gave up already. Now I'm willing to learn how to deal with hardware, which seems a job more difficult to automate. Only people that already well stablished in their careers have nothing to worry.... yet.
Super interesting you still think that! Did you see and hear about how the Devin demo was faked? Or at bare minimum was super misleading? Look into it, doomsday is much further away
@@cody_codes_youtube Well, doing some more research I'm confused. Some claim they don't let us test Devin AI. Others talk about it as if they accessed and tested its capabilities, which did look very impressive for those devs using it, as reported in a blog post titled “Here’s What Developers Found After Testing Devin AI”. Nevertheless, even if we find that Devin AI doesn’t even exist, and every user report is fake, or that in its current stage it can only automate junior tasks, AI is evolving faster and faster. At the same time, our current societal arrangement is still not ready to deal with job displacement due to advancements in technology, and I have no hope it will be better in a few years.
@@patrickjane3315 come on buddy. There is no need for that kind of talk. Humans have been through many phases of automation and streamlining of work. Doomsday sayers will always exist, but also doing more research you’ll notice that AI isn’t all that it’s cracked up to be so far. And as we look into the future, you have to ask the question of, how will this AI train itself? There is a good chance it won’t continue this growth pattern because it already went though most of the good training data for engineering. Hallucinations and mistakes will be more pronounced when we start having AI train on stuff that AI has already created. I’m just saying there is no need to throw your hands up already and believe the doomsday sayers. All that comes out of that is misery and disappointment. The most likely outcome is AI helps, but is not a Jarvis all knowing engine, and our work and collective human creative is augmented, not replaced. Having hope is the only thing that makes sense to me, until I see products or data that tells me otherwise. Even then, I’ll adapt myself.
I am so surprised that there are people out there who are still surprised by this as if no one knew that it's already available, and that it will be updated in two months that it will just be impossible to make flaws. It's such old news....
3:00 I checked that SWE-Bench is a set of about 2K github pull requests from just 12 popular python repositories. So is this benchmark representative enough? Additionally, I'm curious what were the rest 86.14% tasks where it has failed
Does not matter. Remember AI started in the modern form 2017. It gets better. Not good enough? What next year? In 3 years? In 5? Good enough as proof of concept.
I just don't see this doing much more than ChatBotGPT was doing a year ago, not going by the examples. Which could be cherry picked. As no one can test/use this it's impossible to say either way, but that does sure create a lot of hype! Also the SWE-bench benchmark seems a little unfair, to compare an Agent against a straight LLM. Plus I don't see Devin listed on the SWE-bench leaderboard!
@@cody_codes_youtube The 'to-do list' app source code150KB!, which includes unused JS libraries. I just don't understand why they don't demo building an even slightly complex app or website. Plus their website (which is awful) is advertising a SWE job vacancy! Get Devin to build an amazing website, and take the SWE position! The more you research other elements about Devin, the more it seems strange!
@@calebprenger3928 print debugging. If they built or connected an IDE for Devin, then why couldn’t we work with breakpoints and summarize the data while we hit them? Go through the code once vs. cycling 10 time to figure out the code. I was also disappointed to find out what open source projects they were using for the benchmark. It’s only 12 python apps. Which isn’t nothing, but only python? Only 12?
Hi boy, I am beginner and i am learning front end development,, what should i do now? I also have interest in digital marketing.. Is digital marketing safe than coding from AI?? Should i skip coding and convert to digital marketing???? Waiting for your reply.....
I can’t tell you what to do. I will say that learning what AI I coming out with and what tools are being created is a great start. As long as you are solving problems for businesses, then that’s where you will find demand for work
There should be one that architects solutions...devin or any kind of ai can only throw rough work at you and someone has to decide if it is correct or not. This is the work that is done by senior folks. So when senior folks are done with their term...who's next.... obviously it can't be ai.... apprenticeship will and should sustain may be it automatically types the code for recurring problems but the point to remember is some has to set a target and press the button to release the bomb
Absolutely correct. I believe we will turn until like orchestrators or operators in the future. The better you can structure the foundation and distill the needs correctly, the faster you can your code to 80% and crank through the last 20%!
Bro I haven’t even started out my career in this field and I’m already so scared. Should I even get into this field or do something else bc I don’t wanna waste my time on something that can’t make me money at all
There’s always money to be made. Coding is just a skill. There is a lot of FUD right now (fear, uncertainty and doubt). It’s only a waste of time if you learn nothing and coding is definitely something that teaches you how the internet works
I don’t think it will. It will be a messy transition, but the industry will change and new needs and jobs will open up. Keep your head up man, keep learning and adapting. That’s all you can do. Worrying about the future will only cause you pain today
@@cody_codes_youtube They want everyone at the office this bad Lol. Maybe Software engineers should quit working in their replacement. I think its time to work for ourselves. These people wont event hire entry level anyways. You need a job before getting a job this is too anoying for me.
The chart comparing success rates in solving open source project issues is a bit dishonest, because it’s comparing Devin with LLMs. Devin is not an LLM, it’s an agent system that uses an LLM. A better comparison would be to compare Devin with other agent systems, such as AutoGPT, GPT Engineer, AgentGPT etc
NOTHING is worth it on a 10 year timeframe. All jobs will be run by robots or AI - unless we hit some ceiling that totally does not look like it now and then... all jobs are worth it still.
LLM's are really sloppy and error prone. They also show little ability to create original ideas. People assume that they will get better, but that requires quality input from talented humans. Once it starts feeding on its own output, the system crumbles. ChatGPT is starting to decline in quality.
Remember all these self-driving cars incidents? And their impact on the law putting restrictions on this tech? Yeah, waiting for Devin to mess something up
i honestly feel so bad i have a master degree in cs im graduate in 3months, i don't think seniors will ever have problems but for juniors with the actual job market + ai it look terrible
It does look bad now. But you will also look better than most with the masters. The market is adjusting and gauging how much of an effect this will all have.
This is a microcosm of what I despise about VC AI hype. Only positive statistics and cherry picked performances are discussed. No mention of how much it costs to run such an amalgamation of models. No mention of security concerns. No mention of the specific interviews that it went through or why they're significant at all. They're also very vague about the actual meaning of the benchmarks at hand and whether they were maximised for. And on top of all of this the cringey marketing of it as a "software engineer". This is a mockery of the whole profession. But the real reason why it's frustrating is because it's so cool at the same time. I would love for something like this to actually work the way they advertise it. It's just not cutting it. No one fucking wants another calculator app builder or basic game of life builder(there's quintillions of code examples online for it to copy paste)..
You get it. It’s amazing for greenfield coding projects. Yes. But anything else? When hallucinations could cause real business collapse? When the interface is human language that can be misinterpreted, you’re going to have unexpected consequences
@@TrusePkay oh I wasn’t arguing. Sorry dude. Yeah the debug is pretty solid. I would think though this is a kind of lame path: they could have used an IDE and debuggers cleaner I think. But I have no idea the implementation of how Devin works
@@cody_codes_youtube Let's be serious. The people saying AI will take programming jobs don't know how programming works. As far as we still use standard I/O and AI will need new data to improve on anything, there's nothing to fear. Studies have shown building ML apps using AI generated data causes problems. So we need new real-world data every year. Data is what all these AI-scare proponents don't mention. Thus it is those who understand how to work with data that will remain relevant in the industry.
@@TrusePkay absolutely. That’s another layer of my confidence. The AI snake will probably end up eating its own tail in 3-4 years and data will get WEIRD
@@ewjkzgty I’ve spent a decent amount of time in the security space. And it sounds dramatic, but it’s kind of like a war zone. Tools that developers have now, are also available to attackers. So, as of today, there is not absolute tool that just “removes the need for security”
It's fascinating - the inflationary complacency. Let's recap A computer program can semi-autonomously complete job interviews! - response meh A computer program can earn you money in seconds vs days/weeks of work! - response meh Their retort: "But can it solve cancer? It can't! So see - there! We still need computer engineers - my response lmao
I think there is a disconnect with your experience. The engineers and developers I know ARE actively using AI. And after that experience, we are all saying “meh” It’s exciting if this pace of progress (which is not guaranteed) continues and makes our jobs easier
@@cody_codes_youtube My experience is 30 years of programming, starting with Basic and Apple IIe to more recently full-stack enterprise architecture design with AI-Agentss as log monitoring and triage systems throughout the stack and moving into RBAC for AI Agents research - no disconnect. My commentary sought to highlight the disconnect between excellent, given where I sit, automation can reduce AI engineering budget costs with one or two Human Senior Engineers co-piloting/pair programming with several AI agent development teams - the dev cycle from idea to test to market release is a tremendous boon to entrepreneurs, small businesses, and corporate America - where the human intelligence pool (never enough engineers) has just been democratized. To go a step further - Microsoft would release broken/incomplete Windows software all the time - and iterate to a completed product while retaining customers and market share... in my experience, an engineer's engineer will want perfection. Still, the market needs stop-gap releases to keep their problems at bay, and who pays the engineer but the marketplace. They've already proven that good enough is enough.
Right, that’s what I’m saying in the video. When it advances to be more mature, then that’ll be cool. So far, it’s just a little cool, but still has a long way to go
@@johnsonlin4940 yeah, and all this is like supercharging peoples skills. Not necessarily taking over anything yet. Either way, it’s interesting to watch
The probleme lies more on the demand side of SoftwareEngineers due to higher productivity and possible automation. A replacement of the complete Software Departement seems unrealistic in the future
This is analogous to how artists still have to know how to use the AI in order to get the result they want. The skill isn't manually drawing/painting anymore, but rather telling the AI what to do. After all, the real gem is in the IDEA -- what the art means, not how good it looks (I mean look at Picasso's paintings). Similarly, this Devin will just do some of the repetitive work giving software engineers more time to be innovative.
As an artist who was the earliest adopter of AI, a good amount of my time is spent on dealing with quirks and details that AI can’t get quite right, but in terms of skills I’m already lagging behind while it only gets better and better. I love the idea of hybrid pipeline when artists use technology to amplify their creativity and imagination, but majority of artists and managers seeing it as a complete artists replacement, which makes it a rather odd self fulfilling prophecy, but at least for now there’s no single package that wouldn’t make mistakes, and creativity is still exists in the minds of creators. I think this also applies to the coding AIs, let’s hope that this revolution will be rather gradual
@@CMak3r the more I think about it, the more I realize that this will take longer than expected. I think there will soon be laws around what these can do.
you're not wrong, artists will not be replaced, I will use myself as an example. I needed a logo for my course and instead of an artist I just use ideogram AI. Like that's a case where an artist lost a customer cuz of AI. The better AI gets and is easier to access by masses, the less jobs for artists in the future. Now there's that prices will get cheaper, and artists will do 10 works per day instead of '1' a day/week/month and , they will end up winning same amount of $$ either way.@@CMak3r my point was that as of right now, there is no AI available that understands human psychology cuz certain design elements manipulate the eye, and the mind of the viewer, on a certain part, to give insight for 'buying the product', or 'pressing click' , or whatever the artist wants right? That high fidelity in art cannot be done 'yet' by AI, not to speak that the AI is incapable most of the times to do a 100% complete job, it needs assist, even I have to use adobe firefly to modify some things and edit here and there in photoshop to do somewhat of an acceptable job. Def NOT a sign to be depressed for devs or artists, but I think, at what point the AI will actually be able to replace a human, 5,10,15y? 20y? Doesn't matter the answer. Cuz whatever the answer is, if you're still working 10-15y later in that domain, and you can't retire, and the AI TRULY got to that point of 100% replacing the human. Then those who didn't prepare in time their exit from that dying job they will be burned hard and end up working at mcdonald or something (nvm, fastfood will be automatized 100% in the future) RIP LUL. Until we reach that 100%, everyone will be in the future a "Supervisor" job: Artist supervisor, Software supervisor, Team Lead AI supervisor (an AI that supervise everyone in the team, and someone that supervise that AI). This is how we will have it for the next 5-20y. So my only advice, is invest in stocks, invest in new skills, invest in something that will make you either be able to retire early or... at least change jobs, get into a leadership position for e.g, or make ur own company (Etc). Doomsday is not here yet (and if they are not able to do AI hardware (Etched AI), then that prophecy will never occur)
@@CMak3r yeah that makes sense. I'm not an artist myself, but I just heard this from other artists online. Do you think in the future, when there is a way to differentiate human vs AI art, human art will be much more valuable?
@@cody_codes_youtube What are some good resources to keep up with the technologies I should be learning? I feel like what im learning in class isn’t helping me learn things I need in the industry like using git and frameworks.
The level of koolaid drinking around this topic is getting ridiculous. It shows such a lack of critical thinking and healthy skepticism that it boggles the mind. Now we even have stock phrases like “This is the worst Devin will be” floating around in discussions, and it's becoming such a cliche. Let's be real: for all we know, we’re being shown the very best it can do under optimal and controlled circumstances. It’s a possibility that this is the best Devin or its equivalents will ever be. Even in the possible scenario that it improves, it takes leap of faith to assume that AI, and LLMs in particular, will keep improving at the same pace they have. They might or they might not. We don’t know and therefore most of the doomsaying is as speculative as the naysaying.
Thanks for putting this on a top level comment! I really appreciate it. I saw you make this point on a reply as well (i think). This is such a good point.
@@dan-cj1rr i am pretty interested in compsci, not so much other types of engineering to be honest. altho i was already planning to go for something besides software engineering/programming, so i wonder how it'll turn out. thanks for answering
If you’re interested in it, then go for it. This stuff is incredibly complicated and there will be a need for people knowing how it all works. The only thing I think will change is you need to learn how these new technologies work and be able to adapt and work with them
Banning ai and it’s killing skills creative thinking.. this is scaring if all the ai company comes in one hand (singularity )and everyone with no income .. imagine our future
I think all AI tool is goo to make the joiner engineer take a good step when start he's career and increase the routeen writing task and give the engner more time to thinking for problem and possople way to solve it and AI make apply the solution more easy
@@lactobacillusshirotastrain8775 meh. Maybe. I’ve worked on 29 engineering teams now and one thing that has shown me to be true for each and every team: there was always more work to be done, too few people because of too small of a budget, and the work really never ends because the business could never be satisfied. For time to deliver, to levels of feature robustness. Reflecting on that experience, I don’t see any of these jobs satisfied even if we all 10x our output.
Coding demand will dry up eventually when an AI tool makes everybody (including average joes) able to do what a seasoned programmer is able to do. Why would I pay someone 100k salary when I get the same output by someone in Asia and pay them 20K.
If it’s supposed to be a “replacement for all software engineers” and if the story is “it’s an AI engineer”, then that’s the story. That’s my argument in the other video I had on the topic about security. A lot of people shot back that “well of course everything will be sandboxed!” But then you’re treating this “AI engineer” as more of a tool or library that you use in a closed system way… so what’s the benefit? There are a lot of variables and I think you’re right that the security side of things will be a huge deterrent from this being a popular option for companies. Therefore, that’s why I keep thinking it’ll not be a replacement for many years to come
This is fantastic and it's only the begining! I'm extremely excited about technology. Finally, we'll reach new heights. Just imagine, in 10 years, life will be completely transformed. With technology, many things will change for the better and some for the worse, but I'm truly thrilled! 🎉 Whoever believes technology won't replace humans, better wake up.
@@cody_codes_youtube I disagree and believe unless they can somehow implement serious restrictions but I don't see how they can effectively monitor AI then it will be the death of capitalism. For example, when AI can enable anyone to build their own products and services then said products and services become trivial to build and therefore worthless, completely destroying future startups. This same issue can be applied to every industry AI is applied to. Ultimately, the future leads us into a world where UBI is a must but then you have to ask, is it fair for these billionaires to keep their wealth and power while no one else has the opportunities that they had to get even close to their level of wealth? That is a problem and the only answer is for them to be stripped from their wealth which requires a total economic reset. And this is coming from someone who loves capitalism and hates socialism/communism, but nothing lasts forever and AI is causing us to choose our hand.
We're doomed, it's over. The minute I read about Devin AI yesterday, I looked up clinics in Switzerland and made an appointment for voluntary euthanasia. We've had a good run, everyone, but it's over.
Tech people were not giving a shit when they were taking away other's job of artist, copywriters. When it comes to them they are worried. It was inevitable tbh.
This is such a cliched answer by now. For all we know, we’re being shown the very best it can do under optimal and controlled circumstances, so it’s a possibility that this is also the best it’ll ever be. It takes a huge leap of faith to assume that AI, and LLMs in particular will keep improving at the same pace. They might or they might not. We don’t know.
@@alexyari6036 Do you really not see a future where companies compete to produce the best AI engineer out there. Imagine how much they can charge per month for such a service.
@@ramzan7563 it’s not a question of whether the companies invest. It’s a question of whether it can be assumed that progress will continue just because money is being thrown around. Assuming that AI will just progress as it has been is a fallacy.
Crazy how fast AI is advancing. I didn't think companies will see this as a tool that engineers can use... I think they will view this as a replacement for engineers
Haha. Then they will have their velocity go down when it’s not as easy as they think, or worse yet, the AI does something stupid a human wouldn’t do
i am a dev and i have no idea what the hell i'm doing
companies hire devs to have somebody to scapegoat for the CEO's mistakes and mental retardness
primarily
and mediocre coding output as the secondary
let's see how CEOs find a way to blame Devin
If software engineers get replaced, your job will be only one problem. You can expect the vast majority of other office work to go before it. And then what do you do with all those unemployed people?
@@Niblss such a good point! That’s what will make things interesting. But this isn’t a new problem. Innovation and jobs do become obsolete. And the vast majority of jobs today never would have been dreamed of back 70 years ago, ya know?
@@cody_codes_youtube I understand what you're saying but i don't think real general intelligence ("AGI"), if it comes, will be anything like past automation. Once you replace humans "generally" every new task you can come up with (and there certainly are nearly infinite tasks to create) can also be done by the AI. This is something i see a lot of people having trouble wrapping their heads around.
The only questions I have about this scenario is: AGI is possible? and if so when? No one knows the answer to this, lots of Sillicon Valley folks seems to believe its imminent, or at least that's what they say, whether it's a true belief or just something to generate hype and buzz i honestly can't tell.
I used to believe the LLM/Transformers tech would plateau (this was my reaction to seeing the old BERT models and stuff like GPT-3) and i still would rationally expect them to, but i'd be lying if i didn't say that all my previous expectations were wildly surpassed repeatedly and that I have no idea where we're headed or how fast.
I have been working in the field for close to a decade, just so people reading this can get a better perspective of where i'm coming from
2005: “Coal Miners should learn to Code”
2024: “Coders should learn to Mine Coal”
Woof. Going backwards!
I'm about to cry, this is my future bro, I need a software engineering job to be able to live bro.
Oh I understand. And I was hoping this video would at least show that I’m not worried about this yet. It’s going to be a long time before we don’t need engineering knowledge workers.
As long as you’re not pinning your hopes on your only job is being a software coder, and that’s it. At some level, we will all be operating with AI and doing more work than ever. Some random dude in a business suit will not be able to operate an AI agent without knowing the risks and ways they could destroy a business. Us software engineers will have to be the operators, knowing what is the best way to prompt and guide these technologies.
@@cody_codes_youtube thx bro, im ngl this kinda cleared my head
@@XJinfy no worries homie. Knowing the technology, there’s also multiple layers of things that AI is going to have to overcome in 2-5 years that, from a deep technological sense, I don’t know how they can overcome. If it helps, I have kids and I don’t worry about being able to provide for them. I’ll continue to be an engineer
Join the infantry
Learn to code.
Hooked Devin directly to my team’s JIRA board and now I work 1 hr a week AMA.
Got fired after Devin updated all customer records on our User microservice. AMA
The companies should realise that the entire life cycle of Corporate lies on the shoulders of people buying goods and services. If you snatch the ability to earn then you won't get the money in circulation. The layoffs and cutting down works for now but in the distant future if the people don't get the alternative to earn obviously they can never spend.
For sure. The circulation and spending and borrowing is what indicates a good economy.
its defintely over. what do you think will be in 5 or 10 years?
Depend on what you think is over. 50 years
it’s not over guys you have to become project managers who understand the intricacies of programming you’ll be fine.
Exactly. I prefer the term like operator because it sounds cooler. Haha
I think we always need to look an eys on news as technology is adapting like crazy fast. It's a concern also as the stability of job is unsure in the hand of AI
That’s another thing. It’ll be a long time before trust has been built fully with using these tools. Full autonomy with business data and process? That’s even further away.
@@cody_codes_youtubedo you think I should still pursue a career in technology like cybersecurity, web development, software development, data analytics and pursue a degree in IT and comp science? Because Fck man this is my first year in IT and CODING and idk if I should switch or not. Also a lot of people are saying the market is over saturated
@@Cal827 don’t listen to people, listen to job openings and technology outlooks. Data and not fear mongering. I can’t tell you what to do, but if my kids were going to college this year and they still wanted to get into computer science and coding, I would tell them to go for it. You have to enjoy it too, don’t forget that. It’s up to you, but I’m not scared for new grads.
People have been saying that law school is saturated, but, guess what? We still need lawyers. We still need engineers.
@@cody_codes_youtube thank you sir, appreciate your advice
Man of course that is the first version... thousands of companies are coming up to make the AISE even better...wait 2 months and see!
I’m not holding my breath. This feels like VC hype, and over promising.
@@cody_codes_youtube AI is here and forever it's not a hype ;)
@@iammobenal it’s here, but it’s not doing its job yet. When it does its job, that’ll be exciting.
It still only builds todo app or similar. Have not seen AI demo that builds enterprise apps.
@@johnsonlin4940 for sure. That will take a while
Companies will adopt this.. because all they need is a copy of their code base in a sandbox environment and let devin run through it. Once uat is done all they need is to ask developers to go through the whole project and review.. after few successful reviews they can put it on a production server.. and the biggest challenge is that it will fix itself exponentially fast
You have WAAAAY too much optimism for this technology and WAAAAAY to little skepticism about real corporate systems. I don’t know who you’ve worked for, you know some things, but it’s crazy to think “sandbox UAT” is all it needs. Probably 20 out of the last 28 teams I’ve worked on deal with many applications that are integrated, on top of 3rd party sandbox environments that also are not standardized.
Companies will definitely adopt this. And things will go south, very fast, and confidence will be lost quickly. So early adopters will burn their hands, resulting in many other Devin AI companies to pivot or crack down
The day we will have an AI able to develop micro saas and small apps that will be a game changer. Will humble developers who are feeling untitled to mess with people’s projects without any integrity and care.
Assuming the AI is also leaning to a more integrity and pragmatic driven purpose. There’s also the world where AI has been trained to make things MORE convoluted, because of the training data
I want to join to conversation, i was stressed out too with all AI buzz, and was panicking, but started to investigate and wanted to understand how it works.
1. It is based on mathematics probabilities, calculus and etc. so it measures and set probability, hallucinations comes from predictions
2. Based on physics and etc. seems mind blowing but realisation for me about this calmed me down a lot :)
3. SORA blew my mind too, but is based on pixels shadowing (bits) how tree looks like and set a lot probabilities again, so it can generate some images and videos based on that pixel like data.
When i realised I’m not astonished yet by it, it was expected :) and last thing a ton amount of money is put on it to make that buzz :) it can of course benefit for us for sure in many ways like health and etc., but still where is big money big buzz is happening. We live in interesting future for sure! And as SWE we will need to review what code was written, because it is trained on bad data too.
You get it! Yeah it's definitely something that feels scary but isn't too bad. But also doesn't mean you can't stop following it as it grows. Thanks for the comment!
So what can I learn actually? Front end, backend,devops everything is going to be done by AI it seems, they told AI engineers will be in demand but now AI can create its own training data and tweak its learning algorithms....not only that AI can write songs,generate videos etc so someone who replaced swe with AI,its just a matter of time before his/her profession/business too is taken over by AI? So where do we find moksha/ solace? We are doomed
You still need engineers to be operators. You still need to have engineers to recognize when AI hallucinates and does the wrong thing. You still need engineers to understand the tech for the business (they aren’t watching these videos and have no idea what it all means). You still need engineers to understand and work with the existing architecture that AI might build.
We are only doomed if AI (after years of real business use) gains 100% confidence from its users. We are only doomed if our governments agree that this is safe. We are only doomed if AI can satisfy all legal and regulatory requirements for our infrastructure and business practices around commerce, user data, and bulletproof records.
I could go on, but you understand where I’m coming from now
@@cody_codes_youtube The problem with all your statement is the "still". What about in 3-5 years?
@@ThomasTomiczek no one can know for sure what is hype, what is practical, and what will actually happen. So my thoughts are you shouldn’t put your life on hold thinking we are all doomed in 5 short years. Even if coding is 90% gone by then, the knowledge of how computer systems work is a valuable foundational step for any job in the future. I just refuse to be a part of the group where people throw their hands up and say, well it’s all over. Because I don’t think it is
Definitely not there yet, but clearly a full programmer replacement is on the horizon. Senior devs will have jobs for several more years yet, at the very least as the last on staff. I find it really hard to believe that there won't be something equal to the output of a CS degree and 5-ish years experience tool available about a year from now.
You’re not wrong, but you’re also not thinking about the big picture. How else will senior engineers be created? Half of the job of senior engineering is understanding the business. And if junior engineers have the ability to learn to work with Devin agents and learn along the way, then everyone wins.
@@cody_codes_youtube what's the need for senior engineers when AI will do the whole process in a few years, I guess in less than 10 years ? The only way people will be able to hold a job in the field will be by implanting a chip in their head and having GPT running in their brain 24h non stop. Doesnt seem like a nice future.
@@cody_codes_youtube what are devin alternatives
@@Brodragon2225 I don’t know. This seems to be the first that I know of. Still an early version.
I think AI will be just like a tool for SWE and also if people should not be worry so much since in the future they will need people who can actually build those AI. Just take the AI positive change just like the internet changed our lives
For sure. It’ll be a weird transition but I agree with you
I feel like suiciding cause I’m
Not able to code that good.. I have no option to learn also. I love UI/Ux designing that also getting replaced by AI so how and what can I do.
You won’t be. Take a breath. Things will be okay. At some point you’ll use AI to make you a better coder and designer. It will be okay brother.
Ai will open many more jobs dw bro. I’m 17 years old so I’m in one of the worst eras of the job industry but I’m tryna keep my head up!
@@randomfellow1483 thanks for the reply. I’m feeling better now.. I was watching many motivational videos but nothing is helping me.. Financial problems and the job everything is making my heart and mind to die.. I have no one to ask for help..
@@randomfellow1483 the jobs it will open will soon be replaced as well
Yes A.I will indeed make you unemployed it was inevitable, but suicide is not an answer.
It's a path of cowards.
Devin and whatever AI 'software engineer replacement' they come up with is just the latest trend. Sure it will be able to code, even effectively, but it will always be a niche product used by a small number of companies (and even then, for very specific tasks). Think 3D printing, self driving cars etc.
I think of self driving cars a lot. Especially with Musk saying “it’s 1 year away” and has said that for 6 years
But that is because Musk was too optimistic. The other self driving companies like Waymo are basically saying it will take a long time for self driving cars to be realised. On the other hand advances in AI like ChatGPT are just staggering. Just look at the progress the first version of ChatGpt to the latest version. And now we have Devin and Sora on the horizon…
@@rchua72 🤷 it’s all hype to me so far. It’ll be interesting when it’s public and VC money has dried up. If it maintains adoption and is cost effective, that’s what’s most interesting to me.
What's the thought process of engineers who build such tools? What's there end game?
Money. Automation. Promise for more business
Beautiful question, I have also wandered what they stand to derive from all these
The society’s standard of living will go up massively thanks to AI automation. The prices of most things will go down. It’s like a second industrial revolution.
Microservices don’t need to be in their own code repositories, they just need to run in their own space.
Tomato-potatoe
True.
Although I would like that path better, it does take away some of the benefits of micro services.
People also had concerns with cloud when it started. This will mature fast with time. Also, they have said it will assist software development, no where it claims to replace the software engineer role. However, this will have big impact to the number of people needed in next few years. May be divide by 100
No way. After 15 years of software development, when we are able to make it better and faster, we just also pushed the limits of complexity and the number of problems we solve. AI will expedite that trend, and we need more people to do these new crazy things.
@@cody_codes_youtube I'll comment something similar from my previous reply. How are we going to need more people if AI will do the whole thing ? Maybe now it can't do that, but the potential is there. Our only hope is that AI development hits a plateau.
@@jr-yn4lk my theory is that the entry point for customer web solutions will increase. With a huge spike of software applications, the need for “operators” or engineers will increase. AI will never be perfect. I think that’s a fantasy. What happens when it breaks or does something you didn’t want? What if you don’t know how to add to the system because, you’re not familiar with databases, networking, cloud computing, etc? You’ll need some educated and trained to do all that. My prediction is that engineering teams will probably get smaller, but their responsibility and scope will increase.
@@cody_codes_youtube
Let's take the case of car manufacturing automation by robots. While types of cars and numbers grew during past 10 years, did net employment in this field grow significantly?
When the productivity grows exponentially, development can't catch up with it beyond a certain limit. Yes, some people will be needed but we will have net reduction in employment.
Also, will people need web portals and apps at this scale in future if AI can do the job for us on a voice based interaction (virtual customer service representative)?
@@mydreams8092 car automation isn’t a great comparison, because of the product and the audience. Every single business on the planet can use software and the applications are completely limitless. The more we automate in software, the more imagination is engaged and markets jump to tech. Although it can be fun to dream of the technology automating all our life, I think it’s still really far off.
I kind of gave up already. Now I'm willing to learn how to deal with hardware, which seems a job more difficult to automate.
Only people that already well stablished in their careers have nothing to worry.... yet.
Super interesting you still think that! Did you see and hear about how the Devin demo was faked? Or at bare minimum was super misleading? Look into it, doomsday is much further away
@@cody_codes_youtube Well, doing some more research I'm confused.
Some claim they don't let us test Devin AI. Others talk about it as if they accessed and tested its capabilities, which did look very impressive for those devs using it, as reported in a blog post titled “Here’s What Developers Found After Testing Devin AI”.
Nevertheless, even if we find that Devin AI doesn’t even exist, and every user report is fake, or that in its current stage it can only automate junior tasks, AI is evolving faster and faster. At the same time, our current societal arrangement is still not ready to deal with job displacement due to advancements in technology, and I have no hope it will be better in a few years.
@@patrickjane3315 come on buddy. There is no need for that kind of talk. Humans have been through many phases of automation and streamlining of work. Doomsday sayers will always exist, but also doing more research you’ll notice that AI isn’t all that it’s cracked up to be so far. And as we look into the future, you have to ask the question of, how will this AI train itself? There is a good chance it won’t continue this growth pattern because it already went though most of the good training data for engineering. Hallucinations and mistakes will be more pronounced when we start having AI train on stuff that AI has already created.
I’m just saying there is no need to throw your hands up already and believe the doomsday sayers. All that comes out of that is misery and disappointment. The most likely outcome is AI helps, but is not a Jarvis all knowing engine, and our work and collective human creative is augmented, not replaced. Having hope is the only thing that makes sense to me, until I see products or data that tells me otherwise. Even then, I’ll adapt myself.
I am so surprised that there are people out there who are still surprised by this as if no one knew that it's already available, and that it will be updated in two months that it will just be impossible to make flaws. It's such old news....
Haha
This is exactly what I think, sure I saw one guy using ChatBot GPT API to do something similar, but this was months ago.
3:00 I checked that SWE-Bench is a set of about 2K github pull requests from just 12 popular python repositories. So is this benchmark representative enough?
Additionally, I'm curious what were the rest 86.14% tasks where it has failed
Thank you for checking that! I’ve always been curious but hadn’t googled
Does not matter. Remember AI started in the modern form 2017. It gets better. Not good enough? What next year? In 3 years? In 5? Good enough as proof of concept.
I just don't see this doing much more than ChatBotGPT was doing a year ago, not going by the examples. Which could be cherry picked. As no one can test/use this it's impossible to say either way, but that does sure create a lot of hype! Also the SWE-bench benchmark seems a little unfair, to compare an Agent against a straight LLM. Plus I don't see Devin listed on the SWE-bench leaderboard!
For sure! And also, the more I learn about this benchmark, the more it disappoints me. Interesting data point, but not very convincing.
@@cody_codes_youtube The 'to-do list' app source code150KB!, which includes unused JS libraries. I just don't understand why they don't demo building an even slightly complex app or website. Plus their website (which is awful) is advertising a SWE job vacancy! Get Devin to build an amazing website, and take the SWE position! The more you research other elements about Devin, the more it seems strange!
@@cody_codes_youtube what exactly disappoints you?
@@calebprenger3928 print debugging. If they built or connected an IDE for Devin, then why couldn’t we work with breakpoints and summarize the data while we hit them? Go through the code once vs. cycling 10 time to figure out the code.
I was also disappointed to find out what open source projects they were using for the benchmark. It’s only 12 python apps. Which isn’t nothing, but only python? Only 12?
Hi boy,
I am beginner and i am learning front end development,, what should i do now? I also have interest in digital marketing..
Is digital marketing safe than coding from AI??
Should i skip coding and convert to digital marketing????
Waiting for your reply.....
I can’t tell you what to do. I will say that learning what AI I coming out with and what tools are being created is a great start. As long as you are solving problems for businesses, then that’s where you will find demand for work
"Is digital marketing safe than coding from AI??" - THAT field is already nearly taken over by AI.
I’m waiting for devin to run up a company aws build with trail error good luck 😂
OH MY GOD YOU KNOW IT WILL HAPPEN 🤩 😂 😆
There should be one that architects solutions...devin or any kind of ai can only throw rough work at you and someone has to decide if it is correct or not. This is the work that is done by senior folks. So when senior folks are done with their term...who's next.... obviously it can't be ai.... apprenticeship will and should sustain may be it automatically types the code for recurring problems but the point to remember is some has to set a target and press the button to release the bomb
Absolutely correct. I believe we will turn until like orchestrators or operators in the future. The better you can structure the foundation and distill the needs correctly, the faster you can your code to 80% and crank through the last 20%!
Bro I haven’t even started out my career in this field and I’m already so scared. Should I even get into this field or do something else bc I don’t wanna waste my time on something that can’t make me money at all
There’s always money to be made. Coding is just a skill. There is a lot of FUD right now (fear, uncertainty and doubt). It’s only a waste of time if you learn nothing and coding is definitely something that teaches you how the internet works
we need to start an anti-ai protest to protect jobs, like how the hell are we gonna make money if ai takes over everything
I don’t think it will. It will be a messy transition, but the industry will change and new needs and jobs will open up. Keep your head up man, keep learning and adapting. That’s all you can do. Worrying about the future will only cause you pain today
LOL, literally all tech bros were laughing at artists when they were against AI.
Techbros now:
@@agus.lorenzo digital artists still have work, you know… tech bros still have work too. The work is just going to be different now
@@cody_codes_youtube They want everyone at the office this bad Lol. Maybe Software engineers should quit working in their replacement. I think its time to work for ourselves. These people wont event hire entry level anyways. You need a job before getting a job this is too anoying for me.
@@computernerd8157 true man. The market is bananas right now
The chart comparing success rates in solving open source project issues is a bit dishonest, because it’s comparing Devin with LLMs. Devin is not an LLM, it’s an agent system that uses an LLM. A better comparison would be to compare Devin with other agent systems, such as AutoGPT, GPT Engineer, AgentGPT etc
Yeah that’s also true. And the benchmark content is also has plenty to be desired
I love that a coding bootcamp ad came up on this video😂
OF COURSE!
I wonder if getting into software engineering is worth it now
I still think so. I have many reasons to believe so, but ultimately it’s up to you and your own ambition and willingness to learn
NOTHING is worth it on a 10 year timeframe. All jobs will be run by robots or AI - unless we hit some ceiling that totally does not look like it now and then... all jobs are worth it still.
LLM's are really sloppy and error prone. They also show little ability to create original ideas. People assume that they will get better, but that requires quality input from talented humans. Once it starts feeding on its own output, the system crumbles. ChatGPT is starting to decline in quality.
Yeah. I like to call it the snake eating its own tail
Remember all these self-driving cars incidents? And their impact on the law putting restrictions on this tech? Yeah, waiting for Devin to mess something up
That’s exactly the same conversation I’ve had like 20 times already. That’s exactly right. So I’ll continue being an engineer
i honestly feel so bad i have a master degree in cs im graduate in 3months, i don't think seniors will ever have problems but for juniors with the actual job market + ai it look terrible
It does look bad now. But you will also look better than most with the masters. The market is adjusting and gauging how much of an effect this will all have.
This is a microcosm of what I despise about VC AI hype. Only positive statistics and cherry picked performances are discussed. No mention of how much it costs to run such an amalgamation of models. No mention of security concerns. No mention of the specific interviews that it went through or why they're significant at all. They're also very vague about the actual meaning of the benchmarks at hand and whether they were maximised for.
And on top of all of this the cringey marketing of it as a "software engineer". This is a mockery of the whole profession. But the real reason why it's frustrating is because it's so cool at the same time. I would love for something like this to actually work the way they advertise it. It's just not cutting it. No one fucking wants another calculator app builder or basic game of life builder(there's quintillions of code examples online for it to copy paste)..
You get it. It’s amazing for greenfield coding projects. Yes. But anything else? When hallucinations could cause real business collapse? When the interface is human language that can be misinterpreted, you’re going to have unexpected consequences
@@cody_codes_youtubewhat shocks me is that it can debug itself 😂😂😂
Let's stop arguing
@@TrusePkay oh I wasn’t arguing. Sorry dude. Yeah the debug is pretty solid. I would think though this is a kind of lame path: they could have used an IDE and debuggers cleaner I think. But I have no idea the implementation of how Devin works
@@cody_codes_youtube Let's be serious. The people saying AI will take programming jobs don't know how programming works. As far as we still use standard I/O and AI will need new data to improve on anything, there's nothing to fear.
Studies have shown building ML apps using AI generated data causes problems. So we need new real-world data every year.
Data is what all these AI-scare proponents don't mention. Thus it is those who understand how to work with data that will remain relevant in the industry.
@@TrusePkay absolutely. That’s another layer of my confidence. The AI snake will probably end up eating its own tail in 3-4 years and data will get WEIRD
The first AI software engineer 😂😂😂😂😂😂😂🤣🤣😂🤣😂😂😂🤣🤣🤣😂😂
Hahahahhahahahhah
should i choose cse or ece for engg?
I really couldn’t answer that to be honest. I would say if you want to get into web dev, then doing that in side projects will be beneficial
In your opinion. I keep learning cyber security Engineering or not .I really felt worried when I heard that Ai will take place of the programmer😢
No, don’t stop. AI will never remove the possibility that bad dudes will break things.
@@cody_codes_youtube I'm encouraged by your words ❤️
@@ewjkzgty I’ve spent a decent amount of time in the security space. And it sounds dramatic, but it’s kind of like a war zone. Tools that developers have now, are also available to attackers. So, as of today, there is not absolute tool that just “removes the need for security”
It's fascinating - the inflationary complacency.
Let's recap
A computer program can semi-autonomously complete job interviews! - response meh
A computer program can earn you money in seconds vs days/weeks of work! - response meh
Their retort:
"But can it solve cancer? It can't! So see - there! We still need computer engineers -
my response lmao
I think there is a disconnect with your experience. The engineers and developers I know ARE actively using AI. And after that experience, we are all saying “meh”
It’s exciting if this pace of progress (which is not guaranteed) continues and makes our jobs easier
@@cody_codes_youtube My experience is 30 years of programming, starting with Basic and Apple IIe to more recently full-stack enterprise architecture design with AI-Agentss as log monitoring and triage systems throughout the stack and moving into RBAC for AI Agents research - no disconnect.
My commentary sought to highlight the disconnect between excellent, given where I sit, automation can reduce AI engineering budget costs with one or two Human Senior Engineers co-piloting/pair programming with several AI agent development teams - the dev cycle from idea to test to market release is a tremendous boon to entrepreneurs, small businesses, and corporate America - where the human intelligence pool (never enough engineers) has just been democratized.
To go a step further - Microsoft would release broken/incomplete Windows software all the time - and iterate to a completed product while retaining customers and market share... in my experience, an engineer's engineer will want perfection. Still, the market needs stop-gap releases to keep their problems at bay, and who pays the engineer but the marketplace. They've already proven that good enough is enough.
Broo...like come one...i just got into college for Computer Science...now this is happening ...🙃
Completely understand. I’m working on a video for this topic as we speak!
What Devin can do now is not the problem. The problem is what Devin can do tomorrow.
Right, that’s what I’m saying in the video. When it advances to be more mature, then that’ll be cool. So far, it’s just a little cool, but still has a long way to go
Exactly, Devin in a year will be wild.
Or in a year, it may build a slightly improved todo app that we have seen many times in the last 18 months that people demo what AI can do.
@@johnsonlin4940 yeah, and all this is like supercharging peoples skills. Not necessarily taking over anything yet. Either way, it’s interesting to watch
I think we’re suffering from cognitive dissonance. AI is leading us to the unemployment line.
I’m actively trying to find and understand where it’ll take place, and how it would do it. I just don’t see it. I don’t want to get blindsided!
The probleme lies more on the demand side of SoftwareEngineers due to higher productivity and possible automation. A replacement of the complete Software Departement seems unrealistic in the future
I agree. And I also think the greed and demand for more money/market share/business offerings in every single business will never be quenched
This is analogous to how artists still have to know how to use the AI in order to get the result they want. The skill isn't manually drawing/painting anymore, but rather telling the AI what to do. After all, the real gem is in the IDEA -- what the art means, not how good it looks (I mean look at Picasso's paintings). Similarly, this Devin will just do some of the repetitive work giving software engineers more time to be innovative.
For sure. You get it!
As an artist who was the earliest adopter of AI, a good amount of my time is spent on dealing with quirks and details that AI can’t get quite right, but in terms of skills I’m already lagging behind while it only gets better and better. I love the idea of hybrid pipeline when artists use technology to amplify their creativity and imagination, but majority of artists and managers seeing it as a complete artists replacement, which makes it a rather odd self fulfilling prophecy, but at least for now there’s no single package that wouldn’t make mistakes, and creativity is still exists in the minds of creators. I think this also applies to the coding AIs, let’s hope that this revolution will be rather gradual
@@CMak3r the more I think about it, the more I realize that this will take longer than expected. I think there will soon be laws around what these can do.
you're not wrong, artists will not be replaced, I will use myself as an example. I needed a logo for my course and instead of an artist I just use ideogram AI. Like that's a case where an artist lost a customer cuz of AI. The better AI gets and is easier to access by masses, the less jobs for artists in the future. Now there's that prices will get cheaper, and artists will do 10 works per day instead of '1' a day/week/month and , they will end up winning same amount of $$ either way.@@CMak3r
my point was that as of right now, there is no AI available that understands human psychology cuz certain design elements manipulate the eye, and the mind of the viewer, on a certain part, to give insight for 'buying the product', or 'pressing click' , or whatever the artist wants right?
That high fidelity in art cannot be done 'yet' by AI, not to speak that the AI is incapable most of the times to do a 100% complete job, it needs assist, even I have to use adobe firefly to modify some things and edit here and there in photoshop to do somewhat of an acceptable job.
Def NOT a sign to be depressed for devs or artists, but I think, at what point the AI will actually be able to replace a human, 5,10,15y? 20y? Doesn't matter the answer.
Cuz whatever the answer is, if you're still working 10-15y later in that domain, and you can't retire, and the AI TRULY got to that point of 100% replacing the human. Then those who didn't prepare in time their exit from that dying job they will be burned hard and end up working at mcdonald or something (nvm, fastfood will be automatized 100% in the future) RIP LUL.
Until we reach that 100%, everyone will be in the future a "Supervisor" job: Artist supervisor, Software supervisor, Team Lead AI supervisor (an AI that supervise everyone in the team, and someone that supervise that AI). This is how we will have it for the next 5-20y.
So my only advice, is invest in stocks, invest in new skills, invest in something that will make you either be able to retire early or... at least change jobs, get into a leadership position for e.g, or make ur own company (Etc). Doomsday is not here yet (and if they are not able to do AI hardware (Etched AI), then that prophecy will never occur)
@@CMak3r yeah that makes sense. I'm not an artist myself, but I just heard this from other artists online. Do you think in the future, when there is a way to differentiate human vs AI art, human art will be much more valuable?
Is it still worth it to major in CS? 😭
I think so. It’s just worth learning what’s going on in the industry.
@@cody_codes_youtube What are some good resources to keep up with the technologies I should be learning? I feel like what im learning in class isn’t helping me learn things I need in the industry like using git and frameworks.
The level of koolaid drinking around this topic is getting ridiculous. It shows such a lack of critical thinking and healthy skepticism that it boggles the mind. Now we even have stock phrases like “This is the worst Devin will be” floating around in discussions, and it's becoming such a cliche.
Let's be real: for all we know, we’re being shown the very best it can do under optimal and controlled circumstances. It’s a possibility that this is the best Devin or its equivalents will ever be.
Even in the possible scenario that it improves, it takes leap of faith to assume that AI, and LLMs in particular, will keep improving at the same pace they have. They might or they might not. We don’t know and therefore most of the doomsaying is as speculative as the naysaying.
Thanks for putting this on a top level comment! I really appreciate it. I saw you make this point on a reply as well (i think). This is such a good point.
yeah, it hitting a plateau is a possibility, and the safest one for humans. Otherwise, humanity will become obsolete in less than 20 years, I guess.
wondering if i should go on with a compsci degree with all of this happening
i wouldnt, waking up everyday wondering if youre gonna lose your job is not fun. Go in healthcare.
@@dan-cj1rr healthcare isnt really an option for me as the salary is very low here and it is much more expensive to study…
@@bloobluee the thing is that no one know the future so pursue what u like, other kind of engineering (civil, mechanical) are good also
@@dan-cj1rr i am pretty interested in compsci, not so much other types of engineering to be honest. altho i was already planning to go for something besides software engineering/programming, so i wonder how it'll turn out. thanks for answering
If you’re interested in it, then go for it. This stuff is incredibly complicated and there will be a need for people knowing how it all works. The only thing I think will change is you need to learn how these new technologies work and be able to adapt and work with them
Banning ai and it’s killing skills creative thinking.. this is scaring if all the ai company comes in one hand (singularity )and everyone with no income .. imagine our future
I’m imagining. I don’t think it’ll ever get that far
The prices of things will collapse thanks to AI automation. It will increase living standards.
Wish it bee
I think all AI tool is goo to make the joiner engineer take a good step when start he's career and increase the routeen writing task and give the engner more time to thinking for problem and possople way to solve it and AI make apply the solution more easy
For sure
Self driving cars and 3d printing will change everything they said...
I mean, there is a lot of improvements in those categories, but the point you’re making is right: we still aren’t there, are we?
garbage in => garbage out
Devin, now talk like a pirate
Devil AI
Accurate
It would make mediocre, average coders obsolete. Top tier software engineers would still keep their jobs.
Or make those mediocre coders 10x better. But they wouldn’t know the basics and rely too much on the AI magic
If that's true the massive productivity boost would cause an over supply of programmers which would cause job loss in the field.
@@lactobacillusshirotastrain8775 meh. Maybe. I’ve worked on 29 engineering teams now and one thing that has shown me to be true for each and every team: there was always more work to be done, too few people because of too small of a budget, and the work really never ends because the business could never be satisfied. For time to deliver, to levels of feature robustness. Reflecting on that experience, I don’t see any of these jobs satisfied even if we all 10x our output.
Coding demand will dry up eventually when an AI tool makes everybody (including average joes) able to do what a seasoned programmer is able to do. Why would I pay someone 100k salary when I get the same output by someone in Asia and pay them 20K.
@@lactobacillusshirotastrain8775 that can happen today, so why are there still American jobs?
Not a replacement yet. Keep calm people and protect Sarah Connor, just in case
Haha. This is probably my favorite comment so far
😂 what if they programmed it to work this efficiently only for that particular use case ,so they could get some funding
But this does looks scary
Hahaha. Thats not out of the realm of possibilities! Wouldn’t be the first time
Yeah.. long way to go from replacing real human. but will it happen in my life time? 😢
I don’t think so, really. I just think things will get really weird in the next 50 years
is it time to start a youtube channel?
Hahaha. Always an option!
you really think companies are going to give an AI root access to all their machines?
If it’s supposed to be a “replacement for all software engineers” and if the story is “it’s an AI engineer”, then that’s the story.
That’s my argument in the other video I had on the topic about security. A lot of people shot back that “well of course everything will be sandboxed!” But then you’re treating this “AI engineer” as more of a tool or library that you use in a closed system way… so what’s the benefit?
There are a lot of variables and I think you’re right that the security side of things will be a huge deterrent from this being a popular option for companies. Therefore, that’s why I keep thinking it’ll not be a replacement for many years to come
@@cody_codes_youtube "Hey Devin, we're going in a different direction. We need to be faster, drop all columns and rows for all tables."
@@darkquaesar2460 Devin: make the site just static html with no JavaScript. That should do jt
@@cody_codes_youtube Just write al your code in Rust like the white house says you should do.
@@darkquaesar2460 all of these are good ideas
This is fantastic and it's only the begining! I'm extremely excited about technology. Finally, we'll reach new heights. Just imagine, in 10 years, life will be completely transformed. With technology, many things will change for the better and some for the worse, but I'm truly thrilled! 🎉 Whoever believes technology won't replace humans, better wake up.
Should be interesting…
Is this the death of capitalism?
Hahaha. Hell no, it’ll soak capitalism in gasoline as everything gets hyped and supercharged to go to the moon
@@cody_codes_youtube I disagree and believe unless they can somehow implement serious restrictions but I don't see how they can effectively monitor AI then it will be the death of capitalism.
For example, when AI can enable anyone to build their own products and services then said products and services become trivial to build and therefore worthless, completely destroying future startups. This same issue can be applied to every industry AI is applied to. Ultimately, the future leads us into a world where UBI is a must but then you have to ask, is it fair for these billionaires to keep their wealth and power while no one else has the opportunities that they had to get even close to their level of wealth? That is a problem and the only answer is for them to be stripped from their wealth which requires a total economic reset.
And this is coming from someone who loves capitalism and hates socialism/communism, but nothing lasts forever and AI is causing us to choose our hand.
no humans, no capitalism. Doesn't seem like a good solution.
This is the birth of a world of abundance thanks to AI automation.
@@Lerppunen let’s hope so. Computers also introduced a huge level of efficiency and was predicted for us to not have to work 40 hours a week. However…
I cant wait when Devin deletes the whole Facebook data centers.
WHOOOOPSIE
We're doomed, it's over. The minute I read about Devin AI yesterday, I looked up clinics in Switzerland and made an appointment for voluntary euthanasia.
We've had a good run, everyone, but it's over.
Hahaha. This is the most wild comment so far. Wish I had known you more!
Never fear, you can pivot to animal husbandry.
I heard they are using Fugure AI new robot to deliver the euthanasia. Should be fun
Time to boycott Ai!
Meh, I still think it’s cool
That mic 😂
I brought it back!
Tech people were not giving a shit when they were taking away other's job of artist, copywriters. When it comes to them they are worried. It was inevitable tbh.
Haha. I think tech people complain more than most
What people should understand is that this is the worst this technology ever will be. AI only improves
This is such a cliched answer by now. For all we know, we’re being shown the very best it can do under optimal and controlled circumstances, so it’s a possibility that this is also the best it’ll ever be. It takes a huge leap of faith to assume that AI, and LLMs in particular will keep improving at the same pace. They might or they might not. We don’t know.
@@alexyari6036 you’re right. I think the thing is speculating on the future can get tiring. But we will just have to wait and see!
@@alexyari6036 Do you really not see a future where companies compete to produce the best AI engineer out there. Imagine how much they can charge per month for such a service.
@@ramzan7563 it’s not a question of whether the companies invest. It’s a question of whether it can be assumed that progress will continue just because money is being thrown around. Assuming that AI will just progress as it has been is a fallacy.
@@alexyari6036 you should google 'machine learning'