@@AmlalElMahrouss1 *Revelation 3:20* Behold, I stand at the door, and knock: if any man hear my voice, and open the door, I will come in to him, and will sup with him, and he with me. HEY THERE 🤗 JESUS IS CALLING YOU TODAY. Turn away from your sins, confess, forsake them and live the victorious life. God bless. *Revelation 22:12-14* And, behold, I come quickly; and my reward is with me, to give every man according as his work shall be. I am Alpha and Omega, the beginning and the end, the first and the last. Blessed are they that do his commandments, that they may have right to the tree of life, and may enter in through the gates into the city.
Our intelligence shouldn't be ranked in IQ anymore, but in costs with o3 as a baseline. For example, my coding level is worth around 12 cents per hour, because o3 would achieve 100 times as much as me with a minimum wage budget.
@@StudyMusicAssist always has been. At least in most countries, there are set values the government will subsidize in order to train a doctor or engineer for example through post secondary / course funding and coop programs.
@@ILoveTinfoilHats that's true, but that doesn't give a direct monetary correlation of their intellectual output. Not to mention (back then) IQ didn't always correlate to higher profit. But with this is seems like that's now the case
Serious question (please dont laugh even if it sounds stupid,trying to learn here) When the engineers who work in maintaining these systems die/retire...would old or legacy systems be able to find juniors in these old technologies proficient enough to replace them? I have noticed that a lot of students just hated C as a language bc of the pointers and the dynamic memory allocation issues that they had during assignments, I myself dont like c but prefer java and c#,but I have seen this trend that most CS students are choosing more modern languages like python or JS, rust or go I still havent met a single person who even knows someone who programs using COBOL....like how would these places migrate their systems if there are no proficient/senior people left? Or maybe there are seniors left but very few ppl to actually do something?
@@G2rtTr they won't find them, and will pay stupidly high wages as an incentive to learn COBOL. 👀 Paying one guy $300k/yr is probably still way cheaper than migrating away from legacy tech...
My God, do I love enthusiasm of people who think AI will become smart enough to independently run global Tech infrastructure, finance system and medicine, but it will not become smart enough to produce a robot capable of fixing air conditioner and installing pipes.
humans can run infra, finance and medicine, but so far, they cannot produce a general robot... at least not right now... anyways, 10 years down the line
@@o1-preview they can, its just not affordable therefore little monetary motivation behind it, military uses em often though, they got transport bots, rescue bots, repair bots, self piloted drones, etc etc
There many variables in the real world, which is the main issue. There are an infinite amount of variables that would need to be processed and understood even from one house to the house next door, let alone a generalized subset of "houses" to tend to. Second, it's extremely difficult to create a multipurpose robot capable of a wide array of intricate actions. It's one thing to be surgical with a scalpel in a stationary room where everything always looks the same... You seem to be conflating decision making and management with actionable positions that, while require the former, are not entirely reliant on them.
16 years in programming i still don't see my job in danger anytime soon...but i love using chatgpt for tedious work like processing json bodies or writing documentation 🙂
You should be pretty safe when taking into account that devin absolutely sucks, and that an AI-agent that can solve a complex multi file codebug seems to be nowhere near with all the technical challenges it would have to overcome
@@ibgib yes, but its not out yet, so I (and others) cant test how good it is. I would expect it to be a bit better than devin tho (which isnt that hard...)
I turned down a job to train and program AI because not only would I be hastening the downfall of humanity, I also would be training a robot to take my place. Now I just train AI models for free with my wizard YT comments.
I worked in tech support for SaaS startups for over ten years. Now, I can't even get a first interview. Guess I've always been ahead of the curve lol.. I started my own independent tech support business, and it pays well, but it's tough getting customers.
1:20 how many times does open air have to release bench marks that significantly over sell the capabilities of their model for people to not trust their bench marks anymore?
We have to wait for real world consequences of corporations spending billions on these AI Agents only to find out they can't actually fix their bugs and firing their engineers was a mistake.
Don't trust any benchmarks for any LLM. Even if done honestly (which most are not), they represent just a narrow line of questioning. It's best to test every model yourself with your own real-life problems and requests.
Here are my baseless predictions for this year also: 1- AI LLMs will slowly keep improving, no AGI for the foreseeable future 2- AI agents will have niche uses as always, security guards will still have jobs 3- Robots will slowly keep improving, not be affordable, maybe dogs. Agree with 10 years. 4- More tech layoffs, more ML jobs 5- Brain chips will continue to get better and reach more people. EEGs won't go anywhere 6- AR/VR will keep it's niche, grow slowly 7- Quantum computing won't break encryption 8- Project tractor to change C++ to Rust, not sure, might work 9- JS frameworks will become more AI friendly 10- Altcoin scams continue, microstrategy loophole will get fixed 11- Trump-Elon won't legitimize cryptocurrency, adaption will continue 12- Monopoly breaking will continue, but giants will still be dominant 13- People will self host a little more, AWS won't be significantly affected or shift their business model
for number 11, if they do, I predict an economic collapse and an east overtake of geopolitical socioeconomic factors I agree with every prediction you made, they sound very conservative, as in, nothing truly wild
I don't want a robot dog with a gun strapped to it's head. Preferably I want the gun preinstalled, built into the actual design of the robot. This isn't fallout, yet.
Right ? We want it to extend out of its eye or nose or something. That's what's wrong with the world these days - we used to have _aspirations_ y'know ?
you cant do that. We at big tech company say that we dont have enough space in the robot to integrate a gun, much less a usbc port. Make the robot bigger? Not possible, we dont know how to stretch the chassis
5:36 - I know people exactly like this :D brings back some nice memories. for example when they used teflon tubing for insulating the gold wiring inside the STM head, but after vacuuming the chamber and lowering it, the teflons broke due to being brittle at such low temperatures. vacuum was gone, had to vent and open the whole thing and bake it in an oven an clean it. takes about 3 months of non stop work.
I swear to God, I checked on your TH-cam account today morning (EAT) and wondered "Where is Jeff? It's been a while since I have seen one of his masterpiece" 😅😅😅 Speaking of the devil... 😂 Nice to have you back sir...
Where are you going after you die? What happens next? Have you ever thought about that? Repent today and give your life to Jesus Christ to obtain eternal salvation. Tomorrow may be too late my brethen😢. Hebrews 9:27 says "And as it is appointed unto man once to die, but after that the judgement
I asked o1 pro to write a simple command line program to use a particular create, and not only did it not use the crate, but the code didn't even compile.
have you consider becoming a cosmic star? they'll be fed sun light radiation... we're still 15 years away from getting to that point... maybe they'll take you as a mechanic in case of major solar storms
It's a lot more than 10x, man. Over the last two weeks, I've done 7 years of coding work. Dropping the sources, a new app and cognitive measuring framework soon.
I was in middle of a python learning session while i saw on youtube, oh cool a fireship video they are really fun to watch. Watches the video. Gets unsettled. Questions learning python. :)
One of the currently intractable problems is moving a piano and pseudointellectuals genuinely believe AGI is feasible in the near future, and even then that would just be a big math model whose calculations can be geared to solve any task potentially better than a human, but that’s not genuine thinking, a precursory reading of Searle puts that idea to rest
If it could solve any task or problem better than any human, then why would it matter whether or not it meets your own personal definition of “thinking”? Surely that’s like saying that planes aren’t actually flying because their wings don’t flap?
@@adampenbrook5751well he probably is talking about known problems in the form of done in a way by humans before, the problem is it cant actually reason, the fundamental Algorithm doesnt suggest it ever can, nor does the current o3 show any real improvement looking at how it failed at super simple things, what we will rather see is that it will be a highly useful general purpose calculator(not exactly that, but if you think about what a calculator can do i think its similar), with maybe a bit room for error, so it will probably end up as a tool for most things, but that wont stop it from deleting alot of jobs, i personaly dont think llms can realistcaly ever reach AGI on their, own or if so, the amount of data and energy required would be way more than we can reasonable expect to get anywhere in the near future(and i am not talking about a bit more than what we have currently), AGI will probably come but not from LLms. What you need to know is that they actually only mimic reasoning, so even the case mentioned is quite unlikely, but maybe they can manage to train it enough that most things can be done by it.
this argument falls apart when you realize freedom of thought has not been proven for the human mind and is still one of the biggest philosophical debates
Although you may sound like a AI sometimes, but I do cherish and love your videos, making semi-tech guy like me to grasp the nerdy development under media radar. You sum it well, cherish those human friends we can get along, and you’re one of it though we have not yet met. 😊
well spitting out too much is easy to fix with prompt engineering if you're working with the API, more of an issue with the way chatGPT is prompting the model that you're pointing out
@@DDracee I also thought that the initial 4o responses had worse reasoning than gpt4, having the gpt work more autonomously complicated the prompting process rather than simplifying, but agreed, if you are a frequent user you may well find more success with the control the API offers for when it matters.
I am a developer, I worked remotely for 1 year, but now I am unemployed due to the lack of vacancies. This topic causes me a lot of discussion because development is not easy, it takes a lot of time to get used to it and delve into the details. I have been studying for 2 years and this is my career and I am very afraid for the future and its development. I don't want to waste so much time. What would you recommend?
you better make something where you can earn money for free. since you are jobless now. why don't you try making a blog website and write a script that will scrape data from other website?
calculate how much money you can keep applying for jobs full time if you run out of money get a job not related to your field keep trying until your either do get a job in your field or live life outside of your field of study might be worth to get into academia or join the army, depending on your situation
tbf 3 years is nothing, people swap careers and "restart" often, I know plenty of people doing well who only started their current career in their mid-30s the only thing you shouldn't to do is nothing, if you absolutely want a job in that field you can still work in a different field while looking, look elsewhere too, if you aren't willing to move for that career then maybe you don't want that career that much to begin with or you can work on a solo project depending on the career, they supply a bit of income and also go great in the resume or just go back to school and get something else, not only will it give you more job opportunities, 2 degrees would also help you stand out, you aren't doing anything anyway right? the world isn't fair rn and only people who go above and beyond will get rewarded
Should us software engineers just find a new career path at this point? Between myself and others I've heard from, its almost impossible to land a job since most of the postings on popular job sites are just ghost jobs.
6:11 this meme used in your React JS video got me to switch to HTMX. I can't thank you enough. I feel like Agent Smith unplugged from the Matrix, a new AI, so to speak.
o3 was tuned (as also shown in the graph) on the same tests it was supposed to beat, so there's no general reasoning, but it's the same old stuff. The o3 tests on ARC is just smoke and mirrors for investors.
most of the AI tests are also made to check if AI is usefull at all, if you look at the example problems of the SWE-benchmark, then you realise that the results from o3 are really not that impressive...
@toocrazy4030 yep. Almost all benchamrks are done on already solved Leetcode issues, or shitty TODO apps. I can be as fast as GPT in copy pasting answers from stack overflow
Funny thing is how we thought AI would help us with doing the Stuff that are supposed to be literal chores so we can spend more time minding stuff that's important, but it seems like Tech Mega Names are trying to do totally opposite and it's not in favour of Humans lmao.
0:41 It's on January 3, 2025 or Rajab 3, 1446 AH and you're watching Fireship Videos about Amazing Future Tech Trends in 2025 on the Code Report Series.
If you see it in the context of "10,000 hours to master a skill" that makes you a master of 8 skills potentially, or an old sore loser, depending on the individual.
you know, at one point I clone the voice of an anime character and used an open source LLM to pretend to be it. it was one of the most surreal experiences I've ever had, since it was avm level of speech and a great friend, unconditional love dystopian sci fi shit
Hey, maybe 2025 is the year to do that :) I know what you mean though, I'm in my 40s, divorced, and have one friend that isn't a relative or neighbor. Definitely lost more friends along the way over the last few years than I've made.
@@sumitsingh-CANDY That's really nice, thanks. Another ongoing issue is the fact that Western nations are being colonized by India (along with every other country). I want to make friends with other White Americans who aren't degenerates, but aren't total squares. That's part of why I don't have many friends; I'm picky about who I want to include in my life. That's also part of the evil of "diversity"; we all become atomized and isolated when the people around us don't have any kind of shared culture or history.
I've been trying to convince myself there will be a fully-functional-out-of-the-box Linux desktop for as long as I've been trying to convince myself I like Primus.
0:26 Why is it called the Year of Linux? Because, Microsoft will End the Support of Windows 10 at the End of 2025 (October 14), and the Computers/Laptops that aren't Supported to Upgrade to Windows 11, will be Switched to Linux or Buyed a New PC (or Macs).
Sadly it's not true though Microsoft released that they will allow support for Windows 11 on all computers. I do wish Linux will come sooner but it doesn't seem to be the case.
Apple won’t be discontinuing the vision os, the exisiting Vision Pro will get a v2 but ultimately they will bring a vision air and vision glasses with ai integrated which is what meta is doing too.
Exactly. Among people i know i have noticed that most of the people spreading the "bye bye devs, AI will take your job soon" ideology are people who are either completely not related to IT, those who wanted to get into it but failed miserably or those who do stuff like wordpress websites using templates and think they're software developers lol
I really appreciate that the ad in the video actually gave me a free book, no strings attached. I’m not fond of being forced to give out my phone number for it, but getting a free book is pretty awesome!
you know the biggest point against LLMs is that it just generates codes it can't really think it cannot connect 2 totally different things to make something new its not really creative even with reinforcement learning but yeah most repetitive jobs can be easily replaced
There have been progressions in this. In very large LLM's some neurons were found to be 'learning' neurons. They kick in if the task at hand is very sparse in the training data or if other neurons express 'doubt'. It is not human learning yet but it does point to an area where some basic type of learning becomes real.
It's still too tricky to use for normal folks, like installing programs over a browser is still way too complicated for people with no IT experience and also most Linux distros are still a pain in the butt.
tbh it's not even an imitation of intelligence, it's just training a neural network model (these things have been around in a working state for over 30 years with a long history of research leading up to it) for a very specific task. Calling it artificial "intelligence" is a very long stretch imo
You could also say that everything I’ve learned in cognitive science suggests that intelligence or superintelligence would be something that outperforms any human in most fields and is capable of generating new knowledge. Essentially, AGI is only valuable if it can create new ideas and produce research output independently. Otherwise, it’s just task automation or logic, nothing more.
@@timbred8532 Sometimes I use DeepL, which is not an LLM but a grammar-checking tool for a decade, even though it is incapable of properly correcting Bokmål, Dutch, or German. Or do you think that just because someone is a non-native speaker, they have to write like a first grader?
@@sad_wrangler8515 I see, I mistook you for a bot. I suspected such because from a 'native' perspective, language models almost always write in a technical and didactic style. I guess translators can't translate syntax.
@@timbred8532 It may sometimes sound like that, especially for German and Scandinavian speakers, because people rarely stop to translate in their own minds, bit by bit, the word order and sentence structure of their native language. As a result, it may sound quite generic or constructed to a native speaker. Most of the time, my English is essentially a word-for-word translation from German or Bokmål into English, without much regard for proper English.
I agree with abandoning the cloud. I had the chance to take and obtain Azure fundamental certificate. And while I was trying out their demo cloud, I was surprised how expensive it was! My conclusion is that if your mid cap size company is doing well, just get computer servers and build spaces... because in the long run, like within five to 10years you are able to recoup the cost. Lastly, cloud is just like renting a machine or car. And that can be stolen and you will be responsible for it.
Your career doesn't have to suck... get the free 80,000 hours guide here 80000hours.org/fireship
let me have my agent read that for me and get back to you
@@JohnneyleeRollins
Jesus is the way, the truth and the life. Turn to him and repent from your sins today ❤️
am I the only one that's human??
@FaselGodbcho-e7w no. see, you are not human.
Im just going to hook a willow chip up to a crypto miner and earn quadrillions of dollers
As a Large Language Model myself, I support this video.
Me too!
As a thinking LLM I will think about it
Dead internet theory at his peak
@@nikhil_jadhav you are the only one I believe is a LLM bot with your fresh 1 year old account 🤖
@@AmlalElMahrouss1
*Revelation 3:20*
Behold, I stand at the door, and knock: if any man hear my voice, and open the door, I will come in to him, and will sup with him, and he with me.
HEY THERE 🤗 JESUS IS CALLING YOU TODAY. Turn away from your sins, confess, forsake them and live the victorious life. God bless.
*Revelation 22:12-14*
And, behold, I come quickly; and my reward is with me, to give every man according as his work shall be.
I am Alpha and Omega, the beginning and the end, the first and the last.
Blessed are they that do his commandments, that they may have right to the tree of life, and may enter in through the gates into the city.
I sadly passed away last year
I'm sorry to hear that, but I'm glad you LLM is working as expected
Rofl... that's so hilarious
We’re going to have to make some behavior patches though. Your LLM contract included use in 3rd party AI dating apps.
rip
Me too, several times.
I'm gay but there's too many comments so no one will ever know
You better not live in the Middle East
They will know!
You cant hide
trump gonna deport u back
We know
"the real treasure is the friends we made along the way"
sometimes we forget the basics.
Jesus is the way, the truth and the life. Turn to him and repent from your sins today ❤️
@@JesusPlsSaveMebro begging 💀
sad to see how many people made the wrong turn in life due to not having a friend at the right time.
*ChatGPT joins the chat*
The One Piece is real
We can be friends 🤣
Ah yes 2025 the year of the Linux Desktop.
This time for sure!
I user mint btw
Aint hapennung tgis year either as far as i can tell
@@hike8932 Hope it never happens.
the year of the linux desktop is always:
date | awk '{print $4+1}'
Our intelligence shouldn't be ranked in IQ anymore, but in costs with o3 as a baseline. For example, my coding level is worth around 12 cents per hour, because o3 would achieve 100 times as much as me with a minimum wage budget.
thats wild, we need LLMs to replace the sportsball players
That's depressing that human intellect can now be objectively measured in monetary terms.
@@StudyMusicAssistits better to be able to do that in order to improve better
@@StudyMusicAssist always has been. At least in most countries, there are set values the government will subsidize in order to train a doctor or engineer for example through post secondary / course funding and coop programs.
@@ILoveTinfoilHats that's true, but that doesn't give a direct monetary correlation of their intellectual output. Not to mention (back then) IQ didn't always correlate to higher profit. But with this is seems like that's now the case
We got fewer buzzwords mentioned. The world is healing indeed.
Ehh, its just that half of the buzzwords were replaced with AI.
I happy he didn't waste a lot of time talking about js frameworks
@@Flosus AI took our -jobs- buzzwords?!
how is AI not a buzzword? 😅😅
The government is trying to move C to R, meanwhile most critical US infrastructure is still reliant on FORTRAN, COBOL or AS400.
As long it is not C
C stands for Communism we cannot tolerate any of that in the army.
is AS400 a language? i thought it was a hardware/system architecture
Serious question (please dont laugh even if it sounds stupid,trying to learn here)
When the engineers who work in maintaining these systems die/retire...would old or legacy systems be able to find juniors in these old technologies proficient enough to replace them?
I have noticed that a lot of students just hated C as a language bc of the pointers and the dynamic memory allocation issues that they had during assignments, I myself dont like c but prefer java and c#,but I have seen this trend that most CS students are choosing more modern languages like python or JS, rust or go
I still havent met a single person who even knows someone who programs using COBOL....like how would these places migrate their systems if there are no proficient/senior people left?
Or maybe there are seniors left but very few ppl to actually do something?
@@G2rtTr they won't find them, and will pay stupidly high wages as an incentive to learn COBOL. 👀
Paying one guy $300k/yr is probably still way cheaper than migrating away from legacy tech...
My God, do I love enthusiasm of people who think AI will become smart enough to independently run global Tech infrastructure, finance system and medicine, but it will not become smart enough to produce a robot capable of fixing air conditioner and installing pipes.
humans can run infra, finance and medicine, but so far, they cannot produce a general robot... at least not right now... anyways, 10 years down the line
Yup, I wish I was dumb enough to believe all the silly AI hype. Seems fun :/
I mean, some new form of Ai in the near-distant future could of course arise but these LLMs won't do the trick xD
@@o1-preview they can, its just not affordable therefore little monetary motivation behind it, military uses em often though, they got transport bots, rescue bots, repair bots, self piloted drones, etc etc
There many variables in the real world, which is the main issue. There are an infinite amount of variables that would need to be processed and understood even from one house to the house next door, let alone a generalized subset of "houses" to tend to.
Second, it's extremely difficult to create a multipurpose robot capable of a wide array of intricate actions. It's one thing to be surgical with a scalpel in a stationary room where everything always looks the same...
You seem to be conflating decision making and management with actionable positions that, while require the former, are not entirely reliant on them.
As soon as Jeff isn't replaced by an AI, we are good. I'm gonna call it Jeff test.
he automated his yt back in 2023
But if he goes who will be here to tell us we're doomed?
We call him yeff here.
Haha the Arc Jeff test
he name jeff
It will be over when we hit the o7 model.
two more weeks
Going on their naming scheme, that's the next iteration 😵💫
@OriginalRaveParty it went from O1 to O3 because O2 is a trademarked name (British telecom company)
you mean the 007 agent?
Maybe o7 will finally know how to push to master.
8:57 the friends don't even pay for THEIR coffee, they sure aren't paying my bills 😭
yeah, time to start my remote forest hermit story arc and become local folklore
16 years in programming i still don't see my job in danger anytime soon...but i love using chatgpt for tedious work like processing json bodies or writing documentation 🙂
I'd take documentation written by ChatGPT over no documentation at all tbh 💀
11 years in the software industry. I want AI take my job so I don't have to do it anymore. It keeps promising but not delivering.
You should be pretty safe when taking into account that devin absolutely sucks, and that an AI-agent that can solve a complex multi file codebug seems to be nowhere near with all the technical challenges it would have to overcome
@toocrazy4030 have you seen Jules?
@@ibgib yes, but its not out yet, so I (and others) cant test how good it is. I would expect it to be a bit better than devin tho (which isnt that hard...)
Feels so good to know we all gonna lose our jobs 😃
About time
Hated it anyway. Just waiting for UBI to kick in the next few years
@jld-ni3vf Thinking UBI would ever happen or stick. Is like thinking remote work would stay lol..
I turned down a job to train and program AI because not only would I be hastening the downfall of humanity, I also would be training a robot to take my place.
Now I just train AI models for free with my wizard YT comments.
I worked in tech support for SaaS startups for over ten years. Now, I can't even get a first interview. Guess I've always been ahead of the curve lol.. I started my own independent tech support business, and it pays well, but it's tough getting customers.
1:20 how many times does open air have to release bench marks that significantly over sell the capabilities of their model for people to not trust their bench marks anymore?
We have to wait for real world consequences of corporations spending billions on these AI Agents only to find out they can't actually fix their bugs and firing their engineers was a mistake.
benchmarks*
Don't trust any benchmarks for any LLM. Even if done honestly (which most are not), they represent just a narrow line of questioning. It's best to test every model yourself with your own real-life problems and requests.
@@krasserTerror i think your comment glitched and ended up in the wrong thread...thanks yt
2024 was actually the year of the Linux Desktop for me and I don't regret it.
This is crazy. So much information in that video, and the sarcasm is priceless! Thank you!!!
What sarcasm??
@@CaptTerrific It's almost impossible to tell but assuming the Fireship guy knows what he's talking about there was a lot of sarcasm in that vid
My gf unfortunately did not survive 2024. RIP Bri, I hope wherever you are you found the happiness you couldn’t here
bruh
My condolences.
My condolences,
My condolences
My condolences
Ah you're alive! That's a relief. Was missing you...
automated server that creates the videos was offline, took him a while to ssh into the server and reboot it
@@o1-preview lost the ssh key to the server in the garage, cant reach it from bed
@@o1-previewhow did he SSH in if it was offline 🤔
@@industrial-wave Yeah.. He had to run his AI agent with verbal commands, to white hat hack into his ssh key.. because beds are comfortable
5:55 I'm a clown connoisseur and a prominent circus visitor, but that TRACTOR thing rewriting C/C++ into Rust with a help of LLMs got me good ngl
In general when watching podcasts or videos like this I have to speed them up by at least 1.25x but not with you Jeff, you're awesome, keep it going!
Here are my baseless predictions for this year also:
1- AI LLMs will slowly keep improving, no AGI for the foreseeable future
2- AI agents will have niche uses as always, security guards will still have jobs
3- Robots will slowly keep improving, not be affordable, maybe dogs. Agree with 10 years.
4- More tech layoffs, more ML jobs
5- Brain chips will continue to get better and reach more people. EEGs won't go anywhere
6- AR/VR will keep it's niche, grow slowly
7- Quantum computing won't break encryption
8- Project tractor to change C++ to Rust, not sure, might work
9- JS frameworks will become more AI friendly
10- Altcoin scams continue, microstrategy loophole will get fixed
11- Trump-Elon won't legitimize cryptocurrency, adaption will continue
12- Monopoly breaking will continue, but giants will still be dominant
13- People will self host a little more, AWS won't be significantly affected or shift their business model
Everything gud but 10 years for robot normalisation is a bit more
for number 11, if they do, I predict an economic collapse and an east overtake of geopolitical socioeconomic factors
I agree with every prediction you made, they sound very conservative, as in, nothing truly wild
"more tech layoffs, more ML jobs".... does not compute
I believe quantum computing will break encryption, but quantum computing won't be really working for this year
In a nutshell: "things will change without changing".
I don't want a robot dog with a gun strapped to it's head. Preferably I want the gun preinstalled, built into the actual design of the robot. This isn't fallout, yet.
Right ? We want it to extend out of its eye or nose or something. That's what's wrong with the world these days - we used to have _aspirations_ y'know ?
why not transformers? a gun that can become a doggo
you cant do that. We at big tech company say that we dont have enough space in the robot to integrate a gun, much less a usbc port. Make the robot bigger? Not possible, we dont know how to stretch the chassis
Dude...thank you for pimpin 80,000 hours. That's a damn good cause, and they deserve more time in the sun.
5:36 - I know people exactly like this :D brings back some nice memories. for example when they used teflon tubing for insulating the gold wiring inside the STM head, but after vacuuming the chamber and lowering it, the teflons broke due to being brittle at such low temperatures. vacuum was gone, had to vent and open the whole thing and bake it in an oven an clean it. takes about 3 months of non stop work.
I swear to God, I checked on your TH-cam account today morning (EAT) and wondered "Where is Jeff? It's been a while since I have seen one of his masterpiece" 😅😅😅
Speaking of the devil... 😂 Nice to have you back sir...
Bro found his yt password
nope, he just took a while to notice the server was offline, fireship automated his videos back in 2023
It was cracked by a quantum computer
?????
most original joke ever
Hey V-Sauce! Micheal here?
Wake me when robot cats have katana's
Where are you going after you die?
What happens next? Have you ever thought about that?
Repent today and give your life to Jesus Christ to obtain eternal salvation. Tomorrow may be too late my brethen😢.
Hebrews 9:27 says "And as it is appointed unto man once to die, but after that the judgement
“Katana is!” 😂
how?! no one will be alive to wake you up. the last thing we'll hear is the sound of a sword leaving it's sheath accompanied by an 8bit-purr. 😐
fuck, now I want a song about robot cats with katana's or maybe even a video game with that in mind
ah yes a kenshi player
2:11 I heard "Whoever can best sell these asians is going to make it killing" 💀
Nahh not the Shein child employees being layed off as merchandise 💀💀💀
Same
love your vision, the faster we create noJS, only Rust based browser, the faster crabs 🦀 will take over this dirty frontend landscape
first fireship video of the year, released 60 minutes ago! let's go! I hope 2025 is amazing, hopefully this doesn't age badly
I survived 2024
Same here 😅
I didn't....
Shame
And all I got was this lousy t-shirt.
The 4 most accurate profets we had pointed to 2025 as the end of this civilization
Díky!
This is no joke, the only TH-cam channel that I allow notifications for.
Happy new year, I’ve missed you lol
Real
I asked o1 pro to write a simple command line program to use a particular create, and not only did it not use the crate, but the code didn't even compile.
If I have to raise chickens and sell hotdogs to robot workers I will
have you consider becoming a cosmic star? they'll be fed sun light radiation... we're still 15 years away from getting to that point... maybe they'll take you as a mechanic in case of major solar storms
@@o1-previewAt this point, I'll just report you as spam lol
It's a lot more than 10x, man. Over the last two weeks, I've done 7 years of coding work.
Dropping the sources, a new app and cognitive measuring framework soon.
Thanks
I was in middle of a python learning session while i saw on youtube, oh cool a fireship video they are really fun to watch. Watches the video. Gets unsettled. Questions learning python. :)
One of the currently intractable problems is moving a piano and pseudointellectuals genuinely believe AGI is feasible in the near future, and even then that would just be a big math model whose calculations can be geared to solve any task potentially better than a human, but that’s not genuine thinking, a precursory reading of Searle puts that idea to rest
"Cursory" big brain, you mean "cursory".
If it could solve any task or problem better than any human, then why would it matter whether or not it meets your own personal definition of “thinking”? Surely that’s like saying that planes aren’t actually flying because their wings don’t flap?
@@adampenbrook5751 indeed, but don't expect an intelligent reply, fragile egos don't like getting out performed
@@adampenbrook5751well he probably is talking about known problems in the form of done in a way by humans before, the problem is it cant actually reason, the fundamental Algorithm doesnt suggest it ever can, nor does the current o3 show any real improvement looking at how it failed at super simple things, what we will rather see is that it will be a highly useful general purpose calculator(not exactly that, but if you think about what a calculator can do i think its similar), with maybe a bit room for error, so it will probably end up as a tool for most things, but that wont stop it from deleting alot of jobs, i personaly dont think llms can realistcaly ever reach AGI on their, own or if so, the amount of data and energy required would be way more than we can reasonable expect to get anywhere in the near future(and i am not talking about a bit more than what we have currently), AGI will probably come but not from LLms. What you need to know is that they actually only mimic reasoning, so even the case mentioned is quite unlikely, but maybe they can manage to train it enough that most things can be done by it.
this argument falls apart when you realize freedom of thought has not been proven for the human mind and is still one of the biggest philosophical debates
Although you may sound like a AI sometimes, but I do cherish and love your videos, making semi-tech guy like me to grasp the nerdy development under media radar. You sum it well, cherish those human friends we can get along, and you’re one of it though we have not yet met. 😊
I still prefer gpt 4 to o1 and o1 mini. The newer models just spit out way too much and I mostly use ChatGPT for fixes and small functions.
it greatly depends on the task, some gpt4 gets it wrong, other o1 gets it
well spitting out too much is easy to fix with prompt engineering if you're working with the API, more of an issue with the way chatGPT is prompting the model that you're pointing out
@@DDracee I also thought that the initial 4o responses had worse reasoning than gpt4, having the gpt work more autonomously complicated the prompting process rather than simplifying, but agreed, if you are a frequent user you may well find more success with the control the API offers for when it matters.
This is the only YT channel I actually get excited for the ads👌👌.
I am a developer, I worked remotely for 1 year, but now I am unemployed due to the lack of vacancies. This topic causes me a lot of discussion because development is not easy, it takes a lot of time to get used to it and delve into the details. I have been studying for 2 years and this is my career and I am very afraid for the future and its development. I don't want to waste so much time. What would you recommend?
you better make something where you can earn money for free. since you are jobless now. why don't you try making a blog website and write a script that will scrape data from other website?
calculate how much money you can keep applying for jobs full time
if you run out of money get a job not related to your field
keep trying until your either do get a job in your field or live life outside of your field of study
might be worth to get into academia or join the army, depending on your situation
@@o1-preview wow AI now replies to comments. Anyway yeah joining army is a great choice - it's a win win, and is great for future
The hell are you talking about? There's developer job openings everywhere
tbf 3 years is nothing, people swap careers and "restart" often, I know plenty of people doing well who only started their current career in their mid-30s
the only thing you shouldn't to do is nothing, if you absolutely want a job in that field you can still work in a different field while looking, look elsewhere too, if you aren't willing to move for that career then maybe you don't want that career that much to begin with
or you can work on a solo project depending on the career, they supply a bit of income and also go great in the resume
or just go back to school and get something else, not only will it give you more job opportunities, 2 degrees would also help you stand out, you aren't doing anything anyway right?
the world isn't fair rn and only people who go above and beyond will get rewarded
Should us software engineers just find a new career path at this point? Between myself and others I've heard from, its almost impossible to land a job since most of the postings on popular job sites are just ghost jobs.
We were thinking that by 2025, improvements in tech will skyrocket and here we are with barely improved technologies. AI is really stagnating.
Yah...AI is already taking over basic freelance jobs ... I'll pray for it to stay stagnant
AI solving frontier math tasks is stagnating? ASI might be around the corner.
@@Sibeliu nah...ASI could take at least a decade
Hey Fireship, happy new year! Cheers for a less dystopic 2025
It's over for you, Jeff. We all now have audio of you saying Hawk Tuah. It's about to go down
6:11 this meme used in your React JS video got me to switch to HTMX. I can't thank you enough. I feel like Agent Smith unplugged from the Matrix, a new AI, so to speak.
o3 was tuned (as also shown in the graph) on the same tests it was supposed to beat, so there's no general reasoning, but it's the same old stuff. The o3 tests on ARC is just smoke and mirrors for investors.
most of the AI tests are also made to check if AI is usefull at all, if you look at the example problems of the SWE-benchmark, then you realise that the results from o3 are really not that impressive...
Why did you lie?
@toocrazy4030 yep. Almost all benchamrks are done on already solved Leetcode issues, or shitty TODO apps. I can be as fast as GPT in copy pasting answers from stack overflow
Best Regards from F'n'F Family ☝🏼🤘🏼👊🏼
Funny thing is how we thought AI would help us with doing the Stuff that are supposed to be literal chores so we can spend more time minding stuff that's important, but it seems like Tech Mega Names are trying to do totally opposite and it's not in favour of Humans lmao.
Either there is no investment money for automating house chores or the metrics for house chores cant be cheated.
Its almost like they are for profit companies and important stuff is more saturated with money. Welcome to capitalism.
As a Large Language Model i am sad to announce that my seeder past away last year, but i still care to learn about Fireship.
0:41 It's on January 3, 2025 or Rajab 3, 1446 AH and you're watching Fireship Videos about Amazing Future Tech Trends in 2025 on the Code Report Series.
Up
Bro u always comment wheb he says the date am i right
ngl the end of this decade (5 years out) looks kind of grim
Why? Curious
Inflation, less white collar jobs available and suppressed wages?
Thanks for the website link really helpful!😊😊
I love the word "entirely free" at 4:10
not many sponsored content are entirely free
@@Nate-tp1go Free... Yet also able to pay sponsors...?
...Did nobody remember the Honey scandal just happened
Super happy to see 80000hours sponsor this video!
Summarising your entire professional career as 80000 hours is unimaginably terrifying
If you see it in the context of "10,000 hours to master a skill" that makes you a master of 8 skills potentially, or an old sore loser, depending on the individual.
Can’t live without the code report. That is my main source of reliable forecasts as an ASI
the friends in the end of the vid, are AI generated friends.
you know, at one point I clone the voice of an anime character and used an open source LLM to pretend to be it. it was one of the most surreal experiences I've ever had, since it was avm level of speech and a great friend, unconditional love dystopian sci fi shit
“Real treasure is the friends we make along the way…” ❤️👏
And if you haven't made a single friend along the way? 😢
Hey, maybe 2025 is the year to do that :)
I know what you mean though, I'm in my 40s, divorced, and have one friend that isn't a relative or neighbor. Definitely lost more friends along the way over the last few years than I've made.
I have no objection in making friends with you guys 😂
@@sumitsingh-CANDY That's really nice, thanks. Another ongoing issue is the fact that Western nations are being colonized by India (along with every other country). I want to make friends with other White Americans who aren't degenerates, but aren't total squares. That's part of why I don't have many friends; I'm picky about who I want to include in my life. That's also part of the evil of "diversity"; we all become atomized and isolated when the people around us don't have any kind of shared culture or history.
Always awesome videos. Happy 2025
I've been trying to convince myself there will be a fully-functional-out-of-the-box Linux desktop for as long as I've been trying to convince myself I like Primus.
Prediction: we'll see AGI _before_ we see a Linux my grandma can use as a daily driver.
@@anonymes2884 why not linux mint?
@@anonymes2884 gta 6 releases before we see great grampa daily driving linux
sick video as usual man !!
Video starts at 00:00
Thank you
holy shit, o4 confirmed
"the real treasure are the friends we make along the way". Keep it up guys happy 2025!
2:37 (which literally cannot even push to main if you tell them to LMFAOO)
You mean master?
Devin my beloved
A couple of those trends actually give me hope, which is a big surprise to me. 😅
0:26 Why is it called the Year of Linux? Because, Microsoft will End the Support of Windows 10 at the End of 2025 (October 14), and the Computers/Laptops that aren't Supported to Upgrade to Windows 11, will be Switched to Linux or Buyed a New PC (or Macs).
You're Correct dude!
Sadly it's not true though Microsoft released that they will allow support for Windows 11 on all computers. I do wish Linux will come sooner but it doesn't seem to be the case.
@@Gozints They finally have to admit that those hardware "requirements" for Windows 11 were nonsensical.
@@couldntfindafreename ye LOL absolute scumbags man
Watching these videos somehow makes me both more and less confident in my software engineering degree path. Thanks, Fireship 👍
ai gon flatten, gpt5 will be the limit for decades
2025 will be the [Y]ear [O]f [L]inux [O]n the desktop, i can feel it
nah man, its the year of bsd
most of us will say goodbye to windows 10
Honestly it's getting pretty good. For me MacOS or Linux for personal used. Windows is gross af
Apple won’t be discontinuing the vision os, the exisiting Vision Pro will get a v2 but ultimately they will bring a vision air and vision glasses with ai integrated which is what meta is doing too.
only non-devs think chat bots (LLMs) will replace them.
Exactly. Among people i know i have noticed that most of the people spreading the "bye bye devs, AI will take your job soon" ideology are people who are either completely not related to IT, those who wanted to get into it but failed miserably or those who do stuff like wordpress websites using templates and think they're software developers lol
I really appreciate that the ad in the video actually gave me a free book, no strings attached. I’m not fond of being forced to give out my phone number for it, but getting a free book is pretty awesome!
Wish I could be there when you discover Project Gutenberg...
@@anonymes2884 Or the countless free pdf sites
you know the biggest point against LLMs is that it just generates codes it can't really think it cannot connect 2 totally different things to make something new its not really creative even with reinforcement learning but yeah most repetitive jobs can be easily replaced
There have been progressions in this. In very large LLM's some neurons were found to be 'learning' neurons. They kick in if the task at hand is very sparse in the training data or if other neurons express 'doubt'. It is not human learning yet but it does point to an area where some basic type of learning becomes real.
The beginning, I caught that. I worked with Jeff Blue on a couple songs back in the day.
0:01 "my names Jeff"
Perfect way to start a year (watching code report)... love it🎉
It won't be the year of the Linux desktop for the next 10 years either at least. I use Linux everyday and the user experience is very far from ideal.
It's still too tricky to use for normal folks, like installing programs over a browser is still way too complicated for people with no IT experience and also most Linux distros are still a pain in the butt.
Thanks for the XAI217 update! I am loving my XAI217!
at this point it is still imitation of intelligence
tbh it's not even an imitation of intelligence, it's just training a neural network model (these things have been around in a working state for over 30 years with a long history of research leading up to it) for a very specific task. Calling it artificial "intelligence" is a very long stretch imo
Jeff you still making your portfolio?
Love the thumbnail featuring a Tom Cardy song (Mixed Messages)!
I know it's a sponsor but glad to see 80,000 Hours on here. Super important that smart people seek out things that better the world ♡
Big shoutout for mentioning XAI217 and SUI. These two look like the future to me. Holding both tight!
Your voice seems AI generated in this video unlike your older video's
yes, the way he said hawk tuah sounded so artificial
Mmm all his videos do sound like that though. He’s even addressed it
he automated his videos in 2023...
@@subyob felt same
As always, love your videos, let's hope for a great year for everyone!
Your AGI benchmark is an AI model that can craft a AAA game title like GTA 6?
You could also say that everything I’ve learned in cognitive science suggests that intelligence or superintelligence would be something that outperforms any human in most fields and is capable of generating new knowledge. Essentially, AGI is only valuable if it can create new ideas and produce research output independently. Otherwise, it’s just task automation or logic, nothing more.
@@sad_wrangler8515 what model was used to generate this reply
@@timbred8532 Sometimes I use DeepL, which is not an LLM but a grammar-checking tool for a decade, even though it is incapable of properly correcting Bokmål, Dutch, or German. Or do you think that just because someone is a non-native speaker, they have to write like a first grader?
@@sad_wrangler8515 I see, I mistook you for a bot. I suspected such because from a 'native' perspective, language models almost always write in a technical and didactic style. I guess translators can't translate syntax.
@@timbred8532 It may sometimes sound like that, especially for German and Scandinavian speakers, because people rarely stop to translate in their own minds, bit by bit, the word order and sentence structure of their native language. As a result, it may sound quite generic or constructed to a native speaker. Most of the time, my English is essentially a word-for-word translation from German or Bokmål into English, without much regard for proper English.
I’m glad the Apple Vision Pro came out because it made Meta update the quest3 to support windows in a similar way
cool what fresh horrors are coming next
Your segues are brilliant.
Waiting for your rust course man
I agree with abandoning the cloud.
I had the chance to take and obtain Azure fundamental certificate. And while I was trying out their demo cloud, I was surprised how expensive it was!
My conclusion is that if your mid cap size company is doing well, just get computer servers and build spaces... because in the long run, like within five to 10years you are able to recoup the cost. Lastly, cloud is just like renting a machine or car. And that can be stolen and you will be responsible for it.
I used to pray for times like this! New Video!! Happy new Year!
yooo lets go 80000 hours sponsor is crazy - such high impact non profit