Not to mention ‘Web3’, ‘AR’, ‘Metaverse’, and other hypetrains SWE influencers have been jumping on in recent years to get quick ad revenue. The SWE job market post-2022 has been atrocious, ‘senior’, ‘senior’, ‘senior’ is literally all I see nowadays.
Well if Devin is successful. Next step would be Melvin who does management work with much better insights for shareholders. People hoping to replace developers must be afraid as well.
I've been a developer for over 20 years and I've seen many hypecycles. This is just another one. What will happen is that an astonishing amount of money will be spent on AI. It won't really achieve what is promised because the promises are always way over the top. Then lots of money will be lost and some AI tools will survive and be really useful to developers. In the video he mentions problem solving skills and that is correct - devs have great problem solving skills. Not only that but they can go and have conversations with product owners, users, etc. and really understand the problem. Additionally they can understand what edge cases and exceptions might exist, because they're human. AI can't do any of that. Also what needs to happen if AI is supposed to replace the devs it means that users will need to use the AI as a tool to produce their solutions. This is something that users hate doing. I've seen this for over 20 years when tools come along that are supposed to replace devs, the users won't engage cause they hate it. Maybe 1-2% will engage and they are usually called super-users. These people are doing this anyway (and already) because it interests them but for the large majority of users - they have no interest in getting into the technical side of things because they don't consider it as part of their job and they don't like it. So really this whole AI hypecycle is exactly that. It just surprises me these days how these hypecycles are created and how nearly everyone jumps on board like this has never happened before. Blockchain, metaverse, cryptocurrency, NFTs anyone?
@@polaris1985 well, it's achieved better results than before and it's probably here to stay. But like many times before when some new came around, people thought it would be the end of the world and they would lose their jobs, the new thing just became a new tool to make people more efficient
You can’t compare AI to a single other “hype cycle” it’s not like it’s just going to hit a wall. That’s not how technology works. It will keep progressing and it will eventually be able to do anything any person can do. This is a CERTAINTY. The only thing is how long it will take but to say it WONT is just not true. There’s no ceiling.
@@flamakespark chatGpt can run code itself as well. same goes for gemini. it's just that chatGpt and Gemini does that in server and doesn't show us what they are doing. but devin is showing. it's like that particular machine is it's own server where devin can use terminal and browser. Gpt4 and gemini advanced is better than devin currently.
I think that accountants was the same when Excel was coming, but we still have accountants today. It's not about coding, it's about solving problems, coding is a tool.
even if AI is just a tool like Excel and not another industrial revolution, then less devs would be able to do more tasks using these ai tools for example one senior developer could do the job of a team of 10 less experienced developers so far less developers will be employed. this is what I'm afraid of because it will lead to massive layoffs and unemployment for programmers and developers
@@lone-warrior-13 Decimation of entry and mid level jobs. Exactly like happening with artists in the game industry, illustration, musicians, video editors, cashiers, writers, actors, accountants... even singers!... Lots of fields and professions, not just programmers. If that's the fear, then the amount of social issues this would generate, would need (it'll happen) a global solution or regulation (to avoid, among other things, riots; history always tells us that economies never develop well with continuous revolts, economic huge unbalance, and consumption stall). Regulation which I firmly believe will come ...not "one" law, but lots of complex regulations and systems to compensate the whole thing. Among _many_ other things, because societies and economies do need consumers (for the rich staying rich, indeed), and in big numbers. That cannot be provided by a majority formed by jobless people. And I am not referring here to the UBI (universal basic income) thing as a solution (or not only that). That is not in the interest of the powers that be, the large companies and govs.... as it kills the system, and so , their income. It's a complex social, economic and politic problem, it is not just technical. And the solutions Altman proposes (the "utopia") are only good on the technical aspects, but he is very wrong about the human and social side of things, and he might have not read much about history. So... for now I would just focus on keep doing what we know and can do, improving, like always. It is too son to make so many predictions. And the time focusing solely on predictions (or taking bad decisions due to those) is IMO time lost, while there are many "unknown variables" as to be hyper focused on that future problem. If we've learnt something in the past years is that predictions are hard to make for everyone, in every field.
A senior dev told me, "I don't worry about AI and etc. I love to code and that's what I focus on. I focus on getting better and not distracted" I agree... become better. Most developers I hear are quitting or losing motivation. Doesn't that mean you should get better and ahead of the ones leaving?
But in some ways it sounds like "be a better horse groom" when cars came out to the world... I'm just a scared man hasn't even started his IT career yet
@@Yohanjin0925 Ai will kill off bad programmers for sure because bad programmers are the ones who copy and paste. Become a master... but how? Pick one language and get good at it, learn about CS - data structures and algorithms, build projects. Then any and every language is the same just syntax is different. I use to be scared too but honestly, we have no idea what will happen - good nor bad. Recently a lot of articles are stating that there will be an AI bubble soon so it doesnt look so good for AI "taking over". Programmers who never made any project are scared but programmers who have yrs of experience are not... so I rather listen to experienced programmers. Become better, fears will just hold you back, just get better with a proper plan.
@@HarpreetSingh-jd3tu Thanks... I am studying software engineering and I really like to code... I was just starting out on learning front web development, but then there is a new framework that makes components easier to make by just copying and paste and it makes me kinda scared... But I will just keep being focused on getting better
Thank you for your efforts! I think you are right, as usual, I stopped looking at AI videos "taking software engineering jobs" because my users usually ask me to create something out of thin air, like "do what I need, not what I'm saying" and they always forget something in the design phase
@@e-jarod4110 kinda messy code just install all of those things black box suggested. Too many conflicts like it already install code mirror then some parts he uses Monaco something like those. The problem here is the ai generating solutions without the consistency in mind. When it can't produce a short solution it will suggest some libraries and some of those are just doing the same.
Wow, this is a great video! Science-based, evidence-based, and you dissected those statistics really well, revealing at least a hint of bias, and possibly even some data manipulation, which, in science, is always a bad sign! I’m an MD with a great passion for web development, and extra trained in evidence-based medicine and statistics (and I teach those at a university too) and exactly this is what we are teaching students to look out for if a representative comes to you with information about a new drug. And I see so much of that in here as well. They prove what they want to demonstrate, there is no insight into how all these statistics are calculated, and again, those are very red flags to be wary of what the author/vendor is trying to promote. If it would be data manipulation (which is impossible to say without peer review), it is misleading at best. Kyle, this is exceptionally well-done work! And just you, taking this all apart, is just a perfect showcase of how AI can be helpful but not more than that. Critical thinking skills, reviewing materials, and generating new ideas: AI is (at this point) no match for a trained human brain. I think this should be assurance to many developers, but also other people in other sectors that fear for their jobs in the matter of AI. We should welcome it as a tool, just like you use a calculator, but it can’t replace a human being that is trained. Great video, great critical review, and great reassurance! Thank you!
I'm sorry, but no. This entire video is cope. And its also very myopic. Even if Devin (or GPT-4) can't take your job today, it doesnt mean that Devin 2 or Devin 3 wont (or whatever other name it or its competitors will have). By downplaying the actual capabilities of AI systems you are not making yourself any favours.
@@TheManinBlack9054even thought I believe AI will take the jobs I will answer to you with the same logic: You have no evidence neither that the curve is exponential and that we didnt or arent close to roof. That just your faith.
@@ginbarato1178 what other further evidence does he need while the proofs for a skyrocket rate of progression are all around us..? - And it's only the beginning...
@@wishmeheaven I am sorry but extraordinary claims need extraordinary evidence. The "exponentiallity" of the curve has not been proven yet, it is natural to believe it, but it is just a belief. I wish it was true, but it stays a belief, no different than thinking we will have flying cars.
@@TheManinBlack9054 and you probably haven't seen all the other hypecycles to compare this with or are ignoring them. Sure the tool will get better but it can't be human. There will be swing towards this tool but eventually the industry will swing back when they realise it can't replace humans.
When developers say that AI will replace them, I want to inform you that not one model (Google, Anthropic, Open AI) does not know what the :has selector is and a bunch of new CSS features.
AI MIGHT help in some cases for things that can be wildly automated, but not per use specific cases. Jee, I saw many newbs using it for „coding“ and they are always asking me why this and that is not working. Well… AI might help and assist in certain things but it always will fall on one logical premise: shit in, shit out. If 80 % of articles about something is full of crap (ie. Russian propaganda), it will take it without logical reasoning like people do (well, some of them) and will claim it as a FACT. Same for code. And gee, there is a LOT of crap out there… Not just code, anything. Just on youtube, try to find some serious info about anything you want. Influencers who are doind advertisement, crappy clickbaits etc.
Correct. But...that's NOT all. Devin and other AI tools STILL DO NOT know how to even parse let alone make the best decisions from FULL CONTEXT using "EXISTING" docs of what frameworks like Next.js are fully capable of. Not even at 10%. How does that make them play to take dev jobs away from human developers? Don't believe the hype. Keep on coding and becoming expert in your area. Companies who fall for this and make bad hiring/not hiring decisions will lose out big.
@@PartneredBrands These are just tools, if they haven’t been taught, they don’t know how, and there are so many tasks that it’s not enough to just feed all the code from GitHub or stackoverflow. Moreover, every 3 months we have updates in support of new browser functions, which means programming becomes easier, but AI developers did not teach this to the full extent (except for providing documentation with a description) because they have few cases to explain to AI how they Using this or that function, we solved the problem easier and, in general, how we can use it. And I’m talking only about web development and even only about the frontend... but how much do we have in total? Backend, applications for iOS, Android, specialized software, etc. Honestly, it’s hard to call it AI, it has no thought process, just an algorithm, do it this way and nothing else! And let’s be honest, not a single company has the capacity to create AI, just a database with a random answer to a question, remember the demo where they ask how much 2+2? he answers 4, the answer is 6, oh, sorry, you're right, 6...
Devin is the not the point, future Devin is the point, and the exponential rate of AI is improving is what you should be worried about. I don’t understand how people are so short sighted. I’m sure companies are thinking the same thing, they are not looking at Devin 1.0 but 3.0….
@@navis462 I agree. Obviously in its current state, Devin is really not intimidating. If it is capable to take our jobs now, then why are we still working as software engineers? The future and the exponential growth of AI is the point. However having said that, I am still not quite sure how long will it take for it to be able to replace us. So called 'self driving' cars has been around for almost 10 years and today we're nowhere near for a full self driving cars that can be adopted by the masses. AI Software engineers are gonna get better, but the real question is how long will it take
I was really hoping that the AI hype cycle was largely over with when we started the new year... but man does it seem to be in full swing again. 🤦♂ I really appreciate you digging into this and providing some insight and perspective. Keep up the great work!
It was always going to arrive and grow, to suggest it was going to be a short term thing would be laughable. Its a game changer. It'll change the world as we know it. Its already doing that, though its in its infancy stages. Just look at how the World Wide Web and the Internet changed the world. This is going to do the same thing.
We're only at the beginning of a fast-evolving AI industrial revolution. From now on, multiple new industries will be hit every year until 70% of population is unemployed.
@@Lykkos-321 glad to see someone awake ! Absolutely spot on... The technology is already there, waiting to be released. The ignorance of people is going to come to an end very very soon. We have let psychopaths at the top get away with too much sh*t without facing any consequences. They are after us and they want it all unless we say NO.
8:32 Just look at the technical complexity of the prompts themselves: "...that initializes two mod_constraints a{1, 5} and b{1, 10} and then computes mod_constraint r = a & b. Then, check that: r.mod is the LCM of a.mod and b.mod. r.v % a.mod == a.v (and likewise for b), and r.v is between 0 and r.mod." I mean, really? Testing libraries are so declarative these days, that if you're going to write a prompt like that, you might as well just write the test yourself!
Not worried about Devin or the next few versions of whatever comes next. I do think in about 5 years tools 100x more advanced than Devin will take our jobs. 2 years ago this was not a topic now we are talking about replacing coders.
The moment I saw news about this Davin thing I went to their website and asked Davin “what model do you use?” And Davin said “Yes, I'm using ChatGPT as my base model. However, I have been customized for specific tasks and goals by the user who created me.” So shrugs. What a bubbly space GenAI is right now.
I must say your video is still the best comparing it to other videos that have talked about Devin. You have said it all. Thank you very much for taking your time ❤❤❤
Dear Kyle Cook , I wanted to take a moment to express my sincere appreciation for the insightful commentary you provided on the Devin AI. Your depth of understanding and thoughtful analysis truly resonated with me. ❤❤❤
At first glance, I got the same vibe we got another tool to boost our productivity like a copilot and get. Anyway, time to find out what's gonna happen next. Thanks
We have WordPress and Webflow that competes web programmers for sometime now but it just add jobs rather take our coding job. And as long us people are afraid of code, we safe.
Reminds me of my professor at university (Translation Studies) who in 1997 claimed that adequate language translation could never be done by a machine.... ;-)
And his claim is still true, the best AI still can't produce human like translations with the nuances that we have in our native speech, you don't need a professional human translator to spot an AI generated translation. Not to mention literary translations which still needs light years for AI to be good enough to compete with us.
@@mo_1023 Well, in my eyes that is irrelevant, because practically the translations of AI nowadays are good enough for 99% of the cases that you need it. Also: the nuances that you are talking about - only the extremely good translators are really capable of translating them very well. So what you are basically then saying is: An average translator will never be as good as a genius translator. Well, i would agree on that one. ;-) PS: Also, the professor was way more radical - he would have never believed in the tranlsation possibilities that exist today. He did not even believe in low quality translations by a machine.
@@mo_1023 Another side note: I have written and published a poetry book in 2009. Now when I use AI today to translate the poems - the quality is better than I myself could produce them. For sure there might be a translator genius that would find even better translations. But there would be way more average translators who would barely reach the quality of AI translations.
This is a good perspective. I'm glad someone pointed this video out to me, because right now, RUclips has it on *Page 29* of the results when I search for "Devin" and "AI" and I would never have found it otherwise. It should really rank a lot higher than a bunch of the "woe is me" doomscroll clickbait garbage that's currently above it.
Humans have a difficulty talking and communicating with humans..... I literally was asked to take off the link surrounding a card, how are people supposed to nav to the page the card references. They respond there is a button. So I remove the surrounding link and ensure the button gets a link instead, then I get now we can't navigate to our pages, turns out they don't necessarily have to fill out the button on the CMS....
I suggest devs do some basic AI machine learning/data scientists courses. The hype will make sense. Especially narrow AI and what Transformers can actually do and what they’ve been made for.
I don't think most software engineers are concerned that Devin will take their jobs. I think they're concerned that Devin is one more step towards an inevitable future where the need for human developers will descend. If Devin or other AI assistants are at their worst now, should we not be concerned about where they'll be a couple years from now?
The biggest problem with AI is it is not creative and I realized this when trying to find "hacks" to bypass certain restrictions. It was incapable of "thinking outside the box". An example, Chrome has actively tried to prevent developers from disabling the form auto completion suggestions. There are lots of threads on stackoverflow discussing hacks to trick the browser into not filling out certain fields. We have an address auto suggestion we need customers to use because it verifies addresses and they aren't entering their own address so trying to pre-populate it with previously used addresses is not desired. I could not get it to come up with any unique solutions like people do on stack overflow. AI is incredibly useful for pieces of code and makes me far more productive. It's incredible for things like, "I have a list of values like ______ and need to filter out all values that contain numbers." It allows me to focus on the big picture and not solving stupid stuff like that where I may have forgotten the syntax.
Cognition Labs team is legendary. A team full of Math and iOi Olympiad gold medalists and Codeforces Grandmasters. These people are the top of the top of their fields in competitive Olympics level mathematics and programming. I trust them. I do believe AI will one day maybe soon replace developers by a large margin and thats not that bad of a thing. It seems most people saying its not a big deal or try not to be scared are just in denial.
What people fail to realize about Devin and similar systems is that most of the power is in the model itself. Devin and other agentic systems are simply getting more out of the model that power them. When we get GPT5 and similar models these systems will become far more powerful simply by swapping out the model powering it
You have helped me with many videos on web development in the last year and you have great credibility with me. Some code youtubers seems to just want to milk the most money they can from their viewers, whichever the most efficient way is, not caring about truth or ethic, spreading missinformation and surfing on hype talking without much thinking about whatever the topic is. "Oh yeah boys, look at this, programming is over, as an ex-faang I say, next year you are unemployed". It's kinda disrespectful with their viewers Its good to have a good source of information to clarify better so much doubts, at least a more honest source of information. Thanks WDS/Cayle
exactly, as a dev myself I will pay for such tool, even though these AIs are meant to be general I find them very useful in tests, pretty much nothing more.
we still have to remember how tremendously fast ai is growing - just look at the huge jump from will smith eating spaghetti to sora in 1 year. and even though devin today may look like a cute silly kid trying to do serious stuff, I can't even imagine what it might be able to do 1 year later. but instead of being afraid of ai or being skeptical about it what we actually need to do is to consider ai as a tool and learn how to use it to be a of a higher skill and value ourselves. because even if ai won't be able to replace you, a guy who can use it definitely will :D
software engineer here - I’m responsible for building and innovating software that ended up replacing me. So I can’t sit here and complain. All we do is innovate. Our time is up 😅
Its not the present that scares me but its the future of how advanced ai could really get in the upcoming years which does. Just like how the era of smartphones took over, and there have been constant advancements since, its exactly how I imagine this ai situation to go. And this is coming from a computer science major who will be entering their third year soon.
In my experience, the Problem in Software Development is not the Coding but the requirements. I've hardly ever met a Requirement Engineer who was good at their job. People just suck at telling clearly what they want. And this Problem will continue with AI because even HAL 9000 needs to know what it is you want him to do.
The evaluation on 25% dataset was probably because the other 75% was used for training the AI. We don't know how exactly cognition chose those 25% data but I think they did multiple training cycles through different subsets until they yield the highest result which is like the technique used in machine learning called k-fold cross-validation
That or understand the fact that the moment software engineering is automated pretty much all the other jobs can and will be automated. If the majority of the people are out of work then companies don't have anyone to do business with, along the chain of B2B companies at the end there is a B2C. Consumers are always at the end of the chain and if the majority of the consumers are broke then it creates a domino effect. "He who suffers before it is necessary, suffers more than is necessary"
@@_b001 I'm not sure what you're getting at. Are you trying to say that because AI revolution will create a dystopian world, the technological advancement will actually be stopped?
I really don't think that AI will replace whole software team ever if software development became that much easy that we only need to write prompt and it get developed tested and hosted automatically. Then why the hell I need an any third party paid software I can make it on my own by just writing prompt it'll collapse whole market then everyone have their own version of softwares . And also if it became that much easy so it's good for us we can launch new startup everyday we just need to imagine an idea
Current "AI" is just a supercharged search engine. Real AI if it ever gets to that point should be able to update itself, once that happens we can fear for sure.
I can't locate anywhere where they specify the architecture - a big important question that is largely unanswered: "is it dependent on api calls to external ai models/services, or is this it's own fully trained model and proprietary?". With how limited the info is, for all we know (*and should be very mindful of) this is a fancy UI wrapping a bunch of api calls and a scheduler of some sort. Even the wording they use in their video and THEIR website are very careful not to specify. For example "Introducing Devin, the first AI software engineer". If it were revolutionary, one would think they would be stressing this aspect to that effect; where in their site/vid, they completely dance around it. Also, with other teams out there with the tools, infrastructure, and highest caliber of engineers in the field...and they haven't yet cracked it, should send up red flags that these guys 'have done it'.
These kinds of tools can never replace or take a job from a skilled developer. But it will definitely take a job from a non skilled developers. All companies will start to expect a highly skilled developers and people like beginners or non skilled will struggle to get a new job. Not everyone will be a skilled developer. Every one can become a skilled developer if given a proper training or experience but The company always needs a skilled one. This is definitely a problem
Having said that, AI will play a very important role in a developer's life. You've done all of it: setting up a project takes all those steps that can and should be automated. Generating unit tests. Etc. and etc. Those tools are invaluable and save time. AI is just a progress in technology (who even writes a quick-sort algorithm nowadays?). So learn to adopt AI.
25% sounds like a hold out test set. They likely trained it on 75% and evaluated on the rest. That’s pretty standard for ml model evaluation and it likely was run on multiple train test splits and reported the best one. If the difficulty of the problems isn’t evenly distributed between train and test they probably are reporting an inflated number
Devin is mildly capable today but the question that should be asked is what is the rate of progress going forward? Are we hitting a plateau or will improvements continue at the same rate?
It reminds me how some people actively trying to optimize their MQL strategies over the strategy tester and wonder, why results good on history, but don't predict future.
The transition from human developers to AI agents will be gradual, taking maybe up to a decade or more. One thing is for absolute sure, human developers are on the way out and the AI agents will be the norm in the relatively near future.
Irrespective of the AI, you should be here for the love for tech and be paasionately working in it, things will progress and get better, if you got the game then you will stay here as long as you play it right, forget about you getting replaced just keep coding out of love and then you'll find problems and you creative thoughts will do the rest! Happy coding! 😊
I am no longer concerned about AI taking my job. At all. I've been at programming for about 18 months now. I had very little exposure to coding prior to that. My company found itself in a position where it needed some services created to replace things being deprecated by 3rd party vendors and I volunteered. I found some free time last week and wanted to get a head start on an upcoming project so I thought to test these AIs. While I didn't have access to Devin, I tried everything else I could get my hands on. They all failed horribly, not because they are bad per se but because they couldn't really understand the intricacies of what I needed for this particular project. Each of them gave me a good head start. None of them even came close to delivering the finished product. I don't see where Devin is going to be any different.
My software engineering tasks are so complicated to the point where my brain melts. I have hard times to shape up local solutions and even harder it is for me to explain the problem in this small context window typing it up. AI looks to me hopeless at this point. And it don't know exactly and this is the biggest problem with it. Sometimes it is enough to just show you a few things so you'll figure out the rest, but it could be very unrelated. Sometimes it could make a fine formula and assist in a thought process. But nothing is guaranteed. How can it become a full developer? By marketing! It makes everything seems better!
I remember laughing at Will Smith eating spaghetti less than a year ago.... then Sora. I'm not really worried about Devin, it's Devin 5.0 that scares me and based on the current rate of progress, that's not coming in some mythical, far off future my kids will have to worry about, it's likely coming somewhere before my next performance review. It's done for junior devs and whooooole load of other industries.
Isn't it time for making a fullstack course Kyle ? I know your focus has shifted towards frontend, but going fullstack will open a whole new world to those of us who want a web developer job, please man, market's lacking courses on this subject.
Thanks for the video, I'm one of your supporters who must push back a bit on your view on this. Sure you've experienced this replacement debate over your career before, but this is different. I'm saying this as a boomer, this is different. Things are changing, in a different way, the average 30 year old with an engineering degree can't easily buy a 4000 sqft home in suburbs like i did. What they're calling immigration today have never been seen before in this country. And AI will change the software development landscape significantly, that's my assessment. I don't have answers or advise for the younger people, I just know this is different.
You're describing late stage capitalism. The ability to not afford homes is at the feet of land owners and land Lords. "What they call immigration" That's a cute racist dogwhistle.
I'm not worried about today or the next 5 years. What worries me is what will happen to my carrer in 10 years, if these big techs keep spending so much money on this. We know what the end goal is, we know what theyre after. And if theyre spending this much money, going this much fast, they will get there.
ChatGPT and Al won't replace you but will make you 10x productive and successful in what you do. Al-assisted employees will be the future and a requirement in most of upcoming jobs in future. Ride the way to future and start learning it now
Between layoffs, inflation, and AI fearmongering, the past 2 years have been fucking exhausting.
how about the War in Ukraine and million death??
@@youtube.master4766 yeah that too, the past 2 years have been fucking exhausting.
Not to mention ‘Web3’, ‘AR’, ‘Metaverse’, and other hypetrains SWE influencers have been jumping on in recent years to get quick ad revenue. The SWE job market post-2022 has been atrocious, ‘senior’, ‘senior’, ‘senior’ is literally all I see nowadays.
dude, the past TWO years? the society has been dying for at least like 5-6 years already. we're rotting from the inside.
Well if Devin is successful. Next step would be Melvin who does management work with much better insights for shareholders. People hoping to replace developers must be afraid as well.
AI cannot take my job, if I don't have one!
I start to understand NEETs.
😢😢😢😢😢😢😢😢😢
Software as a service !
been a NEET for around 4 years now and i'm doing pretty well mentally and financially thank god.
@@timalk2097 What's this NEET?
@@codegrind1NEET - Not in Education, Employment, or Training
I've been a developer for over 20 years and I've seen many hypecycles. This is just another one. What will happen is that an astonishing amount of money will be spent on AI. It won't really achieve what is promised because the promises are always way over the top. Then lots of money will be lost and some AI tools will survive and be really useful to developers. In the video he mentions problem solving skills and that is correct - devs have great problem solving skills. Not only that but they can go and have conversations with product owners, users, etc. and really understand the problem. Additionally they can understand what edge cases and exceptions might exist, because they're human. AI can't do any of that. Also what needs to happen if AI is supposed to replace the devs it means that users will need to use the AI as a tool to produce their solutions. This is something that users hate doing. I've seen this for over 20 years when tools come along that are supposed to replace devs, the users won't engage cause they hate it. Maybe 1-2% will engage and they are usually called super-users. These people are doing this anyway (and already) because it interests them but for the large majority of users - they have no interest in getting into the technical side of things because they don't consider it as part of their job and they don't like it. So really this whole AI hypecycle is exactly that. It just surprises me these days how these hypecycles are created and how nearly everyone jumps on board like this has never happened before.
Blockchain, metaverse, cryptocurrency, NFTs anyone?
you are too optimistic, maybe AI is the real deal... what if it is!
@@polaris1985 well, it's achieved better results than before and it's probably here to stay. But like many times before when some new came around, people thought it would be the end of the world and they would lose their jobs, the new thing just became a new tool to make people more efficient
You can’t compare AI to a single other “hype cycle” it’s not like it’s just going to hit a wall. That’s not how technology works. It will keep progressing and it will eventually be able to do anything any person can do. This is a CERTAINTY. The only thing is how long it will take but to say it WONT is just not true. There’s no ceiling.
Devin is basically chatGPT in a while loop.
But with terminal, browser and IDE built-in
Bro devin needs it because how it will get errors and also run commands and write code etc @@flamakespark
False and false, it s a nanoGPT wrapper which from my knowledge it's at GPT 2 level... So yeah, you do the math.
@@flamakespark chatGpt can run code itself as well. same goes for gemini. it's just that chatGpt and Gemini does that in server and doesn't show us what they are doing. but devin is showing. it's like that particular machine is it's own server where devin can use terminal and browser. Gpt4 and gemini advanced is better than devin currently.
@@eduardmilea2013Andrey karpathy enjoyer spotted 🧏♂️🤫
I think that accountants was the same when Excel was coming, but we still have accountants today. It's not about coding, it's about solving problems, coding is a tool.
Not really comparable, AI in itself is a new industrial revolution - Microsoft Excel was not.
even if AI is just a tool like Excel and not another industrial revolution, then less devs would be able to do more tasks using these ai tools for example one senior developer could do the job of a team of 10 less experienced developers so far less developers will be employed. this is what I'm afraid of because it will lead to massive layoffs and unemployment for programmers and developers
@@lone-warrior-13 Decimation of entry and mid level jobs. Exactly like happening with artists in the game industry, illustration, musicians, video editors, cashiers, writers, actors, accountants... even singers!... Lots of fields and professions, not just programmers. If that's the fear, then the amount of social issues this would generate, would need (it'll happen) a global solution or regulation (to avoid, among other things, riots; history always tells us that economies never develop well with continuous revolts, economic huge unbalance, and consumption stall). Regulation which I firmly believe will come ...not "one" law, but lots of complex regulations and systems to compensate the whole thing. Among _many_ other things, because societies and economies do need consumers (for the rich staying rich, indeed), and in big numbers. That cannot be provided by a majority formed by jobless people. And I am not referring here to the UBI (universal basic income) thing as a solution (or not only that). That is not in the interest of the powers that be, the large companies and govs.... as it kills the system, and so , their income. It's a complex social, economic and politic problem, it is not just technical. And the solutions Altman proposes (the "utopia") are only good on the technical aspects, but he is very wrong about the human and social side of things, and he might have not read much about history. So... for now I would just focus on keep doing what we know and can do, improving, like always. It is too son to make so many predictions. And the time focusing solely on predictions (or taking bad decisions due to those) is IMO time lost, while there are many "unknown variables" as to be hyper focused on that future problem. If we've learnt something in the past years is that predictions are hard to make for everyone, in every field.
Also controllers. The advanced version of an accountant who can look and plan ahead as well :)
@@lone-warrior-13 In economies increasing value production per person is only a good thing. It means that now your salary is even more worth it.
A senior dev told me, "I don't worry about AI and etc. I love to code and that's what I focus on. I focus on getting better and not distracted"
I agree... become better. Most developers I hear are quitting or losing motivation. Doesn't that mean you should get better and ahead of the ones leaving?
like someone said, "i dont need to be better than AI, i just need to be better than others to keep a job" lol
But in some ways it sounds like "be a better horse groom" when cars came out to the world... I'm just a scared man hasn't even started his IT career yet
@@Yohanjin0925
Ai will kill off bad programmers for sure because bad programmers are the ones who copy and paste.
Become a master... but how? Pick one language and get good at it, learn about CS - data structures and algorithms, build projects. Then any and every language is the same just syntax is different.
I use to be scared too but honestly, we have no idea what will happen - good nor bad.
Recently a lot of articles are stating that there will be an AI bubble soon so it doesnt look so good for AI "taking over".
Programmers who never made any project are scared but programmers who have yrs of experience are not... so I rather listen to experienced programmers. Become better, fears will just hold you back, just get better with a proper plan.
@@HarpreetSingh-jd3tu Thanks... I am studying software engineering and I really like to code... I was just starting out on learning front web development, but then there is a new framework that makes components easier to make by just copying and paste and it makes me kinda scared... But I will just keep being focused on getting better
Thanks WDS! This video has made me a little less worried
Thank you for your efforts! I think you are right, as usual, I stopped looking at AI videos "taking software engineering jobs" because my users usually ask me to create something out of thin air, like "do what I need, not what I'm saying" and they always forget something in the design phase
That was a perfect way to explain it
I just wanna tell you my experience in ai. One developer decided to use black box in his job and now we're starting from scratch lmao
What happened? XD
Yup, this pretty much sums up the future.
What's a "black box in his job"?
@@bestopinion9257 black box is like some free chat gpt web for generating code. It's kinda hype for fresh grads
@@e-jarod4110 kinda messy code just install all of those things black box suggested. Too many conflicts like it already install code mirror then some parts he uses Monaco something like those. The problem here is the ai generating solutions without the consistency in mind. When it can't produce a short solution it will suggest some libraries and some of those are just doing the same.
Wow, this is a great video! Science-based, evidence-based, and you dissected those statistics really well, revealing at least a hint of bias, and possibly even some data manipulation, which, in science, is always a bad sign! I’m an MD with a great passion for web development, and extra trained in evidence-based medicine and statistics (and I teach those at a university too) and exactly this is what we are teaching students to look out for if a representative comes to you with information about a new drug. And I see so much of that in here as well. They prove what they want to demonstrate, there is no insight into how all these statistics are calculated, and again, those are very red flags to be wary of what the author/vendor is trying to promote. If it would be data manipulation (which is impossible to say without peer review), it is misleading at best. Kyle, this is exceptionally well-done work! And just you, taking this all apart, is just a perfect showcase of how AI can be helpful but not more than that. Critical thinking skills, reviewing materials, and generating new ideas: AI is (at this point) no match for a trained human brain. I think this should be assurance to many developers, but also other people in other sectors that fear for their jobs in the matter of AI. We should welcome it as a tool, just like you use a calculator, but it can’t replace a human being that is trained. Great video, great critical review, and great reassurance! Thank you!
I'm sorry, but no. This entire video is cope. And its also very myopic. Even if Devin (or GPT-4) can't take your job today, it doesnt mean that Devin 2 or Devin 3 wont (or whatever other name it or its competitors will have). By downplaying the actual capabilities of AI systems you are not making yourself any favours.
@@TheManinBlack9054even thought I believe AI will take the jobs I will answer to you with the same logic:
You have no evidence neither that the curve is exponential and that we didnt or arent close to roof. That just your faith.
@@ginbarato1178 what other further evidence does he need while the proofs for a skyrocket rate of progression are all around us..?
- And it's only the beginning...
@@wishmeheaven I am sorry but extraordinary claims need extraordinary evidence. The "exponentiallity" of the curve has not been proven yet, it is natural to believe it, but it is just a belief.
I wish it was true, but it stays a belief, no different than thinking we will have flying cars.
@@TheManinBlack9054 and you probably haven't seen all the other hypecycles to compare this with or are ignoring them. Sure the tool will get better but it can't be human. There will be swing towards this tool but eventually the industry will swing back when they realise it can't replace humans.
This channel is really great, Kyle has a talent to simplify complex topics. great job man!
Really enjoyed this video and how you went about explaining the reasons Devin can't replace programmers. Love from Nigeria 🇳🇬❤
are you a prince?
@@btm1Yes Prince of his family . King of his grandmother.
@@btm1 😂
great video. It's the first one I've seen that doesn't just hype everything up but provides a critical view by just going over all claims.
When developers say that AI will replace them, I want to inform you that not one model (Google, Anthropic, Open AI) does not know what the :has selector is and a bunch of new CSS features.
AI MIGHT help in some cases for things that can be wildly automated, but not per use specific cases. Jee, I saw many newbs using it for „coding“ and they are always asking me why this and that is not working. Well…
AI might help and assist in certain things but it always will fall on one logical premise: shit in, shit out. If 80 % of articles about something is full of crap (ie. Russian propaganda), it will take it without logical reasoning like people do (well, some of them) and will claim it as a FACT. Same for code. And gee, there is a LOT of crap out there… Not just code, anything. Just on youtube, try to find some serious info about anything you want. Influencers who are doind advertisement, crappy clickbaits etc.
Correct. But...that's NOT all. Devin and other AI tools STILL DO NOT know how to even parse let alone make the best decisions from FULL CONTEXT using "EXISTING" docs of what frameworks like Next.js are fully capable of. Not even at 10%. How does that make them play to take dev jobs away from human developers? Don't believe the hype. Keep on coding and becoming expert in your area. Companies who fall for this and make bad hiring/not hiring decisions will lose out big.
"Hey, this technology is not working correctly nowadays, so it never ever will in the foreseeable future, guys."
@@PartneredBrands These are just tools, if they haven’t been taught, they don’t know how, and there are so many tasks that it’s not enough to just feed all the code from GitHub or stackoverflow. Moreover, every 3 months we have updates in support of new browser functions, which means programming becomes easier, but AI developers did not teach this to the full extent (except for providing documentation with a description) because they have few cases to explain to AI how they Using this or that function, we solved the problem easier and, in general, how we can use it.
And I’m talking only about web development and even only about the frontend... but how much do we have in total? Backend, applications for iOS, Android, specialized software, etc.
Honestly, it’s hard to call it AI, it has no thought process, just an algorithm, do it this way and nothing else! And let’s be honest, not a single company has the capacity to create AI, just a database with a random answer to a question, remember the demo where they ask how much 2+2? he answers 4, the answer is 6, oh, sorry, you're right, 6...
@@jr-yn4lk A few horses still have jobs...
Thanks WDS, we've been waiting your take on this topic
People always overhype everything. It just so happen that it's in our field this time.
WDS just took Cognition back to college induction, 🤝🤝 love you brother.
Thank you so much for covering this topic. I was really worried about that as a beginner programmer. Love from Pakistan
tera to katega
Don't quite
Maqsad 💣💣
@@SushantShekhar-my7eu lol
When I get to use Devin, I have just one task for it: to build Devin II.
Stopped worrying about Devin once I saw their website LMAO
Devin is the not the point, future Devin is the point, and the exponential rate of AI is improving is what you should be worried about. I don’t understand how people are so short sighted. I’m sure companies are thinking the same thing, they are not looking at Devin 1.0 but 3.0….
@@navis462 I agree. Obviously in its current state, Devin is really not intimidating. If it is capable to take our jobs now, then why are we still working as software engineers? The future and the exponential growth of AI is the point. However having said that, I am still not quite sure how long will it take for it to be able to replace us. So called 'self driving' cars has been around for almost 10 years and today we're nowhere near for a full self driving cars that can be adopted by the masses. AI Software engineers are gonna get better, but the real question is how long will it take
@@navis462 I'm not worried till AI takes over 40% of the real jobs.
Yup its like the will smith video generated a year back. lets just wait a year.
@@navis462 Future devin is today's monkey jpegs.
I was really hoping that the AI hype cycle was largely over with when we started the new year... but man does it seem to be in full swing again. 🤦♂ I really appreciate you digging into this and providing some insight and perspective. Keep up the great work!
It was always going to arrive and grow, to suggest it was going to be a short term thing would be laughable. Its a game changer. It'll change the world as we know it. Its already doing that, though its in its infancy stages. Just look at how the World Wide Web and the Internet changed the world. This is going to do the same thing.
this is just the starting point, just wait when companies release LLMs for every engineering field or for chemists, they gonna get bananas
We're only at the beginning of a fast-evolving AI industrial revolution. From now on, multiple new industries will be hit every year until 70% of population is unemployed.
@@Lykkos-321 glad to see someone awake ! Absolutely spot on... The technology is already there, waiting to be released. The ignorance of people is going to come to an end very very soon. We have let psychopaths at the top get away with too much sh*t without facing any consequences. They are after us and they want it all unless we say NO.
8:32 Just look at the technical complexity of the prompts themselves: "...that initializes two mod_constraints a{1, 5} and b{1, 10} and then computes mod_constraint r = a & b. Then, check that: r.mod is the LCM of a.mod and b.mod. r.v % a.mod == a.v (and likewise for b), and r.v is between 0 and r.mod." I mean, really? Testing libraries are so declarative these days, that if you're going to write a prompt like that, you might as well just write the test yourself!
Yeah that's a good catch.
Not worried about Devin or the next few versions of whatever comes next. I do think in about 5 years tools 100x more advanced than Devin will take our jobs. 2 years ago this was not a topic now we are talking about replacing coders.
The moment I saw news about this Davin thing I went to their website and asked Davin “what model do you use?” And Davin said “Yes, I'm using ChatGPT as my base model. However, I have been customized for specific tasks and goals by the user who created me.” So shrugs. What a bubbly space GenAI is right now.
I must say your video is still the best comparing it to other videos that have talked about Devin. You have said it all. Thank you very much for taking your time ❤❤❤
One more "thanks" from me to add to the pile! =)
Good job, Kyle! More voices deflating the AI bubble are always welcome!
AI is already destroying coding jobs by skyrocketing the productivity of today's experienced developers.
Been watching your videos for many years, was nice to hear your take on this : )
I think one should not just think about the current state of these AI tools, but also what will happen as they become better.
i will convert this video in to mp3 and will listen it every time before i sleep🥲
Dear Kyle Cook ,
I wanted to take a moment to express my sincere appreciation for the insightful commentary you provided on the Devin AI. Your depth of understanding and thoughtful analysis truly resonated with me.
❤❤❤
When every RUclips channel is promoting Devin and AI taking over jobs, this guy just cleared all doubts and made everyone fall flat on their face.
Hope they teach him how to talk to the client and understand what the client wants!
Thank you for dispelling the hype. Way too many people jumping to conclusions.
Good review, thank you for making this video! I will continue studying for a web developer then 🙂
At first glance, I got the same vibe we got another tool to boost our productivity like a copilot and get. Anyway, time to find out what's gonna happen next. Thanks
Hey man, just saying thanks for never selling out to the garbage RUclips algorithm-god, and making actual honest and earnest videos.
We have WordPress and Webflow that competes web programmers for sometime now but it just add jobs rather take our coding job. And as long us people are afraid of code, we safe.
Brother I've been looking for some motivation & then you came, Thank God I won't be learning to “How to Wield”
Remember that this just the beginning, imagine in 5 years what it can do.
We don't know where we are on the curve honestly. We could be at the beginning or reaching the plateau.
Reminds me of my professor at university (Translation Studies) who in 1997 claimed that adequate language translation could never be done by a machine.... ;-)
And his claim is still true, the best AI still can't produce human like translations with the nuances that we have in our native speech, you don't need a professional human translator to spot an AI generated translation. Not to mention literary translations which still needs light years for AI to be good enough to compete with us.
@@mo_1023 Well, in my eyes that is irrelevant, because practically the translations of AI nowadays are good enough for 99% of the cases that you need it.
Also: the nuances that you are talking about - only the extremely good translators are really capable of translating them very well. So what you are basically then saying is: An average translator will never be as good as a genius translator. Well, i would agree on that one. ;-)
PS: Also, the professor was way more radical - he would have never believed in the tranlsation possibilities that exist today. He did not even believe in low quality translations by a machine.
@@mo_1023 Another side note: I have written and published a poetry book in 2009. Now when I use AI today to translate the poems - the quality is better than I myself could produce them. For sure there might be a translator genius that would find even better translations. But there would be way more average translators who would barely reach the quality of AI translations.
This is a good perspective. I'm glad someone pointed this video out to me, because right now, RUclips has it on *Page 29* of the results when I search for "Devin" and "AI" and I would never have found it otherwise. It should really rank a lot higher than a bunch of the "woe is me" doomscroll clickbait garbage that's currently above it.
Humans have a difficulty talking and communicating with humans..... I literally was asked to take off the link surrounding a card, how are people supposed to nav to the page the card references. They respond there is a button. So I remove the surrounding link and ensure the button gets a link instead, then I get now we can't navigate to our pages, turns out they don't necessarily have to fill out the button on the CMS....
lol
I suggest devs do some basic AI machine learning/data scientists courses. The hype will make sense. Especially narrow AI and what Transformers can actually do and what they’ve been made for.
As Frontend developer.. I can't wait to have to refactor entire projects after ai :D
I'm sure you will be using AI to do that.
Thank you for doing the research and spreading the truth about Devin.
You update so fast when the new tool come out...I mean I just saw Devin this morning then you publish a video after 12 hr
Actually he’s been the slowest at this. Devin came out about a week ago
Yeah there are already so many RUclipsrs who Made a Video of it
@@yodkwtf True, but his investigative research is more thorough than any other that I've seen. Everyone else pretty much just said, "Look! Cool!"
I don't think most software engineers are concerned that Devin will take their jobs. I think they're concerned that Devin is one more step towards an inevitable future where the need for human developers will descend. If Devin or other AI assistants are at their worst now, should we not be concerned about where they'll be a couple years from now?
The biggest problem with AI is it is not creative and I realized this when trying to find "hacks" to bypass certain restrictions. It was incapable of "thinking outside the box". An example, Chrome has actively tried to prevent developers from disabling the form auto completion suggestions. There are lots of threads on stackoverflow discussing hacks to trick the browser into not filling out certain fields. We have an address auto suggestion we need customers to use because it verifies addresses and they aren't entering their own address so trying to pre-populate it with previously used addresses is not desired. I could not get it to come up with any unique solutions like people do on stack overflow.
AI is incredibly useful for pieces of code and makes me far more productive. It's incredible for things like, "I have a list of values like ______ and need to filter out all values that contain numbers." It allows me to focus on the big picture and not solving stupid stuff like that where I may have forgotten the syntax.
Cognition Labs team is legendary. A team full of Math and iOi Olympiad gold medalists and Codeforces Grandmasters. These people are the top of the top of their fields in competitive Olympics level mathematics and programming. I trust them. I do believe AI will one day maybe soon replace developers by a large margin and thats not that bad of a thing. It seems most people saying its not a big deal or try not to be scared are just in denial.
What people fail to realize about Devin and similar systems is that most of the power is in the model itself. Devin and other agentic systems are simply getting more out of the model that power them. When we get GPT5 and similar models these systems will become far more powerful simply by swapping out the model powering it
You have helped me with many videos on web development in the last year and you have great credibility with me. Some code youtubers seems to just want to milk the most money they can from their viewers, whichever the most efficient way is, not caring about truth or ethic, spreading missinformation and surfing on hype talking without much thinking about whatever the topic is. "Oh yeah boys, look at this, programming is over, as an ex-faang I say, next year you are unemployed". It's kinda disrespectful with their viewers
Its good to have a good source of information to clarify better so much doubts, at least a more honest source of information.
Thanks WDS/Cayle
If they narrow their scope to "test creation tool" I would be 1000% more interested in this project
exactly, as a dev myself I will pay for such tool, even though these AIs are meant to be general I find them very useful in tests, pretty much nothing more.
we still have to remember how tremendously fast ai is growing - just look at the huge jump from will smith eating spaghetti to sora in 1 year. and even though devin today may look like a cute silly kid trying to do serious stuff, I can't even imagine what it might be able to do 1 year later. but instead of being afraid of ai or being skeptical about it what we actually need to do is to consider ai as a tool and learn how to use it to be a of a higher skill and value ourselves. because even if ai won't be able to replace you, a guy who can use it definitely will :D
software engineer here - I’m responsible for building and innovating software that ended up replacing me. So I can’t sit here and complain. All we do is innovate. Our time is up 😅
Its not the present that scares me but its the future of how advanced ai could really get in the upcoming years which does. Just like how the era of smartphones took over, and there have been constant advancements since, its exactly how I imagine this ai situation to go. And this is coming from a computer science major who will be entering their third year soon.
I hope this video becomes viral
Thank you for this! I was ready to figuratively throw my hands up when i saw the title lol
Stopped pursuing my career in software development 😢😢 just can’t compete with AI no more. I take steps forward but AI is taking leaps at a time.
In my experience, the Problem in Software Development is not the Coding but the requirements.
I've hardly ever met a Requirement Engineer who was good at their job. People just suck at telling clearly what they want.
And this Problem will continue with AI because even HAL 9000 needs to know what it is you want him to do.
Hopefully Devin will understand all of it's clients.🤣
The evaluation on 25% dataset was probably because the other 75% was used for training the AI. We don't know how exactly cognition chose those 25% data but I think they did multiple training cycles through different subsets until they yield the highest result which is like the technique used in machine learning called k-fold cross-validation
I don't have a dev job so Devin is free to take it!
You're the only RUclipsr who doesn't quick cut edit your videos. And you never stutter or say "uh" or "like". Crazy impressive and refreshing!
So we can relax for a few months.. till the next thing comes out
That or understand the fact that the moment software engineering is automated pretty much all the other jobs can and will be automated. If the majority of the people are out of work then companies don't have anyone to do business with, along the chain of B2B companies at the end there is a B2C. Consumers are always at the end of the chain and if the majority of the consumers are broke then it creates a domino effect.
"He who suffers before it is necessary, suffers more than is necessary"
@@_b001 I'm not sure what you're getting at. Are you trying to say that because AI revolution will create a dystopian world, the technological advancement will actually be stopped?
Great analysis of Deivin AI, thanks for sharing this knowledge!
Anyone will go crazy just LISTENING to this fellow, he's talking so fast in that robot voice that I cannot comprehend what he's saying!
The more i know about AI, the less worried I am. You still are the brains of your projects.
‘Things’ will never replace human. They just can assist us
wow your analysis is amazing bro.
10 member team did a remarkable work,imagine now if top companies jump in to improve this
I really don't think that AI will replace whole software team ever if software development became that much easy that we only need to write prompt and it get developed tested and hosted automatically. Then why the hell I need an any third party paid software I can make it on my own by just writing prompt it'll collapse whole market then everyone have their own version of softwares . And also if it became that much easy so it's good for us we can launch new startup everyday we just need to imagine an idea
Current "AI" is just a supercharged search engine. Real AI if it ever gets to that point should be able to update itself, once that happens we can fear for sure.
Thank you for this video! It clarifies many things for me
I can't locate anywhere where they specify the architecture - a big important question that is largely unanswered: "is it dependent on api calls to external ai models/services, or is this it's own fully trained model and proprietary?".
With how limited the info is, for all we know (*and should be very mindful of) this is a fancy UI wrapping a bunch of api calls and a scheduler of some sort. Even the wording they use in their video and THEIR website are very careful not to specify. For example "Introducing Devin, the first AI software engineer". If it were revolutionary, one would think they would be stressing this aspect to that effect; where in their site/vid, they completely dance around it.
Also, with other teams out there with the tools, infrastructure, and highest caliber of engineers in the field...and they haven't yet cracked it, should send up red flags that these guys 'have done it'.
These kinds of tools can never replace or take a job from a skilled developer. But it will definitely take a job from a non skilled developers. All companies will start to expect a highly skilled developers and people like beginners or non skilled will struggle to get a new job. Not everyone will be a skilled developer. Every one can become a skilled developer if given a proper training or experience but The company always needs a skilled one. This is definitely a problem
Having said that, AI will play a very important role in a developer's life. You've done all of it: setting up a project takes all those steps that can and should be automated. Generating unit tests. Etc. and etc. Those tools are invaluable and save time. AI is just a progress in technology (who even writes a quick-sort algorithm nowadays?). So learn to adopt AI.
Chap looks exhausted. Cheer up man.
25% sounds like a hold out test set. They likely trained it on 75% and evaluated on the rest. That’s pretty standard for ml model evaluation and it likely was run on multiple train test splits and reported the best one.
If the difficulty of the problems isn’t evenly distributed between train and test they probably are reporting an inflated number
thankyou brother, this was much needed
Devin is mildly capable today but the question that should be asked is what is the rate of progress going forward? Are we hitting a plateau or will improvements continue at the same rate?
No plateau in the foreseeable future.
@@PakchikpakRajaBabu-sp Any reasoning for this or just a wild guess?
It reminds me how some people actively trying to optimize their MQL strategies over the strategy tester and wonder, why results good on history, but don't predict future.
Really needed video from this channel Thanks Kyle Sir..💖💖💖💖
The transition from human developers to AI agents will be gradual, taking maybe up to a decade or more. One thing is for absolute sure, human developers are on the way out and the AI agents will be the norm in the relatively near future.
Irrespective of the AI, you should be here for the love for tech and be paasionately working in it, things will progress and get better, if you got the game then you will stay here as long as you play it right, forget about you getting replaced just keep coding out of love and then you'll find problems and you creative thoughts will do the rest! Happy coding! 😊
Thank you so much for this video🙏
I am no longer concerned about AI taking my job. At all. I've been at programming for about 18 months now. I had very little exposure to coding prior to that. My company found itself in a position where it needed some services created to replace things being deprecated by 3rd party vendors and I volunteered.
I found some free time last week and wanted to get a head start on an upcoming project so I thought to test these AIs. While I didn't have access to Devin, I tried everything else I could get my hands on. They all failed horribly, not because they are bad per se but because they couldn't really understand the intricacies of what I needed for this particular project. Each of them gave me a good head start. None of them even came close to delivering the finished product. I don't see where Devin is going to be any different.
AI is not scary now, but the speed at which it develops is. In the next 3 to 10 years, it will take our jobs.
or an increase of freelancers
Sure
There’s always going to be humans needed to guide the AI and make sure it’s doing what it’s supposed to do.
Yours for sure. 😂
coming back in 3 years and every year after to laugh at this
My software engineering tasks are so complicated to the point where my brain melts. I have hard times to shape up local solutions and even harder it is for me to explain the problem in this small context window typing it up. AI looks to me hopeless at this point. And it don't know exactly and this is the biggest problem with it. Sometimes it is enough to just show you a few things so you'll figure out the rest, but it could be very unrelated. Sometimes it could make a fine formula and assist in a thought process. But nothing is guaranteed. How can it become a full developer? By marketing! It makes everything seems better!
I remember laughing at Will Smith eating spaghetti less than a year ago.... then Sora. I'm not really worried about Devin, it's Devin 5.0 that scares me and based on the current rate of progress, that's not coming in some mythical, far off future my kids will have to worry about, it's likely coming somewhere before my next performance review. It's done for junior devs and whooooole load of other industries.
Isn't it time for making a fullstack course Kyle ? I know your focus has shifted towards frontend, but going fullstack will open a whole new world to those of us who want a web developer job, please man, market's lacking courses on this subject.
Every RUclipsr dev and their mum is giving this topic more airtime than it deserves.
Thanks for the video, I'm one of your supporters who must push back a bit on your view on this. Sure you've experienced this replacement debate over your career before, but this is different. I'm saying this as a boomer, this is different. Things are changing, in a different way, the average 30 year old with an engineering degree can't easily buy a 4000 sqft home in suburbs like i did. What they're calling immigration today have never been seen before in this country. And AI will change the software development landscape significantly, that's my assessment. I don't have answers or advise for the younger people, I just know this is different.
Yeah, I don't think you have any idea what you're talking about
You're describing late stage capitalism.
The ability to not afford homes is at the feet of land owners and land Lords.
"What they call immigration"
That's a cute racist dogwhistle.
I'm not worried about today or the next 5 years. What worries me is what will happen to my carrer in 10 years, if these big techs keep spending so much money on this. We know what the end goal is, we know what theyre after. And if theyre spending this much money, going this much fast, they will get there.
Proud to be a subscriber.🎉
ChatGPT and Al won't replace you but will make you 10x productive and successful in what you do.
Al-assisted employees will be the future and a requirement in most of upcoming jobs in future. Ride the way to future and start learning it now
It's also going to lower the bar so competition will also heat up
Doesn't 10x productive equal 1/10th the jobs?
Their website and that Google form they're using for early access is enough to debunk their claims 😂
Bro is casually roasting Devin😅 and encouraged our confidence
Only an AGI would be a problem - I am fine using GPT 4 or 5 or any .... I am a developer
When AGI comes we have bigger issues to deal with than just losing jobs…
Thank you bro...
I accidentally found your channel... But that was nice💝
But remember what video gen looked like a year ago vs. today.