Lex Fridman Podcast full episode: ruclips.net/video/qCbfTN-caFI/видео.html Thank you for listening ❤ Check out our sponsors: lexfridman.com/sponsors/cv7889-sa See below for guest bio, links, and to give feedback, submit questions, contact Lex, etc. *GUEST BIO:* Donald Trump is the 45th President of the United States and the Republican candidate in the 2024 US Presidential Election. *CONTACT LEX:* *Feedback* - give feedback to Lex: lexfridman.com/survey *AMA* - submit questions, videos or call-in: lexfridman.com/ama *Hiring* - join our team: lexfridman.com/hiring *Other* - other ways to get in touch: lexfridman.com/contact *EPISODE LINKS:* Trump's Truth: truthsocial.com/@realDonaldTrump Trump's X: x.com/realDonaldTrump Trump's Instagram: instagram.com/realdonaldtrump Trump's Facebook: facebook.com/DonaldTrump Trump's TikTok: www.tiktok.com/@realdonaldtrump Trump's Website: www.donaldjtrump.com *SPONSORS:* To support this podcast, check out our sponsors & get discounts: *Ground News:* Unbiased news source. Go to lexfridman.com/s/ground_news-cv7889-sa *Encord:* AI tooling for annotation & data management. Go to lexfridman.com/s/encord-cv7889-sa *Eight Sleep:* Temp-controlled smart mattress. Go to lexfridman.com/s/eight_sleep-cv7889-sa *NetSuite:* Business management software. Go to lexfridman.com/s/netsuite-cv7889-sa *Shopify:* Sell stuff online. Go to lexfridman.com/s/shopify-cv7889-sa *PODCAST LINKS:* - Podcast Website: lexfridman.com/podcast - Apple Podcasts: apple.co/2lwqZIr - Spotify: spoti.fi/2nEwCF8 - RSS: lexfridman.com/feed/podcast/ - Podcast Playlist: ruclips.net/p/PLrAXtmErZgOdP_8GztsuKi9nrraNbKKp4 - Clips Channel: ruclips.net/user/lexclips *SOCIAL LINKS:* - X: x.com/lexfridman - Instagram: instagram.com/lexfridman - TikTok: tiktok.com/@lexfridman - LinkedIn: linkedin.com/in/lexfridman - Reddit: reddit.com/r/lexfridman - Facebook: facebook.com/lexfridman - Patreon: patreon.com/lexfridman
Surely the incompetent morons at the office who can't even fix a printer will steal my application developer job with the help of AI. yes surely they will just magically know how to properly use cloud services, how to do sys admin, how to host, what frameworks to use, what MVC is, and how to use an IDE. yes surely!
Three things to take in consideration here: 1. AI can't replace a programer yet. 2. It will get better at coding. 3.If AI can code at human level it can do any other job.
These idiots trying to hype this thing maybe to sell more stocks. Anyone who has the slightest idea about ‚AI‘ knows it’s overly hyped and is just a glorified search. People with no education about LLM‘s jump the bang wagen and think they are so special.
@@sepmercury5180 Ppl with no understanding of LLM, will think that LLM is just a probability word guessing machine, even LeCun think that way. It is the combination of multimodalities LLM and alpha algorithm search, q learning, nvidia omniverse and much more latest technology integrated into AI, that makes ppl worry. I think you are not the idiot for not knowing all these things right.
@@神-n6b anyone who is worried has not the slightest idea. At the end of the day, it’s all a scam and people like you trying to hype ‚AI‘ so that more people buy these companies stocks. Don’t name those things like a parrot without even understanding what’s behind them, it makes you more of an idiot than you already are.
Another career facing a similar turning point is translation. I was a translator for many years, and I continue to do translation in my semiretirement. From what I hear, some professional translators have already seen their work flows dry up, because LLM-driven translation is nearly as good as what they do and is much cheaper and faster. On the other hand, translators who not only translate text but also help their clients resolve problems related to intercultural communication seem to be holding on. In my case, I use LLMs as tools to help me do my job faster and better. I first prepare a long prompt explaining the purpose of the translation and how I want it done, I run the source text through several LLMs, and then I use their translations as the basis for my final polished version. The entire process is about 50% faster than if I were translating the text all by myself. I think the quality is somewhat better, too, because the LLMs can come up with good translations for words and phrases that I would not have thought of. But I also agree with Lex’s comment that the ability of LLMs is sometimes frightening. I still haven’t gotten over the shock of seeing that AI could do something - in this case translating between languages - that I thought only humans could do well.
I bet we'll soon see AI translators being used when world leaders who speak different languages meet. There's still a lot of room for these AI systems to get better. Right now, they're mostly focused on English, and they haven't been trained enough on other languages to completely outperform human translators. The amount of training data for languages other than English is tiny - probably less than 10% of what's available for English. But it's just a matter of time before AI becomes really good at every language. I think this will happen fast, maybe even within a year. When that happens, human translators will likely be out of a job completely.
I believe translation is different from programming in the sense that you can potentially accept worse than perfect translation for significantly smaller price running it through LLM. 90% accuracy might be good enough for 1% of the cost. In programming 90% accuracy means non functional system so the bigger the project, the higher the expectations are from the quality of LLM output. While for simple projects, AI might be sufficient in short/mid term, it's really hard for me to imagine that it'd be able to substitute people working on huge projects where the devil in the detail is "the factor". My whole career I worked only on big projects and basically the job of programmers is to fix those devils in the detail issues. So as long as LLM can't be near perfect substitute, I'm not too worried it'd replace us completely.
@@martin.janicek our current approach of using llms to try and write human readable code is a limitation that needs to be overcome. A better system would probably write code in a language designed specifically for AIs to write rather than humans, which could resolve a lot of the problems of generating correct and efficient code. We would need a way to map the corpus of human readable code so that an AI trained on it could output code in this new language, but that seems like a lucrative problem.
@tomgally in order to do the job you described you don't need to be a translator, nor have any knowledge about the source language. It seems that any person good with English writing is able to merge AI translations from Chinese, Russian or Klingon. The next step is that even with a poor English writing skill (like myself, not first language) will be able to prompt an AI to merge multiple translations following an specific style or register. Therefore, your job will be dead in 5 years.
@@hackmedia7755why would they? if the people’s income decreases, this means the companies’ expenses decrease by that same amount. Profits aren’t affected.
@@hackmedia7755 no, jobs will slowly erode and people will be poor for a long time before that happens. the government doesn't care if you lose your job to AI, it only becomes a problem if everyone loses their job to AI. that's why implementing a UBI before then is important. but people are too busy being luddites to push for it
You are the monkey tool for companies. Everyone who has HTML in CV are going to get hired by big tech when AI gets good. They would collapse if they let HTML nerds sit home with AI.
Coder here, completely agree. Embrace AI for development, as most of us don't write assembly code anymore, probably we will not write javascript in a close future.
@@shutinalley I don't have a drone job where I sit at some cubicle in some sterile office hellscape doing repetitive and pointless things I hate for people I hate. My job will not be automated by AI or robots but it might be made obsolete as a secondary effect of AI job loss. Or I should say, the day a nonhuman can do my job, that day we are all obsolete. Being a pet is one of the best-case scenarios.
I am a programmer too and I see it this way: programming is the skill of giving instructions to a machine efficiently. Whether it is via C++, python, natural language or even pseudocode (that could be translated by others or machines). Even when prompting, you have to know how a program is structured, which parts it consists of etc. Practically speaking most coding is 2/3 reading other people's code (or the APIs you are using, Plugins, Libraries, etc.) and maybe 1/3 of typing the own code. Now an LLM can make the writing part a bit more efficient, but if one were to rely on natural language only, how are you gonna do the reading bit? Programming languages can describe Software more precisely and shorter than natural language can. ("int" will always beat "any natural number".) So you gain efficiency on one side, lose it on the other, arguably the more important one. Good software is about planning how to put systems together, not about how to write a simple algorithm or function that runs. If you're coding manually, you get familiar with how your program works exactly, while you're doing it. If you're prompting, you're not (or less at least). So Ai is good to be efficient if you know all the basics, but bad for learning them if you don't. I'd recommend everyone to not use that much Ai, but to do research and use the engines, libraries, plugins and open source correctly, that way you have to type even less compared to prompting.
@@jdmjesus6103 You know what AI cannot do? Reverse the possibilities of a human, who has more knowledge and knows more specifically what he wants to create, vs one, who does not. Especially within software, human knowledge (+AI) will always be superior to just AI, because it is about making services and products for people. No matter how intelligent AI gets, it will never become human, there comes a point where it just cannot know what we are/want as well as we do. As stated before, I am not against using AI. But it can lead to a bad workflow, as soon as the prompting human(s) stops understanding how the program works, soon after will come a point where they don't know what/how to prompt anymore. Imagine a single person prompting to make a software, implementing features for 5 hours, everything works fine. The next prompt they try does not lead to the desired result anymore. Then they realize the generated source code so far is too much/complicated for them to read. To figure out the next correct prompt, they'd have to understand the whole thing or they are stuck. And the difficult thing about software engineering was always evolving medium sized programs into big ones, not making websites or flash games. So feel free to prompt, but also make sure to keep up with the results of those prompts. Understanding is the only way of control when it gets more complicated. If you want to make software tailored for humans that is. If you want to make software for AI by AI, just step out of the loop and take a nap.
@@thFaust respectfully, you're wrong. You're really not looking at the bigger picture. AI will rapidly overtake human ability in every aspect and every possible way. It's happening, and pretending it won't won't change that. To use your example, AI could easily rewrite the entire code to suit the new requirements in no time given enough processing power. But the time will come where it won't even need to, it will have already anticipated what you were about to ask of it. I'm not saying this will be happening next year, but it's coming. Pandora"s box is opened.
@@jdmjesus6103 If that is your belief, I have nothing against it. An interesting question would be, what conclusions do you draw from it? What is humanity supposed to do? Do you have any practical use from approaching the topic of AI with those assumptions? Even if your prediction would be more likely to become true, wouldn't it be more intelligent to assume a perspective that is more pragmatic, that can lead to concrete goals of what to do?
I'm a musician.. Ai is about to take over the music industry. It's song writers who are going to feel it the most. I'm a player most of my work involves weddings and teaching students. I was really hoping Ai would take over that too but for some reason Ai doesn't seem keen on lugging a PA system, lights, instruments etc to a van, then drive that van for 2 hours on a Sunday morning to set up and ready to play by 7pm that night. Then entertain 150 people for 4 hours as they progressively get drunker. Then at 1am in the morning pack all that stuff up and drive the 2 hours home all for the handsome some of 300 dollars each. It would have been 400 but like every time the people say we have 5 other bands that will do it for less money. Please Ai take this away from me😂😂😂
Im also a programming now have been learning guitar for some time, now taking it seriously i believe if ai take over i can still survive doing some random gigs 😅
funny how almost all programmers have some interest in music. maybe because it's both about abstract thinking. one tickles the mind, the other puts the mind at work.
Currently in art, the human is missing. Music is auto-tuned, quantized and full samples. Visualarts involve photoshop and stable diffusion/mid journey. Communication is often via Instant messaging. It is the day and age. And art seems to be worthless. But who wants to see a juke box at a wedding? AI generated art may imitate or create pictures, but how can a machine imitate feelings from one human to another? How can an MP3 give us the same feeling as a human,.making mistakes playing an instrument, in front of you. In flesh and bone?
Recently laid off at Microsoft after 10 years in the data space. Two months in and I’m still struggling to find another job. AI is going to kill jobs and salaries. Initially it will be cost cutting in big tech, to fund data centres but in the medium to long term it will destroy the value of experience. Publicly listed companies only care about satisfying shareholders, if they can cut staff to save costs, they wouldn’t even blink.
What im wondering about is how someone who worked for a big company like Microsoft is unable to find another job? So students who just have graduated from CS/SE have 0% chance????
I'm a computer graphics artist. I spent years learning different skills, including programming, and even made a small game. But for about 10 years, I didn't make any money from it. I kept trying, but nothing worked out. Recently, I finally started making some money as a freelancer. But there's no job security in this line of work. Nobody feels bad for me though, and I don't feel bad about the things I learned that turned out to be useless. I get it if you're a programmer who got lucky and landed a high-paying job. You might feel really scared about AI taking over. But just because you've been lucky for a while doesn't make your worries more important than anyone else's. Maybe you need to accept that life is unpredictable. Try to go with the flow and find ways to get by, just like those of us who haven't been as lucky.
The thing is that only programmers can make programs, with AI or without. Someone who isn't a programmer cannot tell whether code generated by an AI makes sense or not, even if it seems to work at a first glance.
That's for now. For example ChatGPT4 can help me a lot with generating code fast. It can't generate the whole working application. Yet! I think it is not that far of doing that too.
I was really concerned about chatgpt when I first used it to generate code, but the more I use it, the more I recognize its weaknesses. Its a powerful tool, but just like a hammer, you still need a master craftsman to weild it effectively. Someone who doesn't understand how all the pieces fit together will not be able to replace a senior developer for a while, especially if you are on the cutting edge of technology.
Can't you see even one step ahead? This is as bad as AI ever going to get. Even if LLMs run out of steam when you scale up this is not going to be the end of AI development. Some other architecture will take us further forward.
@@ManicMindTrick I don't disagree with this, but the key point is that it will be a different tech, not an llm. And even when that tech does arise, if you stay on top of things, you will move into more of a business role that manages the tech. Once the business role is automated, then there is a bigger issue to worry about than your job, because all of society will be crumbling. You can't sell a product to people with no money. Our only hope is society crashed before then or a tech like neurolink turns us all into integrated cyborgs so we have a chance to compete...
@@justinshankle You can't say LLMs can't take us all the way with any certainty. With scaling and further tweaking or perhaps a new innovation it's possible to go much further beyond today's capabilities. I'm agnostic on the topic as it seems very unclear. Hoping society crumbles sooner rather than later in order to limit the negative impact of the collapse is Ted Kascziyski's idea. For a short while humans together with a chess program could beat the best chess program. Now the human in the loop is obsolete altogether. That is how that cyborg stuff will play out as well if it plays out at all when AI reaches a certain level.
I can't say for sure but I think ChatGPT has a high temperature, a bit too high for coding. Unless they recognize you request code and adjust the temperature, I don't know how they've set it up.
Those executive decisions are always made after considering many factors where there’s no clear information of how those information are weighed. The data which informed those decision made is obscure and hidden. Therefore, I just don’t see AI replacing human at that level even if AI were to articulate better reasoning and critical thinking processes, politicians and executive directors may still scoff at that. Imagine Donald Trump having the most scientific and articulate advisor for his policy making. Will Donald Trump actually exactly follow that advice?
@@dontmindmejustwatching I never understood this stance. If all they need to do is prompt the AI, what prevents other people from making their own companies and doing their own prompting? These businesses would have no moat if all they did was prompt a publicly available AI.
@@0x6e95 nah, that's not what i meant. i meant that superiours at work the ones who "prompt" meaning tell people what to do. So they will just propmpt AI, they don't care who to "prompt'. management will just ask AI what to do and not people, they are immune to AI because they call the shots. if that makes sense.
As someone who's a developer and who's experimented with ai, I've noticed ai is precise and detailed with a 'tree' or 'bark' within the forest but is completely incapable of seeing the 'forest' or understanding a layer within a program and how it works within the broad framework. And if you give it a command to abstract and focus outward, it begins to make mistake after mistake.
That leaves out all the once very productive programmers who lost their jobs in the past year or two that are approaching 50! There is nothing out there even those who keep up to date on everything, as you mentioned, and all the latest ChatGPT newest versions, that BTW still hallucinates now and then. Its hard to see so many in my once heavy tech city having to start over with a new career choice because no matter how well they excel, they are still not wanted due to age, the younger ones disposable as so many, and a now mostly vacant tech area downtown. Near empty towers. I know someone in particular going thru this now as well as going thru his savings, and its very difficult. I am not a programmer, but a mom, and speak up for one as he excelled on the job at every level, but now wonders what now kinda thing. All the best to you Lex for at least making it a topic.
Ai is really useful at writing snippets. There are problems that can't be expressed by using human languages, you have to use computer languages to describe the problem in details. This is why computer languages exist in the first place
Remember that humans pay taxes, programmers contribute a lot to that, AI‘s don’t. Many governments are aware of this, some companies too. Economies would have a BIG problem if knowledge workers were to vanish. Companies need people to buy shit, AI does not buy food, medicine, clothing, transportation. Thats a lot of tax dollars missing if AI did replace so many people.
first I thought the same but as AI develops so will society change. If the rich can have their AI driven world that satisfy their needs they don't need to sell anything to anyone.
Thats not how the economy works lol. Software gets cheaper, people have more money to buy other things. That created jobs and income, which created more demand which creates more tax dollars… and so on. The economy is an ecosystem not a linear machine. And it’s definitely not a zero sum game.
@@JonasMunnich what jobs? there will be no jobs if this gets to that point in the near future because all of them can be done by AI.. what means having people with no jobs? You would have a gap between a certain mass of people with jobs and ones that don't and can't afford anything/living I think you got the answer. And this thing might come soon.
Not just programmers...they will replace all jobs....i recently say a automatic cleaning robot in a mall and realised that everything will be replaced...some sooner some later...
What keeps getting overlooked in this issue is that more and more code will get created in the future. So LLMs will be a productivity tool for programmers (not replacement) and although you might fear this still means fewer jobs (one programmer can create more code in shorter time) in fact you'll see just more code getting created. The worlds hunger for code is only limited by the amount that can be created. It's wholly determined by supply, not demand.
What makes you think more code will be created/needed? There is already a competition between sotwares in almost every aspect of the field that not differ much from each other, so there is already an overproduction of sw.
@@fragebogenvbcAs the world continues to embrace technology; the demand for software will and currently does far outpace the number of people who can develop said software. If AI ever does take off and become more then what it currently is; it will enable companies to cater to those needs and the amount of overall production of software will level out with the demand for SE.
@@apricotmadness4850 currently absolue not outpace. there are 100 thousands of unemployed developers in the US currently. And if AI takes off, the increase of the number of human written sw, is only a therory without any solid base.
@@apricotmadness4850 Why would u as a company want to have people who code worse then Ai it doesn't make sense it's much better to have an elite just manage a bunch of Ai since Ai don't need to sleep, time off or even eat it's more productive. I think it's pretty silly pay a programmer when Ai will work faster then that human Alpha 2 already performs better then 85% of coders in 2-4 years coders will be obsolete we saw this in music and art and it will happen to coding too. Don't think companies will not want to cost cut since they already are cost cutting as much as possible like Amazon are replacing 5 workers with 1 and just forcing them to work 7 days a week. It's actually crazy programmers don't understand such a simple concept everyone is replaceable, Amazon is using Ai to replace managers so 1 manager do 5 jobs at once efficiently and just as effective as before if have Ai that can code why would I even bother having a team of humans when 1 human can do the work of a whole team with Ai. If replace 85-95 percent of coders with Ai how much would I save as a company? A good example would be google avg medium salary is 147K a year around 60K employees so multiply 147K times 54,000 you get 7,938,000,000 so why wouldn't I replace most programmers with Ai if I can save almost 8 billion dollars?
@@developit1152 💯 ... expect unemployment figures to tick up monthly as businesses look to be more efficient. Larger companies will restructure. Consolidate departments. The savings in salaries will go to Ai companies. Companies will choose the brightest and most efficient to lead AI agents.
I am a lead software engineer with 13 years of experience. In short, I can say: don't worry, it can't replace developers. What it can do is reduce the number of software developers by up to 30%. And this is still disastrous. Let me tell you why it cannot replace developers. First of all, the current approach of LLMs is learning solved problems and solving similar problems to these. What they can't do is combine many of these solutions to solve your custom problem. If there are just a few requirements, LLMs can write nearly perfect programs by combining solutions to small problems. But if you have many requirements, which can be solved with at least 3k lines of code, then LLMs can't solve these problems. The reason is mathematical. Every time you add a new requirement, the complexity of your software increases exponentially. Without quantum computers, it is not possible to solve problems with algorithms that have exponential complexity. Assuming we will not get quantum computers soon and not change the current LLMs' problem-solving approach, we can say that AI cannot replace programmers entirely.
I tried it and it's good at doing small-ish tasks that have been done many times before. I noticed in the output that most of the code was more or less copy-pasted from existing github repos and I could easily track down the code it "borrowed". I tried adding some requirements, and it got completely lost. I also tried to use it to build a compression algorithm that I had in my head for many years but never had the time to build. It was basically a combination of existing algorithms. The LLM wrote a decent implementation of the known parts, but the result didn't work. One part of my algorithm was using an adaptive Huffmann coder with a growing number of symbols, and the model simply gave it a fixed size memory for those. When I asked it to fix it it did, but then it didn't decode properly. When I told it that, it's idea of solving it was making the output of the huffmann coder fixed size - completely negating its value. That being said, it saved me tons of time in writing the well known algorithms so that I didn't have to and at the same didn't have to feel bad about copy-pasting them from github myself. With its help, I was able to get my algorithm to work rather quickly by modifying its output.
Interesting point of view. Assuming that this sentence is correct: "Every time you add a new requirement, the complexity of your software increases exponentially"
@@bestopinion9257 Actually I was talking about the system complexity, rather than runtime complexity. I mean complexity of building the code increases exponentially for llms, not running the code.
I don't know as somebody who spent several years in college learning computer science I'm horrified that there were certain skills that took me 6 months to learn that AI can do in under a second. I'm sorry the truth is that it's simply making the workforce more compact yes the best coders will always be able to find a job at least in the next 10 years but the labor and groundwork that used to leave open entry level jobs and internships has been taken up by artificial intelligence.
dont worry bro, you will be allright. This is nothing new whats happening, just the next step on evolution, some people do are left behind but that is for the whole society to be better off. Always have a plan B on your mind as well though, in case you are one of those people that are left behind.
Top employees benefit the least from AI. These are also the people who didn’t use Stack Overflow. They are solving problems that have never been solved before. This is true across the board with genAI, lawyers, marketing, … AI is closing the gap between the just below average and just above average. The bottom of the pack cannot identify problems with AI content, so they receive little benefit. As the cost to produce software falls, projects that were priced out will get green lights. There is no way to know if the total developer headcount will be more or less than today down the road.
Will work only for some time. How can a voice actor use AI to do his job better? If it can be done on a similiar level for free, your job will be gone even if you got "better"..
@@Felipe-zl1rj interesting hypotheses .. i am not a programmer .. i help solopreneurs and business owners that are serious about AI to integrate AI in their processes, products & Services and into their customer journey ... many of them are fighting against AI and i help them to use AI to become better and do a better job ... I love that. An by the way .. its not the cost that are killing companies ... its the complexity and therefore the speed that they are able to move forward ... this will be a driver for lay offs .. not the costs that need to be cut.
I would say for a lot of people beginner - intermediate, familiarize yourself with Cursor, but don't use it for everything. For someone like Lex who has the experience and understanding of the tools/languages it will make him faster, but for many it will turn them into prompt engineers rely on prompts to debug.
Im a jr dev. A little over a year of pro exp, but 3 years of programming. I used code generating tools for a long while. Like many, i would run into a bug, and no amount of sweet talking the AI would fix it. And i certainly wasnt skilled enough to fix its hyper advanced solution... I eventually realized that it takes many years to master programming, and far less skill to use prompts. However, in order to know what prompts to ask, you first need the skill to see a bad implementation. So for the next decade i plan to become as good as i humanly can at programing. When ai does take a larger role in code, in some 10-30 years, at least all ill need to learn is prompt engineering.
Perfect attitude, and ironically probably going to be the one that actually goes somewhere compared to the "Well the dev who uses AI will replace the one who doesn't."
The best advice I have ever been given is to learn concepts not syntax if you know concepts now you can have artificial intelligence write code for you and you can rearrange it or tell it what to fix because you understand the concepts but syntax is constantly changing and if you've spent constant hours like most of us learning syntax while that's essentially just wasted time.
@@78alJ0vle The engineers who benefit the most from AI are those just below average. It takes them a little longer to get their work done by hand and they at least have enough experience to realize when AI is hallucinating.
@@vaolin1703 I guess I'm referring to cramming the code like a lot of people do where they're just trying to basically be able to just do any of the problems because they went through it through the website, no I agree like if you understand it from conceptual point of view then you're great
I'm a videographer, I would never have the time to learn programming nor would it be worth my money to develop programs that assist me in my job. HOWEVER, with claude and chatgpt I have built several programs that are specifically tailored to improve the efficiency of my workflow. It is fantastic that basic programming is available to the everyman now.
@evaander For one I use it to summarize and organize interview transcripts. But in terms of programming, ive built a program that essentially organizes my project database according to the exact organization format I use, one that helps manage and organize the file backup process and several more that are more niche in my field that I'd have trouble explaining quickly in a comment section. They all use python with pyqt6 for UI. Only reason I'm bothering to respond is that other dude called me a bot
Coding is a tool for problem solving. Be agnostic to the tech that helps you get to an optimized solution. Don't try and be a good user of some tool. At this point i've been coding for so many years that I don't see code. I see logical connections. Whether I explain these abstract connections with code or natural language won't matter. I'll do whichever is most efficient at the time.
Better advice, don't worry about the coding part so much. Improve your skills by improving your knowledge of the world and industries. Jump into bioinformatics and begin hammering away at some DNA sequences through various algorithms. Take me, I developed a awesome NLU engine for Cicero project that requried me to learn all about NLP in-depth, now I'm just finalizing NeuralLink's compression challenge which took me through the world of audio transformations and varying compression techniques, and so on. I don't know, jump into material sciences, genetic engineer, signal intelligence, or whatever. Expand your development skills that wa instead of just trying to learn the next framework, or whatever.
1. For many decades non-programmers have relied on others to create software for them. Those days are coming to an end and there are astronomically more non-programmers. 2. The market will demand software at a level of sophistication and complexity that vastly exceeds human ability (programming, code-review, maintainance, refactoring, scaling, etc.). Youngsters should investigate opportunties in QA, UX and CS - all relevant disciplines that will be needed long-term. You're welcome.
If programming can be fully done by AI, so can essentially any other job done on a computer. What would prevent it from interacting with CAD programs, collecting data and interacting with Excel, etc.? Who knows what will happen.
i believe we need to rethink the fundamental approach to designing these systems, one last time. as scary as it might sound, if you simply train these models to be self interested you would solve a lot of problems of not generalizing knowledge well.
Yes but probably not as quick A lot more data to feed these LLMs in regards to code than most other things. That’s why they’re already decent vs others things
About CAD: Drafting plans may be gone soon enough, because that’s mostly technical/mechanical work, but a human needs to oversee the design, the overall process, to coordinate and check AI’s output. At some point in time, I assume, it will be possible to create a detailed task for AI (basically, the same as you have in contact between the Current and the Architect, describing what exactly needs to be designed) and AI will create options. If AI will be able to learn all the Federal and Local Codes, plus best practices, you would need not 100 but 10 professionals in the company. And to be among those 10 lucky guys, you gotta step up your game. All mediocrity will be substituted by AI.
What you describe would be easy. The difficult part for a human would be all bullshit syntax you have to know and type correctly each programming language involved in the process to work
I am an Illustrator and 2D animator, recently I decided to go back to school and learn Software development/programming, since I was young, I always wanted to learn programming as an extra skill besides what I already have. I’ll be honest: I’m still a beginner in coding. One day, while racing to finish several assignments, I decided to try using AI. To my surprise, it worked! I took the instructions for my assignments and asked the AI for help, which provided me with the JavaScript and HTML codes I needed, I was too lazy to type them, think deeper etc... I submitted it and passed, but afterwards, I felt guilty. What’s the point of studying this if a computer can do everything for me? I realized that I relied on AI for about 95% of my assignments. I’m uncertain about how AI will be used in more advanced programming contexts, but as a beginner, it honestly freaked me out. I thought about how someone could graduate without truly having proper knowledge of programming. AI is definitely going to change a lot in the lives of programmers, bringing both positives and negatives. Many jobs may be at risk. In 15 or 20 years, senior programmers might retire, creating space for newcomers, but it will take highly skilled programmers to compete with AI. The tech industry, and really, every field, including art, music, and writing, will likely be dominated by AI, which is a sad reality.
Really liked this take, Lex. And I suspect very relevant to people working in the arts as well. As part of my work I have regular consultations with creatives and naturally these feelings towards AI often arise, and finding an approach which feels both compassionate and pragmatic is always one worth seeking.
The question on AI is not so much a deep fear of slipping into irrelevance because that AI will soon exceed people's programming skills but rather that AI will soon reach escape velocity. By escape velocity, I mean that the speed at which AI can learn is faster than the speed at which humans can learn. This creates a scenario where whichever direction that people decide to up-skill towards will be made obsolete by the time the people are finally able to pick up the skill creating an atmosphere of obsolescence, futility, and frustration.
1) learn to use LLM to solve problems 2) get an interview where they ask you about low level shit that no one does at job nowadays even before AI 3) get rejected 4)*confused* 5) *repeat*
I mean, using A.I. still requires understanding of how programming works, as Lex states to edit the code that the A.I. generates, however, programmers feel more like directors now who overview, oversee, and fit all the code generated together. Is it more holistic and less in depth than what programming used to be some years ago, sure, but it still takes skills that much of the population still can't do. Even if the I.Q. level is reduced now to create these apps because A.I. can generate features. I personally love A.I. it makes coding so much easier but of course if it is too easy than you can become redundant! But programmers also had this fear when code editors and debuggers were introduced and if anything just increased the demand!
yes, AI needs guidance. It can make and change as you ask small parts of your application. You are the one that understands how all these small parts are fit together. But that's for now. With improvement one day you will ask AI "make me an eCommerce site" and it will do it in seconds. Then without any programming skill, ask AI to change this color, put more menu buttons here, make these categories of products and so on. You do not need to code or read code at that point.
@@dontmindmejustwatchingHow do you know that? You realize your prediction is radical? I net you have more than a gut feeling to back up a radical prediction. It just sounds like Shapiro talking about AGI 7 months ago.
Investors are asking a different question - will LLM development firms be around in the next one to two years - aleph alpha recently pivoted away from building large language models because the costs are not sustainable, particularly with diminishing improvements and increasing competition as new models from various vendors come out every week. Will the GPU gravy train continue? and for how much longer?
one could certainly run their own open-source LLM, but even that is expensive, whether on the cloud or on-premise. Is the cost really worth the return over time?
on an investment mindset AI is very expensive and returns peanuts. But if you think away from businesses, you'll see we're not competing with businesses, we're competing with China and Russia.
@@jurassicthunder Except we're outspending them by huge margins... honestly I think China and Russia only care about it because the best practical application for current GenAI is propaganda and misinformation... not the ushering in of a new Sci-Fi world that AI enthusiast like to pretend is happening.
This was a really insightful piece on the evolving role of AI in programming. I appreciate the perspective on embracing AI as a tool to enhance our skills rather than fearing it as a replacement. Transitioning from a low-level tinkerer to a big-picture designer resonates deeply. It's a reminder that adaptability and continuous learning are key in any field, especially as technology advances. Thanks for sharing your thoughts and experiences, it's encouraging to see how we can leverage AI to become more efficient and creative in our work!
LLM can’t read your mind. You have to spend significant amounts of time crafting and refining prompts to match your specs. Time that could have been spent designing and writing the code yourself. AI does not boost productivity for real life problems of medium or higher difficulty. At least not until we find a way to feed it all the context it needs to understand a problem in its entirety.
Dumb comment, writing prompts is much faster. Also, AI will improve. You are judging from the earliest versions. It's like someone looking at a Computer in the 1970s and saying it's useless
@@rameeziqbal8711AI actually started in 70s and those were the ealiest versions. What we have today comes after 50 years of developement. With like 10 years of it being used widly in some sectors, like image recognition.
Yeah I haven’t met any working developers that feel this way. Personally, when I was relying on co-pilot full-time for a couple of months, I felt like my code quality suffered heavily. Another anecdote, I’ve seen some of the code that juniors on my team have produced with the help of LLMs and it will dramatically mislead them into a bad direction/design.
What people don't mention enough is how it wastes more time than it saves when you actually need something complex to be deployable. Use it only for specific parts or components of your app or site. It is not going to replace programmers in the next 5 years period. After that it will replace us humans completely so no need to worry about losing your job over loosing your life.
LLMs can't innovate. It can only spit out code that looks like something it has seen in the training set. If you do anything a bit off the beaten path it's not going to be much help. It can save you some typing. On the other hand if your job consists of copy pasting snippets from Stack Overflow, then yeah your days are numbered.
Im a lead engineer, Ive built some things recently i WISH AI could of helped with. Try having AI build anything with drag n drop, or something with lots of events that coordinate together and see if it doesnt create race conditions. like a wysiwyg editor. Ultimately these things i had to build myself. I also built an agent framework that cursor failed to understand how to add support for claude, i also needed to do myself. There is still so much it fails so hard on, when its not some simple task.
Yes, many of us wish AI would work for us as it does for others, creating simple CRUD apps for common entities in a generic framework. In real life, some of us our really trying to raise the bar for our users, but that also means we can't copy and paste some standard app.
Have you seen the AI Doom paper? Zero programming required. The AI just 'imagines' it and it appears. There are no complex algorithms, no state machines, no design patterns, no optimizations - AI makes them all obsolete. Most people seem to be in denial about what the future holds.
@@rickenbacker472 > Human raters are only slightly better than random chance at distinguishing short clips of the game from clips of the simulation What about long clips? Like a whole level.
@@rickenbacker472 yes, I work in AI, the AI doom where nothing was the same enemies fade in and out of existence, the score was inaccurate. And no persistence or consistency. The AI doom paper is solving a different problem. It doesn't generate anything consistent, but generating scenes, worlds, custom on demand. It's built off diffusion model, and is indeterministic.
I think just like knowing lower level languages makes you better with higher level languages knowing programming at all will make you better at utilizing the AI
I love it. A better world awaits. Thanks for the Cursor recommendation, will check it out. I use Claude Sonnet as well, but Gemini Pro is also knowledgeable about some relatively obscure languages.
Concise and clear video - excellent. One point not mentioned : the AI-adept developers will be very productive which is most likely going to lead to layoffs of the less capable. I would expect at least 30% of software developer roles to fade away over the next year or two.
Curious to how others workflows are changing, what tools have you tried? I've been experimenting with Claude-Dev, Cursor, Aider-Chat. Not settled on any one particular tool ATM and just trying out various tools and workflows.
No its just a tool there's not enough energy in the grid to power the tpus and servers that run these LLMs, it would cause major blackouts to replace thousands of employees.
I'd love to see a survey across programmers of different experience levels asking how easily they think AI can replace programming. Or even more debate/discussion around what it means or what can be replaced, but I mainly see sensationalism of "All coding is dead" or at the other end what feels like denial "It will never replace human programmers" haven't seen much good in-depth discussion though yet. I'd love to hear from more career programmers not necessarily associated with AI and hear their thoughts, as to some degree anyone working for an AI company will obviously be very bullish on it replacing programmers.
Code is for humans. If AI is to replace human developers, it shouldn't have to write code. We would just tell it to build a system and it would produce the binaries that solve the problem. As long as it merely generates code, it will always need a programmer to know what to do with it.
I rate AI just above noob level. Syntax is good, but they struggle with systems and comprehending complex apps, especially ones that are dynamic and creative. Good for assisting but very prone to suggesting faulty code.
I have to fix all the code LLM writes, and it has such a subtle bug that I have to really think about what is wrong. Its playing with my perception. And it takes more time to fix the AI-generated code than to write it by myself. It is beneficial to generate boilerplate code. And this way it's really useful. It's not scary.
Keep moving up the value chain. The more output a programmer can produce (eg via AI), the more value created per hour, the cheaper per finite output, the more demand for that output. Elasticity of demand suggests that if the cost per output falls, the demand for such output increases by more than the proportionate decrease in price, leading to an overall increase in total revenue and market size. So as productivity rises and costs fall (e.g. due to AI), new opportunities emerge for consumption, with businesses/individuals purchasing more of the output, further driving economic growth and demand for skilled programmers capable of leveraging these technologies. In short, there will almost certainly be future demand for competent humans who can nagivate this space and help others.
Cursor is amazing. I haven’t had the title engineer in 15 years and I’ve been able build useable software using python. I really don’t know python but I know how to focus engineers to build what I need as a Product manager. I also have general troubleshooting skills and cursor is an excellent pair. Now I need to learn TDD so I can confidently let the AI refactor and make sure the tests pass.
Good answer. I reached the same conclusion. AI has become a part of my workflow as a developer. But sometimes you do find yourself wrestling with the code it generates 😅
From a person who does not code and do programming for a living, coding is just a part of the architecture, and like typing, spelling checking, and translating, it's better to have some help. It's time-saving and good for health.
Hey lex just wanted to ask you that if someone wants to become ML engineer. Is it important that you must have a strong knowledge of DSA. Or if you're not that good at DSA , can you still learn ML
Yep I see coding as the most useful use cases of AI because coding is a completely closed box problem. you have functional requirements and the code can be tested against those function requirements. AI can debug it’s code by running the code repeatedly..
@@rameeziqbal8711 Lawyers will definitely get hit hard. As will many on your list. Profound changes are coming. If half the workforce is not needed, accommodation needs to be made. We can't have rampant homelessness. Will be horrific otherwise. It's already too much.
@@tommynickels4570 How will lawyers get hit hard? They indulge in a lot of human interaction. How you deliver speeches, communicate, ask questions, plays a big role in the outcome of a case.
How is “writing natural language prompts” different than writing a technical specification? We are going back in time, and we won’t need as many developers, especially for code that is more utilitarian like most enterprise software.
It could go either way. If development is cheap enough, all the SMBs that couldn’t afford custom software will suddenly be hiring. Take a look at what happened with bank staff and ATMs
Hi Lex. I am a writer and I am using my creativity to integrate ai in my writing process. I'm not using ai to replace my ability to write But find ways to make me a better writer. It definitely allows me to be more efficient in my writing process. Thank you for your comments about how to deal with AI as a programmer which applies to almost any field that might be affected by the new wave of AI.
I think it’s just a good google search. I don’t think it will replace us anytime soon because it still doesn’t understand complex business rules for your company and in my experience it just spits out garbage code. It’s really good for school projects though or any other basic stuff
I think AI will replace a lot of the frontend UI work, or at least change it substantially. Why pay an engineer to implement your figmas if an AI can do most of it and then the engineer hooks it up to the rest of the app.
Great take - AI will not replace programmers. But we do need to embrace it so do our job better. That being said, the current state of AI does not help us to do our job better (maybe it does, if you're junior developer, but let me tell you, it does not help you in professional fields). And nor do i think it will be substantially better in the near future because we are plateauing in terms of AI progression. But as Lex said, it could change in the far future and we just have to be willing and ready to "adapt".
I user Claude to do the scafolding. Nowadays, I do 80% to 85% less manual coding than a couple of years ago. Yesterday, for instance, Claude did to me in half an hour* what would have taken me 4 to 5 days. (* it took me half an hour to interactivelly request small changes to my code, until I got it exactly the way I wanted.) So, right now, if you don't use Claude (or others), you are wasting time. Eventually, in a couple of years, all of my code will be done in natural language, in interactive sessions. This is going to be the next step in the "High Level Languages", I think. English is the new Python/JS/Rust/C/and-even-assembly.
First want to say I agree and you're right. But here is another fun thought experiment. Aren't we wasting our time right now? We can build something in say a week. And we continue to do this week after week till the end of the year. Then, at the start of next year, the first week of 2025, that's a single week... we can get as much work done as we are doing today in 8 weeks. Then why don't we just chill for 8 weeks right NOW, and just have it generated within 1 week at the start of 2025 ? Maybe we should all just go on vacation for a few months and then generate everything that we would've written this year. And then, we have got just as much done as if we would've spent the time working instead of going on a few months vacation.
"Eventually, in a couple of years, all of my code will be done in natural language, in interactive sessions" . Devil's advocate here - why would anyone pay you a high salary to do that? Any junior or even every product designer would be able to do that. If you have years of deep programming experience they'll become nullified and probably so will your employment prospects. I'm still hoping that like in autonomous driving, the last 1% of accuracy would be impossible for LLMs , thus necessitating humans to still understand and write code from time to time.
@@יואלבלום don’t fall for this hyping. They are all trying to sell AI company stocks and a bunch of ignorant parrots who only repeat a bunch of nonsense they read somewhere (like singularity AI overlords etc) else trying to feel special. LLM‘s are just a glorified search and without developers who know their stuff it’s practically useless for anything more merely complicated. Education is what we need. People need education.
@@יואלבלום Which is precisely why people should change their career aspirations away from being a master programmer. Unless you are aiming at becoming an AI programmer. The writing is on the wall for those paying attention. While I don't like using this analogy since it's apparently a myth, people asserting that LLMs are not a threat are the frog in the pot of water increasing in temperature.
The thing is, AI is a tool-learn to use it. I’ve already used one to help me put a game idea together. It’s not coding, but I now have a 360-page blueprint of an MMO ready to go. Soon, I’ll just need to feed that into an AI to start programming it for me.
I have spent several months learning how to avoid coding. I have managed 98% AI generation using Aider. Full stack.. truthfully front end is still weak though. It can build but it still struggled with UX but the backend stuff, it is almost there.
Has it really through? It really isn’t that impressive. It’s not even chat gpt 5. It’s just the same technology they’ve been using with different methods,
I honestly think that the careers benefiting the most from AI are design and creative ones, I'm a 3d designer and I've been developing insanely useful tools for the 3d artists of the company I work for, because I'm the end user of it, so I know our "pain" as a 3d artist better than any programmer. My brother on the other hand is a programmer and he is expressing the same fear you are talking about in your video.
They always need 'low-level tinkerers' to build something great. But once something is built, they forget that fact and only admire 'big-picture designers.' Crazy times for artisans.
I guess it all ties into the 'control problem'. I think it's inevitable that AGI will replace programmers. It's still a hypothetical, but the 'race to the top' will bring it. The question is: will AI replace carers? If so, that's me out of a job, too. Somehow, I think that's less feasible!
As a non-programmer AI can write me programs. I think getting into programming would be a disadvantage. Not care about the code, but only for if its working, doesnt occupy your mind with irrelevant cognitive resources.
If AI takes my job I will be like, okay, congratulations, I can finally go and have my own farm and have animals there and I will know, AI is changing the world on my behalf
Lex Fridman Podcast full episode: ruclips.net/video/qCbfTN-caFI/видео.html
Thank you for listening ❤ Check out our sponsors: lexfridman.com/sponsors/cv7889-sa
See below for guest bio, links, and to give feedback, submit questions, contact Lex, etc.
*GUEST BIO:*
Donald Trump is the 45th President of the United States and the Republican candidate in the 2024 US Presidential Election.
*CONTACT LEX:*
*Feedback* - give feedback to Lex: lexfridman.com/survey
*AMA* - submit questions, videos or call-in: lexfridman.com/ama
*Hiring* - join our team: lexfridman.com/hiring
*Other* - other ways to get in touch: lexfridman.com/contact
*EPISODE LINKS:*
Trump's Truth: truthsocial.com/@realDonaldTrump
Trump's X: x.com/realDonaldTrump
Trump's Instagram: instagram.com/realdonaldtrump
Trump's Facebook: facebook.com/DonaldTrump
Trump's TikTok: www.tiktok.com/@realdonaldtrump
Trump's Website: www.donaldjtrump.com
*SPONSORS:*
To support this podcast, check out our sponsors & get discounts:
*Ground News:* Unbiased news source.
Go to lexfridman.com/s/ground_news-cv7889-sa
*Encord:* AI tooling for annotation & data management.
Go to lexfridman.com/s/encord-cv7889-sa
*Eight Sleep:* Temp-controlled smart mattress.
Go to lexfridman.com/s/eight_sleep-cv7889-sa
*NetSuite:* Business management software.
Go to lexfridman.com/s/netsuite-cv7889-sa
*Shopify:* Sell stuff online.
Go to lexfridman.com/s/shopify-cv7889-sa
*PODCAST LINKS:*
- Podcast Website: lexfridman.com/podcast
- Apple Podcasts: apple.co/2lwqZIr
- Spotify: spoti.fi/2nEwCF8
- RSS: lexfridman.com/feed/podcast/
- Podcast Playlist: ruclips.net/p/PLrAXtmErZgOdP_8GztsuKi9nrraNbKKp4
- Clips Channel: ruclips.net/user/lexclips
*SOCIAL LINKS:*
- X: x.com/lexfridman
- Instagram: instagram.com/lexfridman
- TikTok: tiktok.com/@lexfridman
- LinkedIn: linkedin.com/in/lexfridman
- Reddit: reddit.com/r/lexfridman
- Facebook: facebook.com/lexfridman
- Patreon: patreon.com/lexfridman
KAMALA2024!!!❤❤❤❤❤❤❤
The description for this video is wrong…
Surely the incompetent morons at the office who can't even fix a printer will steal my application developer job with the help of AI. yes surely they will just magically know how to properly use cloud services, how to do sys admin, how to host, what frameworks to use, what MVC is, and how to use an IDE. yes surely!
Three things to take in consideration here:
1. AI can't replace a programer yet.
2. It will get better at coding.
3.If AI can code at human level it can do any other job.
and you no need total replacement in order to get mass unemployment in this sector.
Can it cool your meal (not a pre-prepared one, from scratch) ? :D Can it cut your hair, shave your balls ?
These idiots trying to hype this thing maybe to sell more stocks. Anyone who has the slightest idea about ‚AI‘ knows it’s overly hyped and is just a glorified search. People with no education about LLM‘s jump the bang wagen and think they are so special.
@@sepmercury5180 Ppl with no understanding of LLM, will think that LLM is just a probability word guessing machine, even LeCun think that way. It is the combination of multimodalities LLM and alpha algorithm search, q learning, nvidia omniverse and much more latest technology integrated into AI, that makes ppl worry. I think you are not the idiot for not knowing all these things right.
@@神-n6b anyone who is worried has not the slightest idea. At the end of the day, it’s all a scam and people like you trying to hype ‚AI‘ so that more people buy these companies stocks. Don’t name those things like a parrot without even understanding what’s behind them, it makes you more of an idiot than you already are.
Another career facing a similar turning point is translation. I was a translator for many years, and I continue to do translation in my semiretirement. From what I hear, some professional translators have already seen their work flows dry up, because LLM-driven translation is nearly as good as what they do and is much cheaper and faster. On the other hand, translators who not only translate text but also help their clients resolve problems related to intercultural communication seem to be holding on. In my case, I use LLMs as tools to help me do my job faster and better. I first prepare a long prompt explaining the purpose of the translation and how I want it done, I run the source text through several LLMs, and then I use their translations as the basis for my final polished version. The entire process is about 50% faster than if I were translating the text all by myself. I think the quality is somewhat better, too, because the LLMs can come up with good translations for words and phrases that I would not have thought of. But I also agree with Lex’s comment that the ability of LLMs is sometimes frightening. I still haven’t gotten over the shock of seeing that AI could do something - in this case translating between languages - that I thought only humans could do well.
I bet we'll soon see AI translators being used when world leaders who speak different languages meet. There's still a lot of room for these AI systems to get better. Right now, they're mostly focused on English, and they haven't been trained enough on other languages to completely outperform human translators.
The amount of training data for languages other than English is tiny - probably less than 10% of what's available for English. But it's just a matter of time before AI becomes really good at every language. I think this will happen fast, maybe even within a year. When that happens, human translators will likely be out of a job completely.
I believe translation is different from programming in the sense that you can potentially accept worse than perfect translation for significantly smaller price running it through LLM. 90% accuracy might be good enough for 1% of the cost. In programming 90% accuracy means non functional system so the bigger the project, the higher the expectations are from the quality of LLM output. While for simple projects, AI might be sufficient in short/mid term, it's really hard for me to imagine that it'd be able to substitute people working on huge projects where the devil in the detail is "the factor". My whole career I worked only on big projects and basically the job of programmers is to fix those devils in the detail issues. So as long as LLM can't be near perfect substitute, I'm not too worried it'd replace us completely.
@@martin.janicek our current approach of using llms to try and write human readable code is a limitation that needs to be overcome. A better system would probably write code in a language designed specifically for AIs to write rather than humans, which could resolve a lot of the problems of generating correct and efficient code. We would need a way to map the corpus of human readable code so that an AI trained on it could output code in this new language, but that seems like a lucrative problem.
@tomgally in order to do the job you described you don't need to be a translator, nor have any knowledge about the source language. It seems that any person good with English writing is able to merge AI translations from Chinese, Russian or Klingon. The next step is that even with a poor English writing skill (like myself, not first language) will be able to prompt an AI to merge multiple translations following an specific style or register. Therefore, your job will be dead in 5 years.
the big fear is if there are 100 jobs in tech today those will be replace by just 10 people working with AI , so 90 will be left unemployed : )
and the companies will go bust because there is less people with income buying stuff.
@@hackmedia7755why would they? if the people’s income decreases, this means the companies’ expenses decrease by that same amount. Profits aren’t affected.
@@hackmedia7755 there are other people too, not everyone is a programmer
@@hackmedia7755 no, jobs will slowly erode and people will be poor for a long time before that happens. the government doesn't care if you lose your job to AI, it only becomes a problem if everyone loses their job to AI. that's why implementing a UBI before then is important. but people are too busy being luddites to push for it
You are the monkey tool for companies. Everyone who has HTML in CV are going to get hired by big tech when AI gets good. They would collapse if they let HTML nerds sit home with AI.
Coder here, completely agree. Embrace AI for development, as most of us don't write assembly code anymore, probably we will not write javascript in a close future.
Pretty soon people will have to find a purpose beyond typing in front of a screen.
@@shutinalley Maybe we can be the pet of our new AI overlords :D
@@ManicMindTrick You're already the pet of wherever you commute to every monday through friday.
@@shutinalley I don't have a drone job where I sit at some cubicle in some sterile office hellscape doing repetitive and pointless things I hate for people I hate.
My job will not be automated by AI or robots but it might be made obsolete as a secondary effect of AI job loss.
Or I should say, the day a nonhuman can do my job, that day we are all obsolete.
Being a pet is one of the best-case scenarios.
@@ManicMindTrick what do you do for a job?
I am a programmer too and I see it this way: programming is the skill of giving instructions to a machine efficiently. Whether it is via C++, python, natural language or even pseudocode (that could be translated by others or machines). Even when prompting, you have to know how a program is structured, which parts it consists of etc. Practically speaking most coding is 2/3 reading other people's code (or the APIs you are using, Plugins, Libraries, etc.) and maybe 1/3 of typing the own code. Now an LLM can make the writing part a bit more efficient, but if one were to rely on natural language only, how are you gonna do the reading bit? Programming languages can describe Software more precisely and shorter than natural language can. ("int" will always beat "any natural number".) So you gain efficiency on one side, lose it on the other, arguably the more important one. Good software is about planning how to put systems together, not about how to write a simple algorithm or function that runs. If you're coding manually, you get familiar with how your program works exactly, while you're doing it. If you're prompting, you're not (or less at least). So Ai is good to be efficient if you know all the basics, but bad for learning them if you don't. I'd recommend everyone to not use that much Ai, but to do research and use the engines, libraries, plugins and open source correctly, that way you have to type even less compared to prompting.
This is an important comment. Thank you for sharing an invaluable perspective on this.
All the naysayers of AI need to remember one thing:
everything you say it can't do now, it will be able to soon.
@@jdmjesus6103 You know what AI cannot do? Reverse the possibilities of a human, who has more knowledge and knows more specifically what he wants to create, vs one, who does not. Especially within software, human knowledge (+AI) will always be superior to just AI, because it is about making services and products for people. No matter how intelligent AI gets, it will never become human, there comes a point where it just cannot know what we are/want as well as we do. As stated before, I am not against using AI. But it can lead to a bad workflow, as soon as the prompting human(s) stops understanding how the program works, soon after will come a point where they don't know what/how to prompt anymore. Imagine a single person prompting to make a software, implementing features for 5 hours, everything works fine. The next prompt they try does not lead to the desired result anymore. Then they realize the generated source code so far is too much/complicated for them to read. To figure out the next correct prompt, they'd have to understand the whole thing or they are stuck. And the difficult thing about software engineering was always evolving medium sized programs into big ones, not making websites or flash games. So feel free to prompt, but also make sure to keep up with the results of those prompts. Understanding is the only way of control when it gets more complicated. If you want to make software tailored for humans that is. If you want to make software for AI by AI, just step out of the loop and take a nap.
@@thFaust respectfully, you're wrong. You're really not looking at the bigger picture. AI will rapidly overtake human ability in every aspect and every possible way. It's happening, and pretending it won't won't change that.
To use your example, AI could easily rewrite the entire code to suit the new requirements in no time given enough processing power. But the time will come where it won't even need to, it will have already anticipated what you were about to ask of it.
I'm not saying this will be happening next year, but it's coming. Pandora"s box is opened.
@@jdmjesus6103 If that is your belief, I have nothing against it. An interesting question would be, what conclusions do you draw from it? What is humanity supposed to do? Do you have any practical use from approaching the topic of AI with those assumptions? Even if your prediction would be more likely to become true, wouldn't it be more intelligent to assume a perspective that is more pragmatic, that can lead to concrete goals of what to do?
I'm a musician.. Ai is about to take over the music industry. It's song writers who are going to feel it the most. I'm a player most of my work involves weddings and teaching students. I was really hoping Ai would take over that too but for some reason Ai doesn't seem keen on lugging a PA system, lights, instruments etc to a van, then drive that van for 2 hours on a Sunday morning to set up and ready to play by 7pm that night. Then entertain 150 people for 4 hours as they progressively get drunker. Then at 1am in the morning pack all that stuff up and drive the 2 hours home all for the handsome some of 300 dollars each. It would have been 400 but like every time the people say we have 5 other bands that will do it for less money. Please Ai take this away from me😂😂😂
Im also a programming now have been learning guitar for some time, now taking it seriously i believe if ai take over i can still survive doing some random gigs 😅
funny how almost all programmers have some interest in music. maybe because it's both about abstract thinking. one tickles the mind, the other puts the mind at work.
Currently in art, the human is missing. Music is auto-tuned, quantized and full samples. Visualarts involve photoshop and stable diffusion/mid journey. Communication is often via Instant messaging. It is the day and age. And art seems to be worthless.
But who wants to see a juke box at a wedding?
AI generated art may imitate or create pictures, but how can a machine imitate feelings from one human to another? How can an MP3 give us the same feeling as a human,.making mistakes playing an instrument, in front of you. In flesh and bone?
Recently laid off at Microsoft after 10 years in the data space. Two months in and I’m still struggling to find another job. AI is going to kill jobs and salaries. Initially it will be cost cutting in big tech, to fund data centres but in the medium to long term it will destroy the value of experience.
Publicly listed companies only care about satisfying shareholders, if they can cut staff to save costs, they wouldn’t even blink.
What im wondering about is how someone who worked for a big company like Microsoft is unable to find another job? So students who just have graduated from CS/SE have 0% chance????
@@ey00000well sometimes a person with 10 years experience can be replaced by someone cheaper.
@@ey00000 i think they have different salary requests
Ai / robotics coming fast. Even if you get a job, it will be temporary. 2025 will be the YEAR of Change.
Do you think Data Engineering or Data Analytics will be in danger? Which will be more in danger?
I'm a computer graphics artist. I spent years learning different skills, including programming, and even made a small game. But for about 10 years, I didn't make any money from it. I kept trying, but nothing worked out.
Recently, I finally started making some money as a freelancer. But there's no job security in this line of work. Nobody feels bad for me though, and I don't feel bad about the things I learned that turned out to be useless.
I get it if you're a programmer who got lucky and landed a high-paying job. You might feel really scared about AI taking over. But just because you've been lucky for a while doesn't make your worries more important than anyone else's.
Maybe you need to accept that life is unpredictable. Try to go with the flow and find ways to get by, just like those of us who haven't been as lucky.
Wow... You just put all that bitterness out there...
@@stinger4712 not wrong tho ai will replace human ses
@@stinger4712 I didn't detect any bitterness, just acceptance.
@@WillyJunior tomaytoes tomahtoes.
@@stinger4712 no, bitterness is not a synonym for acceptance
The thing is that only programmers can make programs, with AI or without. Someone who isn't a programmer cannot tell whether code generated by an AI makes sense or not, even if it seems to work at a first glance.
That's for now. For example ChatGPT4 can help me a lot with generating code fast. It can't generate the whole working application. Yet! I think it is not that far of doing that too.
I was really concerned about chatgpt when I first used it to generate code, but the more I use it, the more I recognize its weaknesses. Its a powerful tool, but just like a hammer, you still need a master craftsman to weild it effectively. Someone who doesn't understand how all the pieces fit together will not be able to replace a senior developer for a while, especially if you are on the cutting edge of technology.
Try claude sonnet, it generates really good code. With less bugs than an actual person.
Can't you see even one step ahead? This is as bad as AI ever going to get. Even if LLMs run out of steam when you scale up this is not going to be the end of AI development. Some other architecture will take us further forward.
@@ManicMindTrick I don't disagree with this, but the key point is that it will be a different tech, not an llm. And even when that tech does arise, if you stay on top of things, you will move into more of a business role that manages the tech. Once the business role is automated, then there is a bigger issue to worry about than your job, because all of society will be crumbling. You can't sell a product to people with no money. Our only hope is society crashed before then or a tech like neurolink turns us all into integrated cyborgs so we have a chance to compete...
@@justinshankle You can't say LLMs can't take us all the way with any certainty.
With scaling and further tweaking or perhaps a new innovation it's possible to go much further beyond today's capabilities. I'm agnostic on the topic as it seems very unclear.
Hoping society crumbles sooner rather than later in order to limit the negative impact of the collapse is Ted Kascziyski's idea.
For a short while humans together with a chess program could beat the best chess program. Now the human in the loop is obsolete altogether. That is how that cyborg stuff will play out as well if it plays out at all when AI reaches a certain level.
I can't say for sure but I think ChatGPT has a high temperature, a bit too high for coding.
Unless they recognize you request code and adjust the temperature, I don't know how they've set it up.
If the AI could replace me, it first would be able to replace all my managers, CTO, CEO...
Those executive decisions are always made after considering many factors where there’s no clear information of how those information are weighed. The data which informed those decision made is obscure and hidden. Therefore, I just don’t see AI replacing human at that level even if AI were to articulate better reasoning and critical thinking processes, politicians and executive directors may still scoff at that. Imagine Donald Trump having the most scientific and articulate advisor for his policy making. Will Donald Trump actually exactly follow that advice?
they stand above it. they are the ones who "prompt" you.
@@dontmindmejustwatching I never understood this stance. If all they need to do is prompt the AI, what prevents other people from making their own companies and doing their own prompting? These businesses would have no moat if all they did was prompt a publicly available AI.
@@0x6e95 nah, that's not what i meant. i meant that superiours at work the ones who "prompt" meaning tell people what to do. So they will just propmpt AI, they don't care who to "prompt'. management will just ask AI what to do and not people, they are immune to AI because they call the shots. if that makes sense.
I liked "big picture designer vs low level tinkerer".
Yep, but we don't need so many of them.
Unfortunately, a "big picture designer" starts as a "low level tinkerer". No opportunities for tinkerers means a future with no big pictures.
As someone who's a developer and who's experimented with ai, I've noticed ai is precise and detailed with a 'tree' or 'bark' within the forest but is completely incapable of seeing the 'forest' or understanding a layer within a program and how it works within the broad framework. And if you give it a command to abstract and focus outward, it begins to make mistake after mistake.
That's BS way to distract urself into doing whatever the fauck u want
@@steveh.7664 Yes, but that is for now.
That leaves out all the once very productive programmers who lost their jobs in the past year or two that are approaching 50! There is nothing out there even those who keep up to date on everything, as you mentioned, and all the latest ChatGPT newest versions, that BTW still hallucinates now and then. Its hard to see so many in my once heavy tech city having to start over with a new career choice because no matter how well they excel, they are still not wanted due to age, the younger ones disposable as so many, and a now mostly vacant tech area downtown. Near empty towers. I know someone in particular going thru this now as well as going thru his savings, and its very difficult. I am not a programmer, but a mom, and speak up for one as he excelled on the job at every level, but now wonders what now kinda thing. All the best to you Lex for at least making it a topic.
Dude, I have 45 and I want to enter the field as a junior. I can make the whole app though and I am eager to learn more.
Ai is really useful at writing snippets. There are problems that can't be expressed by using human languages, you have to use computer languages to describe the problem in details. This is why computer languages exist in the first place
denial stage sucks the most. see you there on the last one.
Remember that humans pay taxes, programmers contribute a lot to that, AI‘s don’t. Many governments are aware of this, some companies too. Economies would have a BIG problem if knowledge workers were to vanish.
Companies need people to buy shit, AI does not buy food, medicine, clothing, transportation. Thats a lot of tax dollars missing if AI did replace so many people.
first I thought the same but as AI develops so will society change. If the rich can have their AI driven world that satisfy their needs they don't need to sell anything to anyone.
@@Qrexx1 you completely missed the point of the comment above ffs, you are hallucinating with the "evil rich" in your mind
@mattb925 The first point is laughable cope. The government prints money. They don't care about your tax dollars.
Thats not how the economy works lol.
Software gets cheaper, people have more money to buy other things. That created jobs and income, which created more demand which creates more tax dollars… and so on.
The economy is an ecosystem not a linear machine. And it’s definitely not a zero sum game.
@@JonasMunnich what jobs? there will be no jobs if this gets to that point in the near future because all of them can be done by AI.. what means having people with no jobs? You would have a gap between a certain mass of people with jobs and ones that don't and can't afford anything/living I think you got the answer. And this thing might come soon.
Not just programmers...they will replace all jobs....i recently say a automatic cleaning robot in a mall and realised that everything will be replaced...some sooner some later...
I ask chatgpt for help all the time, I get it to write big blocks of unit tests
What keeps getting overlooked in this issue is that more and more code will get created in the future. So LLMs will be a productivity tool for programmers (not replacement) and although you might fear this still means fewer jobs (one programmer can create more code in shorter time) in fact you'll see just more code getting created. The worlds hunger for code is only limited by the amount that can be created. It's wholly determined by supply, not demand.
What makes you think more code will be created/needed? There is already a competition between sotwares in almost every aspect of the field that not differ much from each other, so there is already an overproduction of sw.
I don't think you understand economics.
@@fragebogenvbcAs the world continues to embrace technology; the demand for software will and currently does far outpace the number of people who can develop said software. If AI ever does take off and become more then what it currently is; it will enable companies to cater to those needs and the amount of overall production of software will level out with the demand for SE.
@@apricotmadness4850 currently absolue not outpace. there are 100 thousands of unemployed developers in the US currently. And if AI takes off, the increase of the number of human written sw, is only a therory without any solid base.
@@apricotmadness4850 Why would u as a company want to have people who code worse then Ai it doesn't make sense it's much better to have an elite just manage a bunch of Ai since Ai don't need to sleep, time off or even eat it's more productive.
I think it's pretty silly pay a programmer when Ai will work faster then that human Alpha 2 already performs better then 85% of coders in 2-4 years coders will be obsolete we saw this in music and art and it will happen to coding too.
Don't think companies will not want to cost cut since they already are cost cutting as much as possible like Amazon are replacing 5 workers with 1 and just forcing them to work 7 days a week.
It's actually crazy programmers don't understand such a simple concept everyone is replaceable, Amazon is using Ai to replace managers so 1 manager do 5 jobs at once efficiently and just as effective as before if have Ai that can code why would I even bother having a team of humans when 1 human can do the work of a whole team with Ai.
If replace 85-95 percent of coders with Ai how much would I save as a company?
A good example would be google avg medium salary is 147K a year around 60K employees so multiply 147K times 54,000 you get 7,938,000,000 so why wouldn't I replace most programmers with Ai if I can save almost 8 billion dollars?
Thanks Lex for that recommendation. Downloading Cursor now!
The problem is the fear itself
The problem is faucking real
Mass unemployment loading sooon
@@developit1152 💯 ... expect unemployment figures to tick up monthly as businesses look to be more efficient. Larger companies will restructure. Consolidate departments. The savings in salaries will go to Ai companies. Companies will choose the brightest and most efficient to lead AI agents.
@@developit1152 We have to stop it
I am a lead software engineer with 13 years of experience. In short, I can say: don't worry, it can't replace developers. What it can do is reduce the number of software developers by up to 30%. And this is still disastrous.
Let me tell you why it cannot replace developers.
First of all, the current approach of LLMs is learning solved problems and solving similar problems to these.
What they can't do is combine many of these solutions to solve your custom problem.
If there are just a few requirements, LLMs can write nearly perfect programs by combining solutions to small problems.
But if you have many requirements, which can be solved with at least 3k lines of code, then LLMs can't solve these problems.
The reason is mathematical. Every time you add a new requirement, the complexity of your software increases exponentially.
Without quantum computers, it is not possible to solve problems with algorithms that have exponential complexity.
Assuming we will not get quantum computers soon and not change the current LLMs' problem-solving approach,
we can say that AI cannot replace programmers entirely.
have you tried the o1 model? i love trading stage too, it's comforting.
It can go a lot above 30%, but it will never be 100%
I tried it and it's good at doing small-ish tasks that have been done many times before. I noticed in the output that most of the code was more or less copy-pasted from existing github repos and I could easily track down the code it "borrowed". I tried adding some requirements, and it got completely lost.
I also tried to use it to build a compression algorithm that I had in my head for many years but never had the time to build. It was basically a combination of existing algorithms. The LLM wrote a decent implementation of the known parts, but the result didn't work. One part of my algorithm was using an adaptive Huffmann coder with a growing number of symbols, and the model simply gave it a fixed size memory for those. When I asked it to fix it it did, but then it didn't decode properly. When I told it that, it's idea of solving it was making the output of the huffmann coder fixed size - completely negating its value.
That being said, it saved me tons of time in writing the well known algorithms so that I didn't have to and at the same didn't have to feel bad about copy-pasting them from github myself. With its help, I was able to get my algorithm to work rather quickly by modifying its output.
Interesting point of view. Assuming that this sentence is correct: "Every time you add a new requirement, the complexity of your software increases exponentially"
@@bestopinion9257 Actually I was talking about the system complexity, rather than runtime complexity. I mean complexity of building the code increases exponentially for llms, not running the code.
I don't know as somebody who spent several years in college learning computer science I'm horrified that there were certain skills that took me 6 months to learn that AI can do in under a second.
I'm sorry the truth is that it's simply making the workforce more compact yes the best coders will always be able to find a job at least in the next 10 years but the labor and groundwork that used to leave open entry level jobs and internships has been taken up by artificial intelligence.
dont worry bro, you will be allright. This is nothing new whats happening, just the next step on evolution, some people do are left behind but that is for the whole society to be better off. Always have a plan B on your mind as well though, in case you are one of those people that are left behind.
@@Henrique-wy6cv i feel left behind
@@Nicole-m1p4f :(
Top employees benefit the least from AI. These are also the people who didn’t use Stack Overflow. They are solving problems that have never been solved before. This is true across the board with genAI, lawyers, marketing, … AI is closing the gap between the just below average and just above average. The bottom of the pack cannot identify problems with AI content, so they receive little benefit.
As the cost to produce software falls, projects that were priced out will get green lights. There is no way to know if the total developer headcount will be more or less than today down the road.
youll make it
02:40 … do not compete with AI … learn how to use AI to do your job better … THIS MADE MY DAY, tanks @lex 🔥🙏
💯 😊❤
but what is automation?
@@Soulbiez The last step to optimize a certain process or repetitive approach to reach a certain outcome
Will work only for some time.
How can a voice actor use AI to do his job better?
If it can be done on a similiar level for free, your job will be gone even if you got "better"..
@@Felipe-zl1rj interesting hypotheses .. i am not a programmer .. i help solopreneurs and business owners that are serious about AI to integrate AI in their processes, products & Services and into their customer journey ... many of them are fighting against AI and i help them to use AI to become better and do a better job ... I love that.
An by the way .. its not the cost that are killing companies ... its the complexity and therefore the speed that they are able to move forward ... this will be a driver for lay offs .. not the costs that need to be cut.
I would say for a lot of people beginner - intermediate, familiarize yourself with Cursor, but don't use it for everything. For someone like Lex who has the experience and understanding of the tools/languages it will make him faster, but for many it will turn them into prompt engineers rely on prompts to debug.
Im a jr dev. A little over a year of pro exp, but 3 years of programming.
I used code generating tools for a long while.
Like many, i would run into a bug, and no amount of sweet talking the AI would fix it. And i certainly wasnt skilled enough to fix its hyper advanced solution...
I eventually realized that it takes many years to master programming, and far less skill to use prompts. However, in order to know what prompts to ask, you first need the skill to see a bad implementation.
So for the next decade i plan to become as good as i humanly can at programing. When ai does take a larger role in code, in some 10-30 years, at least all ill need to learn is prompt engineering.
Perfect attitude, and ironically probably going to be the one that actually goes somewhere compared to the "Well the dev who uses AI will replace the one who doesn't."
Great take.
@@jefferymuter4659 what a waste of time!
The best advice I have ever been given is to learn concepts not syntax if you know concepts now you can have artificial intelligence write code for you and you can rearrange it or tell it what to fix because you understand the concepts but syntax is constantly changing and if you've spent constant hours like most of us learning syntax while that's essentially just wasted time.
I wish this was there when I was studying CS
@@78alJ0vle The engineers who benefit the most from AI are those just below average. It takes them a little longer to get their work done by hand and they at least have enough experience to realize when AI is hallucinating.
Leetcode is about concepts though
@@vaolin1703 I guess I'm referring to cramming the code like a lot of people do where they're just trying to basically be able to just do any of the problems because they went through it through the website, no I agree like if you understand it from conceptual point of view then you're great
Leetcode is literally concepts. What
I'm a videographer, I would never have the time to learn programming nor would it be worth my money to develop programs that assist me in my job. HOWEVER, with claude and chatgpt I have built several programs that are specifically tailored to improve the efficiency of my workflow. It is fantastic that basic programming is available to the everyman now.
can you add examples on how you use claude and chatgpt in your work
@@evaanderit can’t cause it’s a bot.
@evaander For one I use it to summarize and organize interview transcripts. But in terms of programming, ive built a program that essentially organizes my project database according to the exact organization format I use, one that helps manage and organize the file backup process and several more that are more niche in my field that I'd have trouble explaining quickly in a comment section. They all use python with pyqt6 for UI. Only reason I'm bothering to respond is that other dude called me a bot
Haha 😂 good reply! Sad that in this world of AI you have to prove that you're a Human, not a bot. @@Potaters12
@@Potaters12😂🎉 not a bot
Coding is a tool for problem solving. Be agnostic to the tech that helps you get to an optimized solution. Don't try and be a good user of some tool. At this point i've been coding for so many years that I don't see code. I see logical connections. Whether I explain these abstract connections with code or natural language won't matter. I'll do whichever is most efficient at the time.
Better advice, don't worry about the coding part so much. Improve your skills by improving your knowledge of the world and industries. Jump into bioinformatics and begin hammering away at some DNA sequences through various algorithms. Take me, I developed a awesome NLU engine for Cicero project that requried me to learn all about NLP in-depth, now I'm just finalizing NeuralLink's compression challenge which took me through the world of audio transformations and varying compression techniques, and so on.
I don't know, jump into material sciences, genetic engineer, signal intelligence, or whatever. Expand your development skills that wa instead of just trying to learn the next framework, or whatever.
Tell me more about what you did with bioinformatics, I'm studying bioengineering myself and am interested in these stuff.
1. For many decades non-programmers have relied on others to create software for them. Those days are coming to an end and there are astronomically more non-programmers. 2. The market will demand software at a level of sophistication and complexity that vastly exceeds human ability (programming, code-review, maintainance, refactoring, scaling, etc.). Youngsters should investigate opportunties in QA, UX and CS - all relevant disciplines that will be needed long-term. You're welcome.
Can u explain the acronyms?
@@Ращин-Рулет Quality Assurance, User Experience Design, Computer Science
@@liminal27thanks
If programming can be fully done by AI, so can essentially any other job done on a computer. What would prevent it from interacting with CAD programs, collecting data and interacting with Excel, etc.? Who knows what will happen.
i believe we need to rethink the fundamental approach to designing these systems, one last time. as scary as it might sound, if you simply train these models to be self interested you would solve a lot of problems of not generalizing knowledge well.
Yes but probably not as quick
A lot more data to feed these LLMs in regards to code than most other things. That’s why they’re already decent vs others things
About CAD: Drafting plans may be gone soon enough, because that’s mostly technical/mechanical work, but a human needs to oversee the design, the overall process, to coordinate and check AI’s output. At some point in time, I assume, it will be possible to create a detailed task for AI (basically, the same as you have in contact between the Current and the Architect, describing what exactly needs to be designed) and AI will create options. If AI will be able to learn all the Federal and Local Codes, plus best practices, you would need not 100 but 10 professionals in the company. And to be among those 10 lucky guys, you gotta step up your game.
All mediocrity will be substituted by AI.
What you describe would be easy. The difficult part for a human would be all bullshit syntax you have to know and type correctly each programming language involved in the process to work
yes, AI can do any job basically. just give it time. no magic there. nothing specific to humans that can't be automated.
I am an Illustrator and 2D animator, recently I decided to go back to school and learn Software development/programming, since I was young, I always wanted to learn programming as an extra skill besides what I already have. I’ll be honest: I’m still a beginner in coding. One day, while racing to finish several assignments, I decided to try using AI. To my surprise, it worked! I took the instructions for my assignments and asked the AI for help, which provided me with the JavaScript and HTML codes I needed, I was too lazy to type them, think deeper etc... I submitted it and passed, but afterwards, I felt guilty. What’s the point of studying this if a computer can do everything for me? I realized that I relied on AI for about 95% of my assignments.
I’m uncertain about how AI will be used in more advanced programming contexts, but as a beginner, it honestly freaked me out. I thought about how someone could graduate without truly having proper knowledge of programming. AI is definitely going to change a lot in the lives of programmers, bringing both positives and negatives. Many jobs may be at risk.
In 15 or 20 years, senior programmers might retire, creating space for newcomers, but it will take highly skilled programmers to compete with AI. The tech industry, and really, every field, including art, music, and writing, will likely be dominated by AI, which is a sad reality.
Really liked this take, Lex. And I suspect very relevant to people working in the arts as well. As part of my work I have regular consultations with creatives and naturally these feelings towards AI often arise, and finding an approach which feels both compassionate and pragmatic is always one worth seeking.
The question on AI is not so much a deep fear of slipping into irrelevance because that AI will soon exceed people's programming skills but rather that AI will soon reach escape velocity.
By escape velocity, I mean that the speed at which AI can learn is faster than the speed at which humans can learn.
This creates a scenario where whichever direction that people decide to up-skill towards will be made obsolete by the time the people are finally able to pick up the skill creating an atmosphere of obsolescence, futility, and frustration.
You don't have to worry about AI taking your job... Worry about the guy who is using AI to take your job.
1) learn to use LLM to solve problems
2) get an interview where they ask you about low level shit that no one does at job nowadays even before AI
3) get rejected
4)*confused*
5) *repeat*
I mean, using A.I. still requires understanding of how programming works, as Lex states to edit the code that the A.I. generates, however, programmers feel more like directors now who overview, oversee, and fit all the code generated together. Is it more holistic and less in depth than what programming used to be some years ago, sure, but it still takes skills that much of the population still can't do. Even if the I.Q. level is reduced now to create these apps because A.I. can generate features. I personally love A.I. it makes coding so much easier but of course if it is too easy than you can become redundant! But programmers also had this fear when code editors and debuggers were introduced and if anything just increased the demand!
there will be completely autonomous AI SWE by the end of next year. maybe sooner. See you there on the last stage of grief.
yes, AI needs guidance. It can make and change as you ask small parts of your application. You are the one that understands how all these small parts are fit together. But that's for now. With improvement one day you will ask AI "make me an eCommerce site" and it will do it in seconds. Then without any programming skill, ask AI to change this color, put more menu buttons here, make these categories of products and so on. You do not need to code or read code at that point.
@@dontmindmejustwatchingHow do you know that? You realize your prediction is radical? I net you have more than a gut feeling to back up a radical prediction.
It just sounds like Shapiro talking about AGI 7 months ago.
not everybody wants to learn coding so there will be less demand
We need UBI
The collective intelligence of Stack Overflow exceeds the skills of my cat chasing its own tail.
Investors are asking a different question - will LLM development firms be around in the next one to two years - aleph alpha recently pivoted away from building large language models because the costs are not sustainable, particularly with diminishing improvements and increasing competition as new models from various vendors come out every week. Will the GPU gravy train continue? and for how much longer?
one could certainly run their own open-source LLM, but even that is expensive, whether on the cloud or on-premise. Is the cost really worth the return over time?
on an investment mindset AI is very expensive and returns peanuts. But if you think away from businesses, you'll see we're not competing with businesses, we're competing with China and Russia.
Yes. And they are called Amazon, Google, Meta, Microsoft and Tesla.
@@jurassicthunder Except we're outspending them by huge margins... honestly I think China and Russia only care about it because the best practical application for current GenAI is propaganda and misinformation... not the ushering in of a new Sci-Fi world that AI enthusiast like to pretend is happening.
This was a really insightful piece on the evolving role of AI in programming. I appreciate the perspective on embracing AI as a tool to enhance our skills rather than fearing it as a replacement. Transitioning from a low-level tinkerer to a big-picture designer resonates deeply. It's a reminder that adaptability and continuous learning are key in any field, especially as technology advances. Thanks for sharing your thoughts and experiences, it's encouraging to see how we can leverage AI to become more efficient and creative in our work!
LLM can’t read your mind. You have to spend significant amounts of time crafting and refining prompts to match your specs. Time that could have been spent designing and writing the code yourself. AI does not boost productivity for real life problems of medium or higher difficulty. At least not until we find a way to feed it all the context it needs to understand a problem in its entirety.
Musk’s neuralink to the rescue who knows 10 years down the line people will have implants like that
So requirement gathering, and understanding the business matter all over again😂
Dumb comment, writing prompts is much faster. Also, AI will improve. You are judging from the earliest versions.
It's like someone looking at a Computer in the 1970s and saying it's useless
@@rameeziqbal8711AI actually started in 70s and those were the ealiest versions. What we have today comes after 50 years of developement. With like 10 years of it being used widly in some sectors, like image recognition.
Years ago, being obsessed with money was psychopaths' domain. Today it seems to be everyone's.
Yeah I haven’t met any working developers that feel this way. Personally, when I was relying on co-pilot full-time for a couple of months, I felt like my code quality suffered heavily. Another anecdote, I’ve seen some of the code that juniors on my team have produced with the help of LLMs and it will dramatically mislead them into a bad direction/design.
From VS Code to Curser...So swearing is supported by AI now.
Cool! That's what I needed!!😅
1:42 "and before that it was emacs to VSCode switch". I can relate!
Beautifully said, Lex! Couldn’t agree more!
Thx for the editor
What people don't mention enough is how it wastes more time than it saves when you actually need something complex to be deployable. Use it only for specific parts or components of your app or site. It is not going to replace programmers in the next 5 years period. After that it will replace us humans completely so no need to worry about losing your job over loosing your life.
LLMs can't innovate. It can only spit out code that looks like something it has seen in the training set. If you do anything a bit off the beaten path it's not going to be much help. It can save you some typing. On the other hand if your job consists of copy pasting snippets from Stack Overflow, then yeah your days are numbered.
Im a lead engineer, Ive built some things recently i WISH AI could of helped with. Try having AI build anything with drag n drop, or something with lots of events that coordinate together and see if it doesnt create race conditions. like a wysiwyg editor. Ultimately these things i had to build myself. I also built an agent framework that cursor failed to understand how to add support for claude, i also needed to do myself. There is still so much it fails so hard on, when its not some simple task.
Yes, many of us wish AI would work for us as it does for others, creating simple CRUD apps for common entities in a generic framework. In real life, some of us our really trying to raise the bar for our users, but that also means we can't copy and paste some standard app.
Have you seen the AI Doom paper? Zero programming required. The AI just 'imagines' it and it appears. There are no complex algorithms, no state machines, no design patterns, no optimizations - AI makes them all obsolete. Most people seem to be in denial about what the future holds.
@@rickenbacker472 > Human raters are only slightly better than random chance at distinguishing short clips of the game from clips of the simulation
What about long clips? Like a whole level.
@@rickenbacker472 yes, I work in AI, the AI doom where nothing was the same enemies fade in and out of existence, the score was inaccurate. And no persistence or consistency. The AI doom paper is solving a different problem. It doesn't generate anything consistent, but generating scenes, worlds, custom on demand. It's built off diffusion model, and is indeterministic.
@@lifeofcode Yes I know. But it is part of a larger trend. A trend that is far from over. Let's see what is possible in 5 years.
I think just like knowing lower level languages makes you better with higher level languages knowing programming at all will make you better at utilizing the AI
All of these "Will AI replace X job?" are almost always coming from people who know nothing about that job.
I love it. A better world awaits.
Thanks for the Cursor recommendation, will check it out.
I use Claude Sonnet as well, but Gemini Pro is also knowledgeable about some relatively obscure languages.
Concise and clear video - excellent.
One point not mentioned : the AI-adept developers will be very productive which is most likely going to lead to layoffs of the less capable.
I would expect at least 30% of software developer roles to fade away over the next year or two.
We need more Q&A videos ❤❤
100% agree. I still do design, but I think about my AI prompts as much as I do design these days.
Curious to how others workflows are changing, what tools have you tried? I've been experimenting with Claude-Dev, Cursor, Aider-Chat. Not settled on any one particular tool ATM and just trying out various tools and workflows.
I love claude AI too. It makes me more to an product owner, having very precise requirements and getting very good results for my problem.
Didn't answer the question in the title of the video
No its just a tool there's not enough energy in the grid to power the tpus and servers that run these LLMs, it would cause major blackouts to replace thousands of employees.
denial stage is the worst. see you there on the last stage.
@@dontmindmejustwatchingwhat do you mean can you elaborate?
@@infinityentity3128 ahhh, ask chad gpt about stages of grief?
@@dontmindmejustwatching In 5 years, I'll still be making more money than you by writing typescript on nvim
I'd love to see a survey across programmers of different experience levels asking how easily they think AI can replace programming. Or even more debate/discussion around what it means or what can be replaced, but I mainly see sensationalism of "All coding is dead" or at the other end what feels like denial "It will never replace human programmers" haven't seen much good in-depth discussion though yet. I'd love to hear from more career programmers not necessarily associated with AI and hear their thoughts, as to some degree anyone working for an AI company will obviously be very bullish on it replacing programmers.
LLM as coding assitant saves me a lot of time and effort where I can instead use my energy on the creative parts.
Code is for humans. If AI is to replace human developers, it shouldn't have to write code. We would just tell it to build a system and it would produce the binaries that solve the problem. As long as it merely generates code, it will always need a programmer to know what to do with it.
I rate AI just above noob level. Syntax is good, but they struggle with systems and comprehending complex apps, especially ones that are dynamic and creative. Good for assisting but very prone to suggesting faulty code.
I have to fix all the code LLM writes, and it has such a subtle bug that I have to really think about what is wrong. Its playing with my perception. And it takes more time to fix the AI-generated code than to write it by myself. It is beneficial to generate boilerplate code. And this way it's really useful. It's not scary.
Keep moving up the value chain. The more output a programmer can produce (eg via AI), the more value created per hour, the cheaper per finite output, the more demand for that output. Elasticity of demand suggests that if the cost per output falls, the demand for such output increases by more than the proportionate decrease in price, leading to an overall increase in total revenue and market size. So as productivity rises and costs fall (e.g. due to AI), new opportunities emerge for consumption, with businesses/individuals purchasing more of the output, further driving economic growth and demand for skilled programmers capable of leveraging these technologies. In short, there will almost certainly be future demand for competent humans who can nagivate this space and help others.
sounds like it is a good time to have Stephen Wolfram on your podcast again .. always enjoy listening to you two
So yes
Cursor is amazing. I haven’t had the title engineer in 15 years and I’ve been able build useable software using python. I really don’t know python but I know how to focus engineers to build what I need as a Product manager. I also have general troubleshooting skills and cursor is an excellent pair. Now I need to learn TDD so I can confidently let the AI refactor and make sure the tests pass.
Good answer. I reached the same conclusion. AI has become a part of my workflow as a developer. But sometimes you do find yourself wrestling with the code it generates 😅
what is your advice to a new learner that want to learn programming?
Hey Lex, have you tried Replit yet?
From a person who does not code and do programming for a living, coding is just a part of the architecture, and like typing, spelling checking, and translating, it's better to have some help. It's time-saving and good for health.
Hey lex just wanted to ask you that if someone wants to become ML engineer. Is it important that you must have a strong knowledge of DSA. Or if you're not that good at DSA , can you still learn ML
Yep I see coding as the most useful use cases of AI because coding is a completely closed box problem. you have functional requirements and the code can be tested against those function requirements. AI can debug it’s code by running the code repeatedly..
if ai replaces programmers then everyone will be replaced
AI won’t be replacing that many blue collar jobs. Most white collar workers are fucked though
It's how government handles the new unemployment
Not really; doctors, lawyers, psychologists, philosophers, historians, professors, police force, military personnel, painters, journalists, biologists, chemists, physicists, geologists, astronomers, vloggers, anchors, podcasters, hotel management, chefs, sportstars, actors, directors, project managers, etc will never get replaced
@@rameeziqbal8711 Lawyers will definitely get hit hard. As will many on your list. Profound changes are coming. If half the workforce is not needed, accommodation needs to be made. We can't have rampant homelessness. Will be horrific otherwise. It's already too much.
@@tommynickels4570 How will lawyers get hit hard? They indulge in a lot of human interaction. How you deliver speeches, communicate, ask questions, plays a big role in the outcome of a case.
Short answer: no. Long answer: no. Even longest answer: no.
Most longest: until it does
@@haldorax that cringe unemployed ahhhole giving people advice based on reddit/ internet discussions 😊
How is “writing natural language prompts” different than writing a technical specification? We are going back in time, and we won’t need as many developers, especially for code that is more utilitarian like most enterprise software.
It could go either way. If development is cheap enough, all the SMBs that couldn’t afford custom software will suddenly be hiring. Take a look at what happened with bank staff and ATMs
Hi Lex. I am a writer and I am using my creativity to integrate ai in my writing process. I'm not using ai to replace my ability to write But find ways to make me a better writer. It definitely allows me to be more efficient in my writing process. Thank you for your comments about how to deal with AI as a programmer which applies to almost any field that might be affected by the new wave of AI.
Kinda ironically that programmers thought they would be needed the most (in AGI world), when they're among the first to be taken out.
I have no opinion on the statement, but it is what it is. We think our work are special, but it seems are not.
Lex Friedman for presodent of the United States. 🇺🇸🇨🇦
I think it’s just a good google search. I don’t think it will replace us anytime soon because it still doesn’t understand complex business rules for your company and in my experience it just spits out garbage code. It’s really good for school projects though or any other basic stuff
I think AI will replace a lot of the frontend UI work, or at least change it substantially. Why pay an engineer to implement your figmas if an AI can do most of it and then the engineer hooks it up to the rest of the app.
Soon there will be theme parks so children can look at people in period costumes in offices sitting at computers and writing in C.
I think the kids will love it. Most exciting day of their life.
Great take - AI will not replace programmers. But we do need to embrace it so do our job better. That being said, the current state of AI does not help us to do our job better (maybe it does, if you're junior developer, but let me tell you, it does not help you in professional fields). And nor do i think it will be substantially better in the near future because we are plateauing in terms of AI progression. But as Lex said, it could change in the far future and we just have to be willing and ready to "adapt".
I user Claude to do the scafolding. Nowadays, I do 80% to 85% less manual coding than a couple of years ago. Yesterday, for instance, Claude did to me in half an hour* what would have taken me 4 to 5 days. (* it took me half an hour to interactivelly request small changes to my code, until I got it exactly the way I wanted.) So, right now, if you don't use Claude (or others), you are wasting time. Eventually, in a couple of years, all of my code will be done in natural language, in interactive sessions. This is going to be the next step in the "High Level Languages", I think. English is the new Python/JS/Rust/C/and-even-assembly.
First want to say I agree and you're right.
But here is another fun thought experiment.
Aren't we wasting our time right now?
We can build something in say a week. And we continue to do this week after week till the end of the year.
Then, at the start of next year, the first week of 2025, that's a single week... we can get as much work done as we are doing today in 8 weeks.
Then why don't we just chill for 8 weeks right NOW, and just have it generated within 1 week at the start of 2025 ?
Maybe we should all just go on vacation for a few months and then generate everything that we would've written this year.
And then, we have got just as much done as if we would've spent the time working instead of going on a few months vacation.
"Eventually, in a couple of years, all of my code will be done in natural language, in interactive sessions" . Devil's advocate here - why would anyone pay you a high salary to do that? Any junior or even every product designer would be able to do that. If you have years of deep programming experience they'll become nullified and probably so will your employment prospects.
I'm still hoping that like in autonomous driving, the last 1% of accuracy would be impossible for LLMs , thus necessitating humans to still understand and write code from time to time.
@@יואלבלום don’t fall for this hyping. They are all trying to sell AI company stocks and a bunch of ignorant parrots who only repeat a bunch of nonsense they read somewhere (like singularity AI overlords etc) else trying to feel special. LLM‘s are just a glorified search and without developers who know their stuff it’s practically useless for anything more merely complicated. Education is what we need. People need education.
@@יואלבלום Which is precisely why people should change their career aspirations away from being a master programmer. Unless you are aiming at becoming an AI programmer. The writing is on the wall for those paying attention. While I don't like using this analogy since it's apparently a myth, people asserting that LLMs are not a threat are the frog in the pot of water increasing in temperature.
is JetBrains' AI any good? I haven't tried cursor
If it can replace programmers it can replace all human productivity
The thing is, AI is a tool-learn to use it. I’ve already used one to help me put a game idea together. It’s not coding, but I now have a 360-page blueprint of an MMO ready to go. Soon, I’ll just need to feed that into an AI to start programming it for me.
I have spent several months learning how to avoid coding. I have managed 98% AI generation using Aider. Full stack.. truthfully front end is still weak though. It can build but it still struggled with UX but the backend stuff, it is almost there.
And now o1 has massively increased coding capabilities. Junior positions will evaporate
Has it really through? It really isn’t that impressive. It’s not even chat gpt 5. It’s just the same technology they’ve been using with different methods,
I honestly think that the careers benefiting the most from AI are design and creative ones, I'm a 3d designer and I've been developing insanely useful tools for the 3d artists of the company I work for, because I'm the end user of it, so I know our "pain" as a 3d artist better than any programmer. My brother on the other hand is a programmer and he is expressing the same fear you are talking about in your video.
Truck drivers won't be replaced
They always need 'low-level tinkerers' to build something great. But once something is built, they forget that fact and only admire 'big-picture designers.' Crazy times for artisans.
I guess it all ties into the 'control problem'. I think it's inevitable that AGI will replace programmers. It's still a hypothetical, but the 'race to the top' will bring it. The question is: will AI replace carers? If so, that's me out of a job, too. Somehow, I think that's less feasible!
Please make more videos about ai-coding!
As a non-programmer AI can write me programs.
I think getting into programming would be a disadvantage.
Not care about the code, but only for if its working, doesnt occupy your mind with irrelevant cognitive resources.
If AI takes my job I will be like, okay, congratulations, I can finally go and have my own farm and have animals there and I will know, AI is changing the world on my behalf