Full podcast episode: kzfaq.info/get/bejne/hsp1YNqSuN6RlGw.html Lex Fridman podcast channel: kzfaq.info Guest bio: Stephen Wolfram is a computer scientist, mathematician, theoretical physicist, and the founder of Wolfram Research, a company behind Wolfram|Alpha, Wolfram Language, and the Wolfram Physics and Metamathematics projects.

To paraphrase Bjarne Stroustrup, the creator of C++ : if we had a dime for every time that programming was supposedly dead, we would be all rich.

As a programmer of 20+ years who is evaluating AI I am of the feeling today that probably on 20% of what a programmer does can be handled by AI. Too many proprietary databases, business rules and algorithms out there that are not part of the ecosystem AI draws from. If AI can make programmers 20% more efficient? Fantastic.

LLMs currently, are to a programmer what a calculator is to a mathematician. Being able to use a calculator doesn't remove the need to understand what calculations need to be made. Being able to use a machine to generate code doesn't remove the need to understand what systems and subsystems are required to make useful, stable, secure, efficient, ergonomic software.

It's not code generation that's the hard part of programming. It's getting the correct requirements from your customers.

Every place I've ever worked at had an endless "want" of projects, the only thing stopping them was time and money. With programmers being 50% + more efficient, more will get done, it doesn't necessarily mean we'll have some mass layoff event. This is just another abstraction layer in software development. With AI we can zoom out even further and solve bigger and bigger problems faster.

As a guy who studied linguistics and decided natural language processing was too hard, prompt engineering sounds like my ideal job 😂

Natural language prompts are fine, but just imagine if there was a more precise language we could use to prompt a computer ...

Programming isn't dead, but it is changing. There will always be demand for people that can understand and communicate in a way computers understand. If you think about sci-fi, we've been predicting this AI ascension. Humans and AI will work together, and in that endeavor there will be triumphs, and there will be failures.

When you want something specific, it can be more efficient and more predictable to write it yourself (code or document) than to explain it in enough detail to point the AI in the right direction.

People think coding isnt a usefull skill anymore cause of chatgpt and other things. But they forget that we have calculators and still people learning math. So as long as we need mathematician we also need programmers.

I am building a video game and just learning how to build the game with some friends who know coding at a far deeper level than me has allowed me to understand the types of questions I need to ask, or how to go about implementing a function or what have you. This is from somebody who is technically Savvy but also I'm more on the art side and creative side. Being a full stack developer of any kind is a huge benefit now you can't just know one thing.

4 months ago I knew nothing about LUA. 4 months later I still know almost nothing, but using GPT 3 and especially now 4 I have taught myself enough to make some pretty cool mods for my favorite games. Just a personal example of using it before learning it.

AI to programmers is what an impact wrench is to a mechanic. It's fast, it's effective, and 100% will break things.

As far as my current understanding goes, I believe that the future of prompt engineering will eventually lead back to some form of programming languages. According to OpenAI, a large part of the reason why some prompts work better than others is that they more closely resemble the way the model was trained. Right now models are trained for natural language, which contains a lot of noise, inconsistency and is just overall finicky and easily misinterpreted with all the possibilities for double meanings. If we want to build very specialised models for certain tasks like programming, it only seems natural to try to more strictly define a prompt language for overall better results.

I can't get useful code out of ChatGPT for my day job. The reason, I think, is that I work with proprietary languages with closed APIs. It seems to work for open languages such as Python. This did not stop ChatGPT from claiming it could help me write the type of code I need. None of it would compile, and appeared to be completely fabricated syntax. I'll check back next year or something.

No. Just because an AI can write code does not mean it will easily add code to an existing codebase. Or fix a bug. Will it even understand how to do that? And, are you just going to let the AI commit code unsupervised? Or supervised by a non-programmer? Who will tell the AI what code to write? A non-programmer? And how will the AI do debugging? What if you need to reproduce a bug that is hard to find? I mean, software engineering is far more complex than people think.

The “prompt hacks” discussion of whether we will find some way knowing the structure of LLM reminds me of the neuro linguistic hackers from Snowcrash by Neal Stephenson

This why I always say a CS degree is better than a bootcamp: Change happens rapidly in tech, and sometimes the shift is massive. A CS degree (or a CS level degree education) allows you to pivot much more easily and ride the waves or change as you have the deeper understanding of what's happening.

It will just shift from writing in C#/Java/Python to writing in plain English which funnily enough was what we wanted to happen when COBOL was designed. Those good at writing the specification in clear and concise ways will still be around. I have been writing guides to prompt engineering since Feb and it’s basically a new higher level programming language but it will single the best developers/analysts out now and the space for those robotically writing someone else’s design/spec will slowly fade away. I’ve been in the game since the 80’s and this just seems like the predicted progression talked about in that era too.

I have found while I agree that we will need to do less "repetitive coding" tasks understanding how to actually do these tasks in python or another language has helped make my prompts much clearer for chatgpt and it gets to the answer I want in much less time. Now you dont have to know how to write for loops to tell chatgpt what to do but I think it will just be a benefit to understand what it is likely trying to do in the background when you ask it to perform a coding based task. Also I have just found the process of creating something useful or cool from typing extremely fun so while it may not be a job I think coding will still stay a hobby for some so I dont know that itll ever be dead.

the fun, for me at least, with the kind of engineering I do is the challenge. I don't have all the answers, so researching, applying, testing, and all of that is fun. I am quite sad to see that it may go away as a career, but I will probably still do it for more personal endeavors.

Programming is not dead. It is more important than ever before. Wolfram just wants to promote his no-code platform and it benefits his business to say that humans should become dumb monkeys who click on buttons. Programming is NOT dead. Wolfram loses all credibility when he says things like prompt engineers are more valuable than software engineers in the job market.

2cents: A programming language is a prompt for the computer, we've been writing prompts all along. The thing that's changed is that now it's possible to write prompts in plain English (or your written language of choice) some people are discovering that it's actually still difficult to capture and communicate all the requirements succinctly and accurately in English. Perhaps this alone will eliminate a middle person in some sense, but I think that also does a dis-service to the role of software engineering, which is more complex than just translating requirements into functional code. There are many implicit requirements which must be discovered and adhered to, and their understanding of the technical side of the project and what can practically be achieved helps inform the future requirements of the project. Once you have an AI capable of fulfilling that role effectively, well, you probably won't even "need" "drivers" at that point, they'll be capable of driving the whole thing themselves...

The skill of being good with computers will probably just become even more important. Maybe you dont have to write as much code in the future, but knowing low level code like Assembly and C++ makes you much more skillful with a computer so I would expect it to be a plus.

I see Wolfram, I watch. I am a simple man, he is a legend.

chadgpt's transcript summary: "The speaker talks about the future of programming, where more people will be able to access computation and generate code without prior knowledge. The speaker discusses changes in programming education, and emphasizes the importance of learning practical programming skills. The speaker also highlights the increasing relevance of computation in many fields, and suggests that it will be essential for people to have computational skills in the future."

There's such a lack of agreement about how this will turn out that it's hard to know what to believe.

Another great interview Lex! I believe if we ever use AI generated code without real supervision we will end at a very dangerous result. I think the reality is these tools will be used and perhaps over time write a great deal on its own but there must be human oversight.

As a mediocre programmer, I do not know where this will take me in my career. As someone good at technology, I am excited about the future.

What should I study? Accounting or Information Systems?

This is a great time to be alive. Let’s not put too much trust in AI. Programmers will be needed for a long time.

98% of engineers already copy and paste code and probably only have a vague idea of what the code does, AI is making it easier to copy and paste. The real power in AI is helping to explain code,, what I have found really useful is to give it some code that some other engineer wrote which was badly documented or not at all, and ask it to write a readme file for the code or suggest how the code could be improved. You learn a lot by just doing that.

When I went to university the philosophy department contained the clearest, most logical writers. I don’t know if that is still true. It was a long time ago. I don’t think it matters ultimately. Prompt engineering is a brief phase only.

AI will replace me when you can give AI a Jira ticket and the Git repo link and it'll code Everything you need perfectly with tests, documentation etc

In my unprofessional opinion, after spending time with chat it feels like you always end up with something that closely resembles the almost accurate solution. I believe that computers and programmers are so heavily regarded with respect because of the accuracy and speed a computer can get the right solution…. So If programming does loose traction, it’ll be because a lot of people wanted to trade accuracy for “looks good to me”, leading to overwhelming mishaps and a loss of trust by the world in the community itself.

Apparently there have been "glitch words", like ' petertodd' that for some GPT models seemed to be comparable to an "optical illusion".

Just do the math: In the USA alone, almost 400,000 IT workers were laid off, with the last wave in 2023 primarily affecting software developers. By the end of 2023, half a million software developers will have been laid off. Is this not oversaturation? Are you kidding or high on hallucinogenic drugs? Software Dev like profession is dead.

Software Engineers are paid to solve problems NOT to wrote code. Authors are paid to write plots, characters, character arcs etc NOT write words.

My son started on Basic and a Commodore 64 and has continued through the field via IT and then computer scientist / programmer. He stated the same thing mentioned by Dr. Wolfram when I challenged him by saying Chat GPT would take his job. He responded that he does little programming anymore but now collaborates on the course to get to the goal.

I've been in Software for 35 years, my job still involves some some deep coding, I use AI in some complex modelling of Gas mixing, AI has some useful application, Github Copilot etc, but a lot of what I do is building requirements and executing them with the subject matter experts, tbh coding that is bricklaying and maintenance, that is offshored, tools will evolve that make that work more and more automated, if you are in the West, you cant compete with those rates, even before AI. I would certainly not advise anyone to go into programming, but it is not Dead, you want to think of a Career that is AI proof, many jobs will be dead within 10 years, I could list many that are already being made redundant in the last year.

LLMs are very good in ideation and troubleshooting due to all the data included in their datasets. However, they fall short in areas such as decision-making, progress tracking, and planning that require a bigger picture of the project context, evaluation the consequences of past decisions, and taking decisive actions. LLMs would need to be fine tunes on this kind of workflows, to continuously train on the real-time data (the ongoing project), akin to the evolving intuition a human might develop over time.

i'm old enough to remember the introduction of spreadsheets (Visicalc) a lot of the current discussion about AI reminds me of conversations back in the day about whether we could trust spreadsheets it's a long time since i heard anyone ask 'how can you be sure that this thing is calculating stuff correctly?'

As with all click-baity questions the answer lies somewhere in the middle. Programming won't die becuase someone has to peer review the code and be held accountable for multi-million dollar errors. Although my prediction is that AI will displace far more jobs than it creates and those displaced jobs won't magically be replaced by something else like with technology in the past. Further more these displaced jobs will be middle class knowledge worker jobs (in the past it was lower class jobs replaced by tech) which will be the nail in the coffin of the middle class who have suffered already from the housing crisis, inflation, austerity, higher taxation and lower quality of life. Overall I don't see how AI doesn't create a more unequal economy.

In the 90's coders stopped using assembly language and started using high-level languages without knowing what the machine was actually doing in the lowest level. Today AI will simply replace these coders.

Prompt engineer is a DOA job, I don't understand how people think of that as a profession, once we know better how models work internally, we won't need prompt engineers, also natural language is too imprecise for generating code, it may work for a boilerplate, but the actual reasonably precise (we could argue that assembly could be considered for the fine details) description of a computer program is source code, not natural language. It makes more sense to have something like copilot as an AI based completion tool.

Okay, let's look at it this way. Assume AI will subsume every human occupation at some point. How will the encroachment likely unfold? Now that AI has stretched its "legs" in art and programming it's fair to say we (most people) have miss calculated the starting point for this encroachment. It seems obvious now, but AI is seeing significant progress in these areas because these areas exist completely or substantially in the digital world. The digital realm is AI's native environment. Taking this framework it is logical to conclude that the occupations that will likely be human-dominated the longest will involve highly variable physical environments with a low paying wage associated with these occupations. This is because, apart from the obvious variable aspect, the time needed to create humanoid robots at scale that are capable of performing highly variable physical tasks while being cost-effective will take a very long time. So it seems the occupations that will have a substantial amount of time remaining human-dominated, while paying well, will include a highly technical aspect and a highly variable physical environment to operate in. In the end, ironically, the jobs that will last longer than any others will be the ones humanity has looked down upon. There is only one path that I can see that may be able to circumvent AI entirely, that is until we are all slaves to AI lol, which is investing/trading. The real issue is participation. AI becoming competent and cost-effective at your skill set means you can not participate, generate income, maintain independence, etc... So the traditional paradigm for raising/earning capital is SKILL SET = CAPITAL, but if we are outcompeted within our respected skill sets we have another option fortunately. Instead, of using a traditional skill set/occupation to generate capital, we use capital to create capital (think along the lines of real estate/equities/trading, etc... = capital). As long as we are free to allocate our capital the way we see fit we can generate more by utilizing the capital markets by allocating our capital to assets generally.

AI will likely be a better prompt engineer. The question comes down to "who can ask a better question or more logical string of questions?".... AI

Why is everyone happy to say "programming is dead"? And isn't AI enabling the "idea guys"? Why it only happens in computer science? Have you ever seen professionals in mathematics or physics rejoice that "we found some ways to never need to look at number and symbols again, yey!!! I will be a math expert without needing to stare at nerdy things!". This aversion to programming only happens among programming experts. How many articles I've seen "you will not have to code again and better catch on with it because the future is bright".

GPT-4 Summary: "In a conversation between Stephen Wolfram and Lex Fridman, they discuss the changing nature of programming and its potential future. Wolfram suggests that as computational language becomes more accessible and intuitive, users may not need a deep understanding of the coding process. With advancing technology, users can give relatively simple commands, like comparing two images, and the computer generates a complex piece of computational code. Wolfram speculates that people may eventually only look at the results of the code, trust the generated language, or review tests to ensure accuracy, rather than reading the actual computational language code. He also brings up the interesting concept of learning to use computational language before truly understanding it, a process quite different from other disciplines like mathematics. They further discuss the changes in computer science education over the years, with the shift from a theoretical focus to a more practical one, driven by demand for software engineering skills. Wolfram points out that as different fields became computational, the main focus shifted to learning programming language-type programming. However, with advancements in technology, if users don't need to understand the under-the-hood mechanics of telling the computer detailed commands, then what is it that they need to learn? Wolfram suggests that users need to understand the computational possibilities and have a clear picture of what they want to achieve. Fridman and Wolfram also explore the notion of 'prompt engineering,' which involves clear, expository writing to communicate effectively with AI, hinting at a future where this could be a key skill. They conclude by considering the future of computer science departments, given these changing dynamics."

That optical illusion analogy is fascinating.

At guess it all depends at what point of the exponential curve we are. But regardless of that, we will always be able to program for fun.

honestly I don't think it will replace much at the moment. I am a programmer and create apps, for small clear functions it can do stuff (sometimes wrong, so you still need to know how to fix it and what it did wrong). But the entire architecture, how to connect it all, how to add to it in a logical way, how to make it flexible etc, it does not do that at all. I wonder if this is at all possible at some point, maybe, but I think it would still not be possible for someone that does not understand how to code. sure, creating generic apps or something might be possible, but something new, i doubt it.

I was thinking about learning programming as a career but am afraid by the time I finish AI will replace people and all the time has gone to waste

The fear that people will no longer learn programming or computational languages I think is unfounded given that we already have visual interfaces for many tools not using AI, so that's not a new thing, it's just that now it will be much more versatile and useful and general purpose and easy to use so more people will have access to it, its a net positive

two very articulate people talking about giving instructions in natural language - might work ok for them, not so sure about the average communicator

In it's current form, ChatGPT4 will generate a wild and vast variety of methods and code style depending on a few words in the request. You change one word or two and the result is completely different. With the correct words, the result is impressive. I really feel like my programming skills are multiplied! Saving so much time writing the structure of the code, then integration is way simpler. This is cool to see the features working so fast!

People stopped learning how to properly engineer and manage the software development process over a decade ago. Maybe two! The result, is company's who are bogged down and spending countless hours of programming time trying to maintain and extend their existing systems, and also, users having to deal with basically broken and unintuitive software. Hopefully automation will correct these problems in the same way that autonomous driving will greatly reduce auto accidents and traffic jams. Humans have the potential to be great, but money and greed usually ruins that potential across the board.

I feel sorry for the people who are really good at programming. I’m a newbie developer and I suck at it. I don’t think I would ever have the potential to become any better than average and I’m ok with that. I got in to it for the money and the flexibility. So if my job is replaced by AI it won’t be any great loss. But for all the 10x developers out there who live and breed this stuff it would be a bummer if they were to lose their jobs to advancements in technology. However I don’t think this will be the case as these people will adapt and the average Joe devs like myself who have to basically Google everything will likely be replaced.

"Computer Analyst? Jeez, these days even the coputers need analysts." Ben Parker- 2002

What worries me is that I think coding is one of the most fundamental things students need to learn. It literally re formats the brain.

The LLMs are to programmers as automatic checkout lanes in supermarkets are to cashiers. Yes, you can check out through the machine, but most people still prefer to be served by a cashier because (1) it requires less effort and (2) when anything should go wrong it's so much less pain in the ass. For the same reason, good companies will retain their good programmers while shitty companies will generate "something" with LLM (just like they employ mediocre programmers today).

Consider integrating programming into non-technical fields as computation becomes more accessible (00:02). Encourage learning computational language code through experimentation and iteration (00:43). Explore the potential of generating tests alongside code to validate computational outcomes (01:17). Prepare for a shift in education to focus on understanding computational possibilities rather than the mechanics of coding (02:05). Recognize the importance of clear exposition in communicating with AI and teaching prompt engineering (05:36). Anticipate the need for AI Wranglers and psychologists as human-like AI interactions increase (09:12). Investigate mind hacks for language models that may emerge from understanding the science of LLMs (09:55).

I knew zero about coding, picked up chatgpt and brute forces my way through building an app that I required for work. I now have a fully functional app built in python and react that would normally have taken a real software programmer weeks at least. It’s truly game changing

I was an engineering student in 1980. We took one semester of Fortran, key punch cards and waiting in line to get your program run. Thank God for beer.

A machine will never have a soul. We have the divine spark. I don't care how advanced the AI gets; I will continue to try and live a good life.

This is why people often say in the industry that coding is becoming less and less your main job as a software engineer. It is your main part when you start out, but if you really track down your coding time 5 years later, you'll find that it's mostly down by 50% while you will be actually delivering 200% more value.

Me out here procrastinating my JavaScript course

The fact that people make a living off of twitch doing nothing still blows my mind.

I don’t think it’s dead because you will still need understanding to even communicate with AI about whatever it is your trying to do. So a person with experience in a specific tech stack will have a much easier time with AI assistance than someone who doesn’t know anything. However salaries might go down and or demand might go down OR up if companies compete with eachother by hiring more technical ppl 🤷‍♂️

Programming is 90% getting precise requirements from your stakeholders and then transforming them into a form that can then be translated into an algorithm. Then you go and translate that into supportable machine-readable code, which is the remaining 10% (not of time but the impact). And then comes the arduous support and expansion path. Most of the stuff that has an impact on the companies' performance isn't automated yet. And asking LLMs "what precisely should our HR department want" is a path to mediocre, buggy solutions.

I don't think we're ever going to fully replace programming with natural language "translators". Our natural languages are not good at expressing imperative ideas that could be useful in computer software: just recently I had to get into and document an uncommented part of the codebase no one really touched for a long time. I was shocked by how much time it took me to give a poor verbal description to a 10-line function that I understood perfectly. Our natural languages didn't evolve to be used in these context and I believe they will never be a fulfilling alternative. If even specifically designed computer languages didn't fully push out C and assembly - English doesn't stand a chance :)

I code for the cognitive stimulation. I will still code.

As to the car metaphor, those who can work on and fix the stuff under the hood make better drivers who can also get you out of trouble when something goes wrong.

if the search space is all code, it will include the whole spectrum from trash to pearls, including antipatterns, best practice code and outdated language features. so i guess the irony here is to "know" exactly what you want AND being able to formulate this question in a concise manner. but this is half the answer even if you program it yourself.

AI will develop code and AI models to more efficiently solve problems they want solving. Code is not always the best solution, NN is not always the best solution. AI can help identify the best solutions and refactor the solutions and the requirements change.

programmer -> software developers -> software engineers -> software architect

Anyone who says yes probably hasnt coded at an advanced level or worked on a huge system where AI would destroy everything at worst and fill the codebase with cruft at best if you actually let unskilled devs copy/paste code generated from ai.

As a software developer I see ChatGPT as Eliza with more data to draw on. I see Wolfram Alpha as qualitatively different. Programming is not dead, yet.

Hearing Wolfram himself nod to expoz writing is SO exciting. These large language models are giving those who are more linguistically inclined a chance at the playing field traditionally dominated by mathematical minds.

Programming isn't dead, its actually coming alive, because people with ideas but couldn't code can now put their ideas to life with chat gpt.

Coding isn't going anytime soon bro, GPT-4 the current iteration isn't good enough yet. It is just a useful tool. If it can replace a developer it can replace every job. Still very far off buddy.

"is walking dead?" no, you need it to start running

in my experience right now chatGPT is able to give me code if it doesn't require any legacy business knowledge but the code it gives usually has flaws and either I have to manually correct it (if I lose patience) or I have to ask him a few times to correct things. If I ask him for anything that requires any amount of legacy knowledge or business context it fails flat on its face. Worse thing is that regardless if it's correct or not, it's giving me answers with 100% confidence. I'll be honest, after using it for a few months I haven't found it that much better than going to stack overflow or reaching for the source docs, it's useful in the sense that it is one centralized interface with all the info, but often the info is wrong and I have to look it up anyway. I think it's giving people a false sense of confidence that they are able to learn some topic faster, but actually it's often skipping or ignoring crucial details that will bite you down the line, and you don't know because, well you're using it to learn. I see most people hyped by AI are saying these problems will be fixed in the next years, but it's not very clear to me how? From what people like Sam Altman is saying we pretty much reached the limit of expanding the parameters, it's already able to consume all the info on the web to construct a response and the responses really aren't that great. We already have things like Auto-GPT and you can use it to automate things but that also has a lot of flaws and requires a lot of human involvement. So what's this magical potion we're gonna give it that will make it actually be able to replace programmers?

Programming isn`t dead nor it will be really, but the job market for programmers is going to get crushed by AI, because instead of having a 100 programmers to do something, companies will need 1.

coding isn't going away anytime soon buddy.

Why would you have the AI generate the code for you, instead of just having the AI execute the task the code is meant to execute?

I like Wolframs practical approach of expository writing. It generally aligns with my experiences with ChatGPT, but he makes it more clear what one should aim for.

I hope this never happens. I watched what people do with tech. It's terrifying to give programming power to the general public. The internet was a wonderful place until 1992 when mosaic came out. Then the general population had access to the interchange of any information. Look what they have done to what was once reserved for academics and universities. There should be a prerequisite for people to own computer hardware and use compilers. That power in general use reveals the worst of humanity and fosters its growth.

How often do programmers look at Assembly? It’ll be just like that

Isn’t C++ dead already? 😮😅😂 Seeking your opinion on this: 🤓 I just heard from my Instructor that XP programming (Pair Programming) is very much useful today. I questioned her and said “Smart Compilers and AI generating codes” can now be your peer coder to check quality codes, can’t it? Why do we still need a 2nd coder person to do the same thing the 1st coder does?” My instructor said that that it is still a great practice today even said “Mob Programming” is still on. I thought 💭 What the? What is your opinion on this? What is practiced out there? (I am interested to know - a CS Student/Novice)

9:35 he is talking about "adversarial". For example for image classification models you can design seemingly random noise texture that you layer over the input to get any output you want (e.g. Neural net classifies a car as an ostrich)

programming isn't dead but it will die eventually 100% the question is exactly how far we are from that point and with current trends things seem to be rather unpredictable. programming can die next year or in 15 years but it will die.

I want to become AI psychologist😂. Getting to the deep truth of its unconscious mind 😉. From my experience in psychotherapy l think what we employ are not "manipulative" techniques, but knowing a person, responding to him/her in an attuned manner in the context of deep, trustful relationship that is based on mutual respect. Intellect, emotions, and intuition are involved. Also love. Self-knowledge (transference, countertransference). Additionally crafting of language in communication. Like in any deep, fulfilling relationship. The difference is that a therapist is "a tool" for help in the treatment and development of a patient. It is an awfully fascinating, very humbling work.

As a programmer and analyst, I'd say it is far from dead. When it comes to having a high level of control and accomplishing something that requires the ability to have a tool that is not limited by simply having the functionality that has been made available to you, you will find that coding is very important. The more I deal with low code and automated solutions, the more I know that I need to have the flexibility of writing things in code. It's the difference between using powerBI and building a real analytics pipeline. It's the difference between a square space website or using a framework, vs knowing the CSS and JavaScript to build exactly the features you want.

The customized program a human can create, It'll take a lot of time for an AI to replicate the way we communicate and want things to be. It's way too random.

I firmly believe that majority of people who are saying that AI will make all of us jobless in next 2 years are jobless people who simply want for entire mankind to be in their situation.

accountants didn’t disappear when calculators were invented. AI is a calculator. engineers just got their own version of the calculator.

Coding is the hardest things human beings do. That's exactly why so many bad software is selling because people are better at marketing than coding.

Any experienced software engineer know that AI will replace jack sh!t. Study how ChatGPT actually works and you will know