Replit CEO Amjad Masad says learning to code is a waste of time, citing Dario Amodei's prediction that AI may generate essentially all code by next year.

Someone has pointed out it would be lot easier for AI to replace a CEO then a developer.

Who is this guy? Some CEO ? Isn’t it more cost effective to replace him with ai?
- What’s weird is he’s the ceo of replit.
  Replit’s product is a website where you can write a snippet of code and run it without having to install anything. An activity that human developers would do to test out something.
  So if his prediction comes true, his product will lose all value.
  
  Maybe his expectation is that companies will buy his product because people will have to feed the AI-generated code into it to test it, instead of having humans manually review everything. Basically telling people to create a problem so he can sell his solution.
- There is no AI products to replace CEOs, currently?
  
  Literally any chatbot, probably
  
  (You don’t need ai for that)
  
  Even the simpler AI models can simulate 'management'.
  What we don't have is AI's that simulate your everyday asshole bosses.
  What we don't have either is AI's that simulate fair, decent, smart, and empathic bosses.
  
  Until AI can sexually assault their workers, can it really be said they're qualified for the position?

Coding is totally obselete, bro. AI can totally write all the code, trust me bro. You just gotta know how to tell it what code to write, like learn some keywords and stuff, bro. Like, as long as you check how it produces looping mechanisms and tell it when it should use polymorphism and stuff, it'll totally do all the work bro. You don't need to know how to code, just the right sequence of keywords and commands so the AI can write all the code.

Like AI generated stories or art you're going to need someone to edit them.
- Writing code is easier than understanding and reviewing another's code. There is good reason code reviewers aren't the interns and new hires.
  My question to others is, why would you want to turn into a code reviewer for AI code? It's a shitload harder. And if the goal is anything but a weekend project, you damn well better be understanding and reviewing it critically, otherwise one is shitting up the code base and forcing others to clean up your mess.
  
  Because shit for brains says we just need to train our AI models until they're sentient. So that means we need all our workers reviewing the output of AI as much as possible until that happens.

As a coder, the majority of my job isn't writing code. It's translating the bullshit management says and the broken specs we're given into what they both actually want, not what they said. There is never going to be an AI that fixes that
- There is never going to be an AI that fixes that
  Don't be so negative. Of course AI, if sufficiently trusted, could fix the existence of the human race and by extension the existence of bullshit management.

The idea of LLMs putting coders out of work at a large scale seems inherently self-defeating.
The LLMs needed to ingest a massive volume of code to get to their current level of proficiency. What will happen if they put all the coders out of work and Stack Overflow is down to just a small number of hobbyists? Will the LLMs just stop advancing?
I'm sure Sam Altman would say they are just about to have reasoning capabilities that will allow them to improve. But Sam Altman is not credible.
- It's sadly already happening in regards to stack.
  
  I can explain this chart: SO and AI both give me questionably useful example code, but AI isn't as much of an asshole about it as the average SO user.
  
  Outdated and no source..

ok then.
all developers should quit their jobs now. like right now. also, don't bother with foss anymore. AI can generate it all by itself. all foss projects should close all their public repos entirely. no more public code repos, no more software development.
everybody go home, AI has it covered. we can just get new more fulfilling jobs in farming, logging, or construction.
bye greedy AI fuck-sticks. We'll be back when you assholes snap your AI neck trying to suck your own dicks.

We will find out just how good AI is with coding in a few months when it replaces the SSA's, mostly, COBOL based system.
My guess? Catastrophic.
- And instead of going back, they'll say some kid deleted the old stuff after copying it into the new system.
  
  Of course they will delete the old system entirely. Saving it would just be wasteful, and intelligent.

Waiting for AI to take over CEO positions because they do nothing and you can replace them with a series of shell scripts.
- That's a more viable solution than replacing most software engineers, honestly.

I installed Manjaro on one of my computers and I wanted to see, if ChatGPT, Claude, Gemini or Mistral where useful in eliminating some quirks you always encounter after a fresh install. So I asked them whenever I stumbled upon an issue how I could solve that issue.
None, I’d like to emphasise this: NONE of their tips was helpful.
And Mr Masad wants me to believe AIs would be able to program whole applications within a year?
MUAHAHAHA

Lol, yeah. Keep paying us developers to write that philosopher stone. For writing general AI my rate is 100x because it's magic you can't understand without being able to write code.

AI won't code everything by next year, blnot inn5 years either as it requires understanding context and actual reasoning which AI doesn't have and won't have for a long time to come, but the day that AI can code itself is the day that humanity is done for

As a developer, I literally laughed hard enough to choke a little.
- I actually dare them to try. I'm really looking forward to the massive paychecks I'm going to get when companies are panicking to try to untangle all the absolute nonsense bullshit these AI companies are about to unleash into corporate codebases. The AI-slop bugfest will make the Y2K issue seem trivial. I'm so excited, the future looks very bright for human software developers.
  My advice: Practice going over other people's code with a fine-tooth comb looking for bad architecture, flaws and inefficiencies. You won't always be right, you won't find them all, but you'll learn lots of skills you'll need in the future. Whatever you do, don't undersell yourselves, remember that your experience is valuable, and AI has no experience, it just has a huge library it can shotgun "solutions" out of. Half the time they don't even compile, nevermind work properly, or efficiently.
  
  My advice: Practice going over other people's code with a fine-tooth comb looking for bad architecture, flaws and inefficiencies.
  I agree. Funny story, I wasn't allowed to do code reviews at my current job for about 2 years because they thought my comb was too fine. Suddenly software quality is something they are really valuing and they're allowing me to do code reviews again. Funny, that.

Listening to this guy talk shit is a waste of time

LOL, my friend no. Hahaha 🤡

The main issue I’ve encountered is with troubleshooting. Initially, working with cursor was smooth when dealing with a single file and script project. However, as I tried to extend it to handle dependencies like a typical project, the code generation began to spiral out of control, resembling a cancerous growth that keeps producing more and more code. This problem intensified when I started interacting with multiple libraries, making the situation even more chaotic. It must be extra directed to stay on track and even if it tends to always create extra.
But what is interesting with those company CEO is that they still have developers. How come? If AI will replace them you don’t need any. Actions vs words

AI is probably going to transform how code is written, but I don’t think AI will fully replace programmers. At least not in the foreseeable future.
Most of a programmer’s work is maintaining existing code. This is something current AI models still struggle with.

Masad's comments have come up before and sparked huge outrage before and just like before people are missing the hugely important context here.
He added that coding may become obsolete, but people will still need to continue to work on their fundamentals: “I’m at this point, like agents pilled. I’m very bullish. Like, I sort of changed my answer even like a year ago. I would say kind of learn a bit of coding. I would say learn how to think, learn how to break down problems, right? Learn how to communicate clearly, with as you would with humans, but also with machines.”
The way I see it, he's thinking that the current-day approach to coding is likely to go the same way that coding in assembly language went when high-level languages and compilers became good and common. The vast majority of programmers never need to think about individual registers or the specific sequence of opcodes needed to perform operations or access memory, the compilers handle that and they do a great job. Only a handful of specialists really need to go down to the metal like that any more.
So too will it be for a lot of the programming that current day programmers do. It'll still be useful to know how it works so that you'll know what to ask for and what to do when something goes wrong, but 99% of the code will be done by AIs and will hardly even be looked at by a human. There'll still be people who are experts at working with programs but the current approach to how that's done is likely to be obsolete.