Stability AI CEO: There Will Be No (Human) Programmers in Five Years | Hacker News Reader

Stability AI CEO: There Will Be No (Human) Programmers in Five Years

decrypt.co

13 pointsfoobazgt16 comments

At some point the people making absurd claims like this need to be called out. So many people are just scared of looking foolish in the other direction (discounting a world changing technology at an inflection point).

I think it's more about fundraising. "Omg guys, this tech is so amazing and terrifying, it's a total paradigm-changer! ...Better get in on the ground floor $$$."

There are people whose entire careers got reset due to changes in laws. Those individuals I’m sure are absolutely trying to send the same fear into other industries. Its definitely worth calling them out on though. Tired of having people try to poke holes in my career.

I can imagine far in the future we will have multi purpose AI models generating models that are narrowed to a specific task, for example a model trained on the entire corpus of text available to humanity (legal cases/libraries/blogs/debate ToS/the trash on the internet/etc.) to generate a model that is faster and more accurate to do inference work for legal cases.

Replacing all human programmers in 5 years is a bit too far fetched even for a CEO who is trying to do fundraising.

For programming, I can imagine in 5 years we will have AI trained on massive codebases that enterprises have (+10m lines of code) and the machine would detect duplicates/crossing boundaries, etc. but to replace those who develop creative solution to problems in mere 5 years... :D

Human Programmer (Me): There will be no Stability AI in 2 years.

Those kind of BS remarks simply undermine the credibility of the person uttering it. Or it is deliberate BS aimed at convincing no clue VC’s to throw money at a startup with no real business model.

>In March 2023, he signed an open letter calling for "all AI labs to immediately pause for at least 6 months the training of AI systems more powerful than GPT-4".[11]

Deliberate attention-seeking bs. I would be a significant amount of money he was a crypto bro before this latest hype train.

Edit: Yupp, crypto bro turned ai evangelist... https://aithority.com/technology/cryptocurrency/ananas-found...

I'll take the other side of that bet. Between the technological limitations (at least so far) and the legal challenges, I'm pretty sure the odds are against the CEO's prediction.

> AI CEO bullish on AI.

This is not the absolute hill one should die on.

As long as you have to check the code that is being generated by this AI, it means you don't trust it. Thus, there will always be human programmers. But I would also say there would be less programmers but not none.

Here's one prediction that will likely become true:

Stability AI will lose the Getty Images lawsuit in under 5 years and will reach a settlement with Getty in licensing the images used to train Stable Diffusion.

JS/TS/node madness is too complex even for AI ;)

https://twitter.com/mitchellh/status/1674498970842972160

Preach! And why I’ll have a job for a very long time ;)

Data from GitHub reveals that "41% of all code right now is AI generated," Mostaque remarked.

Yup, you heard him - "all code" in the whole dang universe, he said.

Apparently this canard derives from a recent quote from Microsoft's Scott Gutthrie [1] which includes the breathtaking mention of "40 percent of code checked in by Github Copilot users."

Shine on, Mostaque. The more you pump this bubble, the faster and harder it will come crashing down, and the better off we'll all be.

[1] https://www.microsoft.com/en-us/Investor/events/FY-2023/Morg... / https://news.ycombinator.com/item?id=35161866

90% of for loops were generated by Ai? have they gone too far?

No human CEOs is more likely.

Lol what an idiot

Reply on news.ycombinator.com