3 ms·
It's still too early to say what will the outlook be like in the next ten years, wrt to AI, but one thing I am perfectly sure of is: AI will always produce 'err
by jan_m_savage 1mo ago
It's still too early to say what will the outlook be like in the next ten years, wrt to AI, but one thing I am perfectly sure of is: AI will always produce 'errors' due to the fundamental fact that it's heuristics-based.
Maybe fewer and fewer, but the mistakes will never be totally eliminated. This is why those wanting to learn how to write software shouldn't get discouraged that 'Oh, AI will replace human programmers'.
- frotaur 1mo agoI fail to see why this is a good argument against AI replacing humans, given that humans also produce errors.
- jan_m_savage 1mo agoIf a human makes errors and discovers it, the error gets fixed. With AI, it usually fails to fix the error totally on first try, even on nth tries. If on the off chance it does fix it, it generates other problems.
- bigstrat2003 1mo agoHumans do produce errors. But they produce far fewer than LLMs, and they can also learn. Once a person makes a given mistake and realizes what he did wrong, he will take steps to prevent that mistake in the future. LLMs do not do this and therefore are less trustworthy than humans, even if both are prone to error.
- exitb 1mo agoThere's a class of programming/configuration tasks where I specifically use AI because it makes less errors than I do.
- jan_m_savage 1mo agoYes but it will make other kinds of errors you wouldn't make.
- CrimsonRain 1mo agoSo humans never make errors? The very same thing that allows humans to make errors, find errors, learn from it, fix it...AI can/will do too. Then what? You lack imagination.
- jan_m_savage 1mo agoYou lack manners. That's even worse.
- someguynamedq 1mo agoTool use.