Thread with 8 posts
jump to expanded post
it has been 0 days since someone made an unhelpful comment on the touchHLE GitHub that's just a verbatim ChatGPT response
I'm going to hold a grudge against OpenAI et al for a long time. To steal a phrase, these tools are raising the noise floor of our perceptual environment… they're certainly doing it for me. And what am I getting out of it? Nothing! Absolutely no good is coming of it for me!
For touchHLE in particular, ChatGPT, GitHub Copilot and many other machine-learning-based systems are at best unhelpful, wasting my time with coherent-looking nonsense, and at worst sabotage the project by sneakily violating copyright. They actively make my life worse.
if you want to learn programming, or really any skill: I wouldn't recommend these tools because they'll definitely give you bad advice without you being able to tell it's bad, and they can't explain themselves if they say something you don't understand, even if it looks like it.
that said, if you insist on using them, please don't pass off their work as your own. especially on my projects. thanks
@hikari i tried putting a haskell question into it and it just. gave me the wrong function so confidently. then i gave it the type error. another confident but wrong answer, apologetic. give it another type error. another wrong answer.
@hikari , there's some good coming from generative AI, at least in my view. People have done some intriguing things with gen AI -- e.g., I create the images accompanying my in-joke post series "Great Moments in Software Engineering" #gmise using MidJourney. You can make a pretty good case that gen AI has fueled a lot of creativity.
And, all that being said, I wish we weren't rushing headlong into ubiquitous use of an extremely dangerous technology that we don't fully understand.