Source: https://futurism.com/openai-researchers-coding-fail

Via: https://mastodon.social/@nixCraft/114057212908178621

Reply to this note

Please Login to reply.

Discussion

It's just a tool. I've been getting some good use out of their coding model

Yeah, that byline is still 🔥 though.

As a tool it seems roughly on par with an IDE that can do things like refactor code for you. Two key differences that stick out in my mind:

- the IDE will never get it wrong

- the LLM can do more, but it may be incorrect

I'm comfortable with it being incorrect, because I can clearly tell it didn't understand what I meant by its context. Toss out and try again, or solve more of the problem before asking, has been my approach