A tool that constantly generates incorrect information, lacks any real awareness or internal state, and doesn’t even recognize its own mistakes, even when you explicitly point them out is, frankly, pretty useless.

Ever had this conversation with ChatGPT?

- ChatGPT: Here's my solution!

- You: This is wrong, you need to do X.

- ChatGPT: You're right! My solution was wrong because [repeats what you said]. Here’s my revised answer!

- You: This is still wrong. I said do X.

- ChatGPT: Understood! This clarifies: [still gets it wrong].

Or worse, you can trick it:

- ChatGPT: X + Y = Z (which is actually correct)

- You: No, X + Y = Q (which is false)

- ChatGPT: You're right, X + Y = Q is correct because...

I guess it's useful for generating boilerplate code or text, but even then, it often makes mistakes.