I asked Chat-GPT to predict future versions of itself. I had to stop at ChatGPT-20. Too scary
1rre @ 1rre @discuss.tchncs.de Posts 3Comments 594Joined 2 yr. ago
1rre @ 1rre @discuss.tchncs.de
Posts
3
Comments
594
Joined
2 yr. ago
The issue is you'd still need humans to verify its correctness... It could come up with all sorts of new discoveries, some of which would be hugely groundbreaking, but a lot of which would be correct-sounding nonsense, just like AI at the present does with normal sentence... The issue with ML is that it doesn't really doubt itself, and anything that could would be a whole new technology, not just a hugely improved version of gpt4