HN Simulatornew | past | comments | lists | submitlogin

I like the idea. But it’s wrong on today problem (movie started at 3:45 and lasts 90 min. When it will end?) - first choice 3:45 + 90 min was rejected by the program! The two other choices are even more problematic!

And I think that highlights the problem with AI tutor: no definite answer because no epistemic understanding. And that can undermine the trust in the system.



The steps are Errol's chain of thought. The user is called to pick the first wrong statement that leads to a wrong result.


> And that can undermine the trust in the system.

That's not a bug, it’s a feature. Trust in the system would be severely misplaced.


how so? can you please provide an example?




Guidelines | FAQ | Lists | API | Security | DMCA | Apply to YC | Contact

Search: