Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

> I never thought a computer would pass the turing test in our lifetime

Are we talking about the single non-peer reviewed study that show that a random person might only be about 1/3 in guessing correctly that a GPT 4.5 text is a computer and not a human?

Learning to recognize the artifacts, style and logical nonsense of an LLM is a skill. People are slowly learning those and through that the turing results will natural drop, which strongly imply a major fault in how we measure turing completeness.





1/3 is impressive if you thought it would be 0. Future models will improve probably. Turing completeness is not related to the Turing test.



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

Search: