Hacker News new | past | comments | ask | show | jobs | submit login
In Largest-Ever Turing Test, 1.5M Humans Guess Little Better Than Chance (artisana.ai)
4 points by mellosouls on June 10, 2023 | hide | past | favorite | 1 comment



This mostly proves that untrained humans suck at software testing.

It also demonstrates that evaluating people or bots takes time.

That waitress who smiles at me may trick me into thinking she cares about me, only later for me to realize that to her I'm just another rando who might dispense a big tip. Or maybe not. I can't know.

I once had a ten year acquaintance with a person who turned out to be (in the further evidence of 15 more years of experience) an opportunist and sociopath. In a world where it's hard to know people, it's also hard to make snap judgments about bots-- except that we KNOW bots ARE NOT actually human-like in their perceptions, judgments, or private experiences. If they seem so, it's merely camouflage.

Camouflage may be more or less effective, but it's still just a trick. If you tell me "here's a con artist who is so good that you will not catch him defrauding you for at least a year!" that's not an exciting development in interpersonal sophistication. It's alarming and depressing. And as that improves to "you will not catch him for 5 years... 10 years... you will never catch him." that is not improvement.




Consider applying for YC's W25 batch! Applications are open till Nov 12.

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

Search: