4 ms·
The paper explains that they used a panel of evaluators to judge the "humanness" of the interactions : )
by mztwo 3y ago
The paper explains that they used a panel of evaluators to judge the "humanness" of the interactions : )
- explaininjs 3y agoSaid panel of evaluators found that AI agents pretending to be humans had more "believable" responses than humans pretending to be AI agents pretending to be humans. So that's... a result.