This paper poses a puzzle about what we think makes us human.
Before I give the answer, try it: You & an AI that looks like a person are in front of a human judge. You can each say only one word. The judge then kills whoever they think is the AI
What do you say? (Don’t peek)
Before I give the answer, try it: You & an AI that looks like a person are in front of a human judge. You can each say only one word. The judge then kills whoever they think is the AI
What do you say? (Don’t peek)
The most common answer was “Love" but that really didn’t help the judge. The best answer was 💩
If someone said 💩 and the other said "love," judges would assume that whoever said 💩 was the human 69% of the time, and kill whoever said "love." "Banana" is also a good choice.
If someone said 💩 and the other said "love," judges would assume that whoever said 💩 was the human 69% of the time, and kill whoever said "love." "Banana" is also a good choice.
The graphic shows all the words given by at least one person, clustered by semantic similarity (yes, that means at least two people chose “moist” and two chose “bootylicious”). Here’s the paper: cocodev.fas.harvard.edu/publications/a…
The most human strategy.
https://twitter.com/kristirjensen/status/1434863147861614593
• • •
Missing some Tweet in this thread? You can try to
force a refresh
