3 ms·
In the world of chess the machines are so intelligent, that this can actually be used to distinguish them from humans. That Niemann has probably cheated in his
by amai 4y ago
In the world of chess the machines are so intelligent, that this can actually be used to distinguish them from humans. That Niemann has probably cheated in his chess carrier has been "proved" by comparing his moves with computer engine moves.
We might see a similar metric in the future when trying to prove that a certain text has been AI-generated. So a text will be marked as AI-generated, if it is highly correlated to the output of common LLMs.
But even in chess this metric is far from decisive: https://chess.stackexchange.com/questions/40695/how-often-do-super-gms-achieve-100-engine-correlation https://chess.stackexchange.com/questions/40695/how-often-do...
- danuker 4y agoWow, with such overlap it looks like optimally-played chess has a much smaller state space than what you'd expect given possible moves each turn. I don't think this applies to language or image models. But that would also allow models to foil detection by "playing" far enough from optimal.