I think the OP's point is that chat GPT-3.5 may have a chess-engine baked-in to its (closed and unavailable) code for PR purposes. So it "realizes" that "hey, I'm playing a game of chess" and then, rather than doing whatever it normally does, it just acts as a front-end for a quite good chess-engine.
I think the OP's point is that chat GPT-3.5 may have a chess-engine baked-in to its (closed and unavailable) code for PR purposes. So it "realizes" that "hey, I'm playing a game of chess" and then, rather than doing whatever it normally does, it just acts as a front-end for a quite good chess-engine.
I see – my initial interpretation of OP’s “special case” was “Theory 2: GPT-3.5-instruct was trained on more chess games.”
But I guess it’s also a possibility that they had a real chess engine hiding in there.