[integricho]

I like these AI threads with everyone's measurement of how good a model is starting with: "it feels like..." and that says a lot on how incapable we are to judge and compare these models.

[topspin]

Indeed. We use similar gauges to judge each other, with no better precision.