6 Feb
2026
6 Feb
'26
11:59 p.m.
Hi David, I could see this being adapted for AI training too — for instance, labeling “impostor” behaviors could generate interesting datasets for anomaly detection models. Have you thought about how different levels of difficulty could simulate various cognitive biases, like confirmation bias or inattentional blindness? https://geometrydashlite2.io