I see a lier, eyes are very wide, head moves from left to right (no) while speaking "truth." Terefore, the AI was trained on liers and bullshiters. There will always be a human bias built into these models.