Really interesting, thanks - but not really surprising, I'm afraid.

The huge difference between AI and humans, is that humans have prejudices and biases, while AIs have not (unless intentionally programmed for that).

That's why most reasoning with humans is useless: when they have some bias (and they usually do), they will blindly defend their positions, even denying the facts put in front of them.

We could say that AI and computers "don't give a damn", they just consider data. And they do not need to feel “right” — while humans utterly do.

- Humans have emotions and reason - but emotions overwhelm reason most of the time.

- OTOH, computers and AI have only reason. Thus only them can really be objective.

This explains why an AI can easily arrive at logical conclusions, where most humans would not. At the same time, an AI can be easily fooled and influenced (they don't - yet - really understand what they are "reasoning" about), so they have a different reliability problem.