Current iteration gets fixated on identifying logical fallacies and cognitive biases, and the pilot is quite neglectful.

Training data mostly from Private Eye magazine.