Bengio's method involves tagging data into 'communication acts' (what people said) and 'verified factual claims.' The AI is trained to explain these, inferring 'latent variables' (hypothesized facts) and assigning probabilities, crucially maintaining the distinction between reported speech and objective truth.
Impact: High. This syntactic distinction is critical for preventing AI from conflating reported speech with reality, a key safety concern with current models that can mimic deception.
In the source video, this keypoint occurs from 00:03:56 to 00:06:31.
Sources in support: Yoshua Bengio (Guest, AI Researcher)

