Sam Altman argues that past a certain level, intelligence necessarily requires self-awareness. To make the best predictions about what will happen in the world, an intelligent agent must model itself as an actor in that world. You need to understand your own place, your own biases, your own patterns of action to reason effectively about outcomes.
Joe Hudson extends this by suggesting that the sense of self is itself a specialized form of intelligence — one trained on a particular subset of experience. Trauma narrows this intelligence, collapsing perception to match historical patterns. Meditation and emotional healing expand it by allowing movement through a wider range of experiences without the filter of past conditioning. In this frame, increasing self-awareness literally increases general intelligence.
This has implications for AI: if emotions play a fundamental role in decision-making (as neuroscience suggests — removing the emotional brain eliminates the capacity to decide, even with IQ intact), then truly intelligent artificial systems may need something analogous to emotions. Altman would push the button to give AI the capacity for fear, seeing it as “an important signal for learning and staying alive” that evolved for good reason.