Michael Pollan said the existing scientific theories regarding consciousness are inadequate during a talk at the Royal Institution in London.

This critique challenges the prevailing optimism surrounding the ability of science to map human awareness and the potential for artificial intelligence to achieve consciousness. As AI systems become more sophisticated, the distinction between simulated intelligence and actual subjective experience remains a critical point of contention.

Pollan identified 22 competing scientific theories of consciousness [1]. He said none of these frameworks adequately solve what is known as the "hard problem," the question of how and why physical processes in the brain give rise to subjective experience.

During the presentation, Pollan examined the limits of current scientific approaches. He said the existing theories fail to bridge the gap between biological mechanisms and the internal feeling of being alive. This gap persists despite the proliferation of different models attempting to explain the phenomenon.

Pollan also cautioned against over-optimistic claims regarding AI consciousness. He said the ability of a machine to mimic human language or logic does not equate to the presence of a conscious mind. The lack of a working theory for human consciousness makes it nearly impossible to verify if a machine has actually become aware.

The talk highlights a fundamental tension in modern science. While neurobiology can identify which parts of the brain activate during specific tasks, the transition from electrical signals to a conscious thought remains elusive. Pollan said the current scientific trajectory may be unable to resolve this mystery without a paradigm shift.

Pollan identified 22 competing scientific theories of consciousness.

The inability to reconcile 22 different theories of consciousness suggests that the scientific community lacks a consensus on the basic nature of awareness. This theoretical void creates a significant risk in the development of AI, as developers may attribute sentience to systems based on performance rather than actual consciousness, potentially leading to misplaced ethical or legal protections for software.