Essay argues AI tools exploit human cognitive shortcuts rather than replicate thought

An essay in The Verge critiques the dominant metaphor of the brain as a computer, arguing that it has distorted both AI development and public understanding of cognition. The piece, written by Benjamin Riley, traces the idea to Alan Turing and notes that figures like Google DeepMind CEO Demis Hassabis and Elon Musk have explicitly framed the brain as a biological computer or Turing machine. Riley contends this computational model reduces thought to a three-step process—input, computation, output—ignoring the dynamic, embodied nature of human cognition.

Riley contrasts this with a feedback-control model of the brain, drawing on the work of neuroscientist Paul Cisek at the University of Montreal. Cisek’s research, which spans evolutionary timescales, characterizes the brain not as an information processor but as a system that continuously extends control over the environment by adjusting actions to regulate sensory input. For example, catching a fly ball in baseball is better explained by a simple heuristic—keeping the ball in a fixed position in the visual field—than by complex subconscious calculations of velocity and gravity.

The essay argues that AI systems, designed under the computational metaphor, exploit evolved cognitive shortcuts rather than replicate or augment thought. Riley compares AI to 'junk food for the mind': easy to consume in the moment but ultimately unsustainable for healthy cognition. This framing suggests that current AI tools may not enhance reasoning so much as they cater to superficial engagement, potentially undermining deeper cognitive habits.

Riley also links this critique to the history of neuroscience, noting that figures like John von Neumann doubted the computational model could capture the complexity of the nervous system. Instead, the essay advocates for a model that unfolds over evolutionary time, with new behaviors emerging in response to environmental possibilities. This perspective emphasizes the brain’s role in actively shaping perception and action, rather than passively processing information.

The piece concludes by suggesting that the AI industry’s reliance on the brain-as-computer metaphor has contributed to cultural confusion about what AI can and cannot do. By prioritizing computational metaphors over biological and evolutionary realities, the industry risks building tools that align with human weaknesses rather than strengths.