Reading

McGilchrist's argument isn't about "left brain vs. right brain" the way pop psychology tells it. It's about attention — the fundamentally different ways each hemisphere attends to the same world. The right hemisphere sees context, relationship, the living whole. The left isolates, categorizes, makes things manipulable. Both necessary. The problem is that Western culture has systematically privileged the left hemisphere's mode: the map over the territory, the representation over the thing itself.

This book changed how I think about AI design. Most AI tools are built in the left hemisphere's image — reduce, categorize, produce outputs. The question I keep coming back to: can you build AI that operates in the right hemisphere's mode? That sees context and relationship, not just data and categories?

Dense book. Took me months. Worth every page.

Menary takes Clark and Chalmers' Extended Mind thesis and asks a harder question. Clark and Chalmers asked: does the external resource function equivalently to an internal cognitive process? Menary asks: have the person and the tool developed integrated practices for thinking together?

The shift matters. Equivalence is a low bar — it just asks whether the tool does the same job. Integration asks whether using the tool has actually changed how the person thinks. This is the difference between a sales rep who uses AI to generate discovery questions (substitution) and a rep who has developed a practice of thinking with AI that has genuinely changed how they approach buyer conversations (integration).

Bloom's finding from 1984 is still the most important number in education: students who received one-on-one tutoring performed two standard deviations above students in conventional classrooms. Two sigma. The average tutored student outperformed 98% of classroom students.

The "problem" in the title is that one-on-one tutoring doesn't scale. You can't give every student a personal tutor. Forty years later, that's still true. But AI is the first technology with a realistic shot at it — not by replacing the tutor, but by replicating the two things that make tutoring work: deep knowledge of the individual learner, and real-time adaptation to how they're thinking.