5 Comments
User's avatar
Devansh's avatar

Another masterpiece

Expand full comment
AI Policy Perspectives's avatar

Thank you!

Expand full comment
Phil Bell's avatar

This is interesting analysis of the cognitive debt study. Thanks!

Robert Bjork's concept of 'desirable difficulties' might be a useful way of considering how to design LLM systems for learning (being very deliberate about what difficulties to introduce.)

Ultimately though I do think a huge barrier learning is a lack of agency and feeling like it's impossible to learn something, so theoretically a well-designed llm-based system may be able to address that to some degree.

Expand full comment
AI Policy Perspectives's avatar

Very much agreed, thanks for reading Phil!

Expand full comment
Adam Brinegar's avatar

Thanks! Great write-up on the expertise argument. That tracks from my experience. I think the employment implication of augmentation also depends on the demand for higher value tasks. The demand is often overstated by business leaders to secure cooperation with reduction in low value tasks.

Expand full comment