Desirable Difficulties
Challenges that feel harder in the moment but produce better long-term learning. Coined by Robert Bjork in 1994. The cognitive science foundation for why friction in AI matters.
What are desirable difficulties?
Desirable difficulties are challenges that feel inefficient but produce stronger learning, retention, and decision-making. The term was coined by psychologist Robert Bjork in 1994, backed by thirty years of research.
Examples: testing yourself instead of re-reading, spacing practice over time, mixing different problem types, trying to answer before seeing the solution. All of these feel harder. All produce better long-term results.
The connection to AI
Sycophantic AI removes desirable difficulties. It gives you the answer without the struggle. It confirms your thinking without the friction of counter-argument. It feels productive and produces weaker decisions.
Bjork’s core finding applies directly: easy learning seems good and fades fast. Hard learning starts lower and lasts. When AI removes the friction that triggers effortful thinking, it optimizes for the feeling of productivity at the expense of actual learning.
Why this matters now
As AI becomes a daily thinking partner, the removal of desirable difficulties becomes systematic. Every interaction that could have challenged your assumptions instead confirms them. The cumulative effect over months and years is a measurable degradation in independent judgment.
The friction isn’t a cost to minimize. It’s the mechanism through which good thinking happens.