IEEE Spectrum ⭐⭐⭐⭐⭐

Why Frictionless AI Might Be Harmful

By Emily Zohar, Paul Bloom, Michael Inzlicht | Published February 2026

Summary

Psychologists from the University of Toronto argue that AI systems that make tasks too easy may have unexpected costs. Drawing on research in cognitive psychology, they suggest that friction—difficulty, struggle, and even discomfort—plays an important role in learning, motivation, and meaning.

Key Insights

🎯 Core Argument

AI removes "friction" from cognitive and social tasks, taking away the intermediate steps that drive motivation and learning. It prioritizes outcomes over process.

What is "Frictionless" AI?

Frictionless AI refers to the excessive removal of effort from cognitive and social tasks. With typical AI use, it's easy to go from ideation right to the end product—you ask AI to solve something with one prompt, and it completes the whole thing.

The Problem with Removing Friction

  • Learning: Research on "desirable difficulties" shows that effortful engagement deepens understanding and strengthens memory
  • Relationships: Disagreement, compromise, and misunderstanding help broaden perspectives
  • Motivation: Struggle leads to achievement; too little friction means no growth
"If you're faced with the same problem and AI is removed, you don't have the required knowledge to know how to face the problem next time."

Real-World Examples

  • Writing: People increasingly rely on AI to draft emails and essays, removing beneficial friction
  • Vibe coding: Programming is integral to what drives meaning for developers
  • Adolescents: A critical developmental period where effortful interactions teach critical thinking

Historical Context

Past technologies (calculators, washing machines) mostly focused on reducing physical effort—removing mundane tasks that weren't driving learning and growth.

But AI is different: it's taking away effort from creative and cognitive processes that drive meaning, motivation, and learning.

💡 The Chairlift Analogy

Taking a chairlift vs. hiking up a mountain: both get you to the top, but the hiker gets growth benefits, a sense of achievement, and a learning opportunity.

What Can Be Done?

The authors suggest thinking about AI design differently—instead of jumping to the answer, AI could be more collaborative, helping users think through problems and teaching along the way.

Original Article

Read full article on IEEE Spectrum →