Why Frictionless AI Might Be Harmful
Why Frictionless AI Might Be Harmful
https://spectrum.ieee.org/frictionless-ai-psychology
Publish Date: 2026-03-22 09:00:04
Source Domain: spectrum.ieee.org
In a rapidly evolving tech landscape, many praise AI tools for making life easier by streamlining both professional and personal tasks. However, psychologists from the University of Toronto caution against frictionless AI, which removes all the struggles and difficulties inherent in cognitive and social tasks. In their commentary, published in Communications Psychology, they argue that “friction”—defined as effort, persistence, and even discomfort—is essential for learning, motivation, and gaining meaningful experiences. The researchers warn that AI’s ability to produce polished outcomes almost immediately strips away the intermediate steps that foster deep understanding. This could lead to over-reliance on AI, which may inhibit vital skills such as problem-solving, empathy, and critical thinking, especially for adolescents who are still developing.
The authors highlight that past technologies have alleviated mundane, non-developmental tasks, but AI eliminates effortful engagement in processes that fundamentally shape character and capabilities. For instance, outsourcing writing to AI diminishes both cognitive and social friction and can decrease the learner’s trust and memory of generated products. The psychologists suggest a more human-centered design where AI incorporates elements that encourage reflection and partial learner involvement, fostering deeper engagement instead of just delivering instant solutions. They argue that such friction-full AI could offer long-term cognitive and social benefits, although it may face resistance from users accustomed to the convenience of current models.
Key Points:
– AI’s frictionless nature could erode essential cognitive and social learning processes.
– “Friction” involves difficulty, struggle, and discomfort that are crucial for deep learning and developing empathy.
– Current AI practices remove effort from meaningful, skill-building tasks.
– Incorporating moderate friction in AI could foster more engagement and long-term cognitive benefits.
– Resistance to such design changes might come from users and companies due to convenience and preference for instantaneous results.
