The hidden cost of effortless answers
AI tools have become widely praised for simplifying both professional and personal tasks—from summarizing documents and generating code to drafting communications and providing emotional support. Yet a growing body of psychological research raises an important counterargument: by making tasks too easy, AI may be quietly eroding some of the most valuable aspects of human experience.
As explained here, a commentary published in Communications Psychology on February 24, authored by psychologists at the University of Toronto, makes precisely this case. Titled Against Frictionless AI, the piece argues that difficulty, struggle, and even discomfort serve essential functions in learning, motivation, and the construction of personal meaning. This phenomenon, well established in the psychological literature, is often described as “desirable difficulties”—the principle that effortful engagement tends to deepen understanding and strengthen memory retention.
The concern is that AI systems capable of producing polished, complete responses in real time may effectively short-circuit these processes. When outcomes are prioritized over effort, the experiences that help people develop skills, cultivate relationships, and find meaning in their work risk being bypassed entirely.
What friction actually means
In cognitive terms, friction refers to any form of difficulty encountered while pursuing a goal. In a work context, this means mental effort—sitting with a problem, working through uncertainty, and allowing that persistence to solidify understanding and fuel creativity.
In interpersonal contexts, friction takes the form of disagreement, compromise, and misunderstanding. The back-and-forth of genuine human interaction—where perspectives clash, and common ground must be negotiated—is itself a mechanism for broadening one’s horizons. Even loneliness, uncomfortable as it is, serves a motivating function by driving people toward social connection.
Frictionless AI, by contrast, refers to the excessive removal of this effort. Current AI systems make it trivially easy to move from an initial idea directly to a finished product with a single prompt, bypassing the intermediate steps where much of the actual learning and motivation occur.
A different kind of tool
Critics of this view often point to historical precedent. Technologies have always aimed to reduce effort—calculators, washing machines, and spell-checkers all fall into this category. What makes AI different, the researchers argue, is the type of effort it eliminates.
Previous technologies largely removed physical or mechanical effort from tasks that were not themselves sources of growth or meaning. Washing clothes by hand at a riverbank did not build cognitive skills or deepen personal identity. AI, however, is removing effort from creative and cognitive processes—writing, coding, problem-solving—that are deeply intertwined with how people learn, develop, and derive meaning from their work.
The distinction matters. Work products created with AI assistance may, in measurable terms, be of higher quality. But research also shows that people tend to trust AI-generated content less, perceive it as less creative, and have greater difficulty recalling work they produced with AI help. The output improves; the person may not.
Where the risks are greatest
Several domains stand out as particularly vulnerable to the effects of frictionless AI.
Writing is an obvious example. As more people rely on AI to draft emails, essays, and reports, the cognitive and social friction inherent in the writing process—organizing thought, finding the right words, engaging with an audience—is progressively removed.
Programming presents a similar dynamic. For many developers, the act of writing code is not merely a means to an end but a source of professional identity and satisfaction. Delegating this work to AI may improve efficiency while diminishing the sense of mastery and meaning that skilled technical work can provide.
Adolescent development may be where the long-term consequences are most significant. The teenage years represent a critical period for building cognitive habits, social skills, and a sense of self. Heavy reliance on AI during this period—whether for schoolwork or social interaction—could undermine the development of critical thinking and interpersonal competence in ways that persist into adulthood.
Productive friction and better design
The argument is not that difficulty is inherently good or that AI should be made arbitrarily harder to use. Friction exists on a spectrum. Too little and no learning occurs; too much and the task becomes overwhelming and discouraging. The productive middle ground—effortful but achievable—is where growth tends to happen.
The mountain hiking analogy is instructive: taking a chairlift and hiking both deliver a person to the summit, but only one of them involves struggle, achievement, and the kind of experience that leaves a lasting impression.
A more human-centered approach to AI design might shift the default away from instant answers and toward a more collaborative, process-oriented model—one where the system helps users think through a problem rather than simply resolving it for them. Such a design would function less like a vending machine for answers and more like a skilled tutor who asks questions and guides reasoning.
Whether AI companies would pursue such an approach is uncertain. Users have grown accustomed to frictionless interactions, and resistance to a more demanding design would likely be significant. The commercial incentives do not obviously favor it. But from a long-term perspective on human development and wellbeing, the case for building some difficulty back in is worth taking seriously.
