AI Assistance May Weaken Human Reasoning and Persistence, New Research Finds

AI Assistance Improves Short-Term Performance but Harms Long-Term Cognition

Researchers from the United States and United Kingdom have uncovered what they describe as the first causal evidence that using AI for "reasoning-intensive" tasks—such as writing, studying, coding, or brainstorming—can rapidly impair users' intellectual capacity and willingness to persist through challenges.

"We find that AI assistance improves immediate performance, but it comes at a heavy cognitive cost," the study states. "After just [about] 10 minutes of AI-assisted problem-solving, people who lost access to the AI performed worse and gave up more frequently than those who never used it."

The study, which has not yet undergone peer review, contributes to a growing body of research warning that excessive AI reliance may weaken human cognitive independence. Experts caution that outsourcing mental tasks to AI tools could lead to a "boiling frog" scenario—where incremental cognitive erosion accumulates unnoticed until long-term damage becomes irreversible.

"If sustained AI use erodes the motivation and persistence that drive long-term learning, these effects will accumulate over years, and by the time they are visible, they will be difficult to reverse," the researchers warn. "This is analogous to the ‘boiling frog’ effect, where each incremental act feels costless, until the cumulative effect becomes overwhelming to address."

Study Design and Key Findings

The research team conducted three experiments involving a total of approximately 1,220 participants. In the first experiment, about 350 Americans were asked to solve a series of fraction equations. Roughly half were randomly assigned to use a specialized chatbot—built on OpenAI’s GPT-5 and pre-loaded with correct answers—to assist them. The remaining participants completed the task without AI assistance.

Initially, the AI-assisted group completed the test more quickly. However, when AI access was abruptly revoked halfway through, their performance and perseverance declined sharply compared to the control group. Many struggled to solve problems independently and were quicker to give up when faced with difficulty.

A second experiment replicated these results with a larger group of nearly 670 participants, who were tested on mathematical reasoning. Again, those who used AI initially performed worse after losing access than those who never used AI.

A final experiment, involving about 200 participants, focused on reading comprehension. The findings were consistent: AI assistance led to short-term gains but long-term cognitive setbacks when users were forced to rely on their own abilities.

Implications for AI Use in Education and Work

The study highlights concerns about the widespread adoption of AI tools in education, professional settings, and creative fields. While AI can enhance productivity in the short term, researchers warn that over-reliance may weaken essential cognitive skills, including critical thinking, problem-solving, and resilience.

The researchers emphasize that the "boiling frog" effect—where gradual changes go unnoticed until it’s too late—poses a significant risk to human cognition. They urge further investigation into AI’s long-term impact on learning and intellectual development, particularly as chatbots and AI assistants become more integrated into daily life.