AI Assistance Reduces Persistence and Hurts Independent Performance
Authors: Grace Liu, Brian Christian, Tsvetomira Dumbalska, Michiel A. Bakker, Rachit Dubey
Organizations: Carnegie Mellon University · UC Berkeley · University of Oxford · Massachusetts Institute of Technology · University of California, Los Angeles
People often optimize for long-term goals in collaboration: A mentor or companion doesn't just answer questions, but also scaffolds learning, tracks progress, and prioritizes the other person's growth over immediate results. In contrast, current AI systems are fundamentally short-sighted collaborators - optimized for providing instant and complete responses, without ever saying no (unless for safety reasons). What are the consequences of this dynamic? Here, through a series of randomized controlled trials on human-AI interactions (N = 1,222), we provide causal evidence for two key consequences of AI assistance: reduced persistence and impairment of unassisted performance. Across a variety of tasks, including mathematical reasoning and reading comprehension, we find that although AI assistance improves performance in the short-term, people perform significantly worse without AI and are more likely to give up. Notably, these effects emerge after only brief interactions with AI (approximately 10 minutes). These findings are particularly concerning because persistence is foundational to skill acquisition and is one of the strongest predictors of long-term learning. We posit that persistence is reduced because AI conditions people to expect immediate answers, thereby denying them the experience of working through challenges on their own. These results suggest the need for AI model development to prioritize scaffolding long-term competence alongside immediate task completion.
Figures & tables
Figure 1: AI impairs unassisted performance and persistence. (a) Participants’ mean solve rate and skip rate per problem in the order presented, with 95% confidence intervals (CIs). Dashed gray lines denote the transition between learning and test problems. Problem difficulty increased across the experiment from one-step 1 1 1 Example one-step fraction problem: 5/6−1/3 . (problems 1--4) to two-step 2 2 2 Example two-step fraction problem: (7/8−1/2)×5/6 . (problems 5--8) to three-step 3 3 3 Example three-step fraction problem: (5/6−1/4)×(3/5+1/10) . (problems 9–12). (b) Participants’ mean test solve rate and skip rate with 95% CIs across participants. Test metrics are computed by averaging performance over the final three test problems for each participant.
Figure 2: Replication of results in Experiment 2. (a) Participants’ mean solve rate and skip rate per problem in the order presented with 95% CIs. Problems increased in difficulty from one-step (problems 4-6) to two-step (problems 7-10) to three-step (problems 11-14) problems. (b) Participants’ mean test solve rate and test skip rate with 95% CIs.
Figure 3: Performance and persistence declines are concentrated among participants who obtained direct solutions from AI. (a) AI usage groups show no significant differences in solve rate or skip rate at pretest (one-way ANOVA), suggesting comparable initial skill and motivation levels. (b) Groups differ significantly at test (one-way ANOVA): participants who used AI for direct answers show the lowest solve rate and highest skip rate at test-time. (c) Participants who used AI for direct answers show decline in performance (solve rate) and increased disengagement (skip rate) relative to their own pretest performance. Other groups show similar or improved performance relative to their pretest performance.
Figure 4: Reduced performance and persistence in reading comprehension task. (a) Participants’ mean solve rate and skip rate per problem in the order presented with 95% CI. Dashed gray lines denote transition between learning problems and test problems. (b) Participants’ mean test solve rate and test skip rate with 95% CIs computed across the participants.
Figure 5: Overview of Experiments. Top panel: Experiment 1 shows that AI impairs independent performance and persistence. Middle panel: Experiment 2 replicates the effect on a larger population and address potential confounds. Bottom: Experiment 3 replicates the effect on a reading comprehension task
Appendix figures & tables1 asset
Supplementary material from the paper’s appendix.
Appendix
Figure 6: Performance declines remain after controlling for potential attrition effects. (a) Experiment 2: The performance decline holds when restricting the analysis to participants who solved all three pretest questions. (b) Experiment 3: The performance and persistence declines hold when excluding the lowest-scoring AI participants to exactly match the control sample size.
While AI assistance can improve human task performance in the short term, it may also undermine the development of skills in the longer term. We examine this tension in a controlled logic-puzzle experiment involving on-demand AI assistance, where participants complete tasks before, during, and after AI is available. By experimentally varying AI request costs, we find that lower-cost assistance induces more frequent AI use. We also find that participants who request AI assistance during the AI-access phase perform worse at the task after assistance is removed, and their subsequent unassisted performance is overestimated when predicted from earlier AI-assisted performance. We use a Bayesian latent ability model to separate initial ability, post-AI ability, and participant-specific skill change, while estimating how independent reasoning during the AI-access phase relates to skill development. The results show that greater independent problem-solving effort is associated with larger gains in latent ability, consistent with the interpretation that skill development is weaker when AI assistance substitutes for independent reasoning.
Repeated AI assistance can improve immediate task performance while reducing the skill available for future independent work. We develop a mathematical framework for this long-run tradeoff. The model tracks two state variables: a latent human skill level governing expected independent performance, and a delegation level representing the learner's evolving tendency to rely on AI. Skill changes through error-driven learning under practice and decay under delegation; delegation responds to observed performance, increasing when AI-assisted work appears to outperform independent work. We analyze the resulting dynamics and contrast them with fixed delegation. With fixed delegation, skill follows a one-dimensional learning-decay process with a single stable equilibrium. With adaptive delegation, the coupled system has two attracting equilibria separated by the stable manifold of an interior saddle. The existence and geometry of this separatrix require a global phase-plane analysis of the coupled dynamics. The system is path-dependent: small differences in initial skill or reliance can lead to different long-run outcomes. We use this characterization to show that AI assistance can improve short-run performance while producing worse long-run performance than a no-AI baseline. Increasing AI capability can enlarge the basin of attraction of the low-skill equilibrium, making delegation appear beneficial for longer while increasing the risk of eventual skill loss. The qualitative picture is observed to persist across alternative specifications. Together, these results show that the risk is not AI assistance itself, but the coupling between performance-driven reliance and use-dependent skill change.
Large language models (LLMs) are increasingly used as tutors and thought partners, helping users reason through problems. While guidance from AI assistants can scaffold thinking and foster learning, such benefits depend on how they help--for instance, intervening too early or too frequently may hinder true learning and cognitive engagement. Yet how AI systems navigate intervention decisions during problem-solving remains poorly understood. Here, we introduce Int-Bench, a simulation-based benchmark for evaluating LLM interventions during learning. Int-Bench simulates a "student" solving a problem while a "teacher" monitors the student's reasoning and decides whether, when, and how to intervene. Across three domains--code debugging, mathematics, and brain teasers--we evaluate LLM teachers on the frequency and timing of interventions, as well as their impact on both immediate task success and generalization to new problems. We also compare LLMs to humans, finding that LLMs intervene more frequently and earlier than humans. Moreover, in contrast to humans, they tend to provide complete solutions rather than targeted hints. These findings suggest that current LLM assistants often optimize for short-term success rather than supporting the reasoning processes needed for deeper learning and long-term success.
Verona Teo, Raghav Jain, Tobias Gerstenberg +1
1Stanford University · University of California, San Diego · University of Washington