Critical Thinking in the Age of Artificial Intelligence: A Survey-Based Study with Machine Learning Insights
New research shows AI's impact on reasoning isn't uniformly negative, but depends heavily on how you use it.
A team of researchers led by M Murshidul Bari and Akif Islam has published a comprehensive study titled 'Critical Thinking in the Age of Artificial Intelligence' on arXiv. The paper, submitted to the International Conference On Power, Electronics, Communications, Computing, and Intelligent Infrastructure 2026, investigates the complex relationship between AI usage and human reasoning through interview-based surveys and logic tasks. The findings reveal a nuanced picture: while participants largely viewed AI as a tool for speed and learning support, many also reported decreased patience for sustained cognitive effort.
Objective performance data showed that reasoning scores varied significantly, with reduced patience and stronger dependence-related tendencies showing a closer association with lower performance than demographic factors alone. Using exploratory machine learning clustering techniques, the researchers identified that AI users are not a homogeneous group but instead form three distinct behavioral profiles: over-reliant users, mixed-strategy users, and balanced support-seekers. This suggests the technology's impact is not uniformly positive or negative but is mediated by how individuals choose to engage with it.
The study's core argument is that effective human-AI collaboration should be designed to support reflection, verification, and sustained cognitive effort rather than simply substituting for these processes. With 5 figures and 2 tables of supporting data, this research provides empirical evidence for what many educators and professionals have observed anecdotally, offering a framework for developing more thoughtful AI integration strategies in both educational and workplace settings.
- Study found reduced patience and AI dependence correlate more strongly with lower reasoning performance than user background characteristics
- Machine learning clustering revealed three distinct user profiles: over-reliant, mixed-strategy, and balanced support-seekers
- Research argues AI systems should be designed to support, not replace, reflection and verification processes
Why It Matters
Provides data-driven insights for designing AI tools and workplace policies that preserve, rather than diminish, essential human reasoning skills.