AI tools make workers less critical and more confident, Microsoft research finds

A study of more than 300 knowledge workers reveals a study on paradox: the greater the confidence people have in AI tools like chatgpt, the less likely they are to think critically about AI output- Even if they find their work easier to accomplish.

The study, published in Chi ’25 by Carnegie Mellon University and Microsoft Research, examines how professionals in a variety of fields use generated AI tools in their daily work. Their findings suggest that while AI can increase productivity, it can also erode basic cognitive abilities.

After analyzing 936 real-world examples of AI tools used in professional environments, the researchers found that “higher confidence in Genai is related to less critical thinking, while higher self-confidence is related to more critical thinking. related.”

This dynamic creates what researchers call a “transformation from material production to critical integration” – workers spend less time creating content from scratch, but more effort is required to validate and refine the outputs generated by AI.

For many professionals, AI tools have become an integral part of their workflow. The study found that workers use AI assistants for everything from writing code and analyzing data to drafting emails and creating presentations. Nearly 97% of participants reported using Chatgpt, while other tools such as Microsoft’s Copilot and Google’s Gemini were also popular.

The study identified three key shifts in how professionals interact with their jobs when using AI tools: information collection becomes information verification, problem-solving transformations into response integration, and evolving task execution into task management.

A participant identified as P147 highlighted the verification challenge: “AI tends to constitute information to agree with any point you want to make, so manual verification takes valuable time.” This experience is common in different professions, workers Reports say they need to carefully cross-reference AI outputs for reliable sources.

Time pressure is an important factor in how professionals use AI tools. Sales Development Representatives in the study noted that encountering quotas often forces them to prioritize speed over verification: “The reason I use AI is because in sales, I have to reach a certain quota every day or the risk of losing my job. Use AI to save time and don’t have much room to think about the results.”

The researchers warn that this pattern may lead to something they call “cognitive atrophy” — and the conventional chances of practicing critical thinking skills are reduced as workers increasingly rely on AI to perform routine tasks. This phenomenon reflects early attention to the use of calculators in mathematics education, but has a wider implication among professional fields.

The study also found that workers with higher self-confidence tend to be more rigorous in AI output, often using these tools to enhance rather than replace their own judgment. These workers report spending more time evaluating and refining the content generated by AI, especially in tasks requiring domain expertise.

Going forward, researchers suggest that organizations need to develop strategies to maintain critical thinking skills in AI-level workplaces. They suggest designing AI tools to actively encourage critical engagement rather than passively accepting output.

The implications of this study go beyond the productivity of individual workplaces to raise broader questions about how AI tools can reshape expertise and decision-making capabilities. As these tools become increasingly complex and widely adopted, maintaining a balance between AI aid and critical thinking in humans can become an increasingly important challenge for organizations and professionals.