Ask Onix
AI tools linked to reduced brain activity in learning tasks
Recent research indicates that heavy reliance on artificial intelligence for academic and professional tasks may weaken cognitive engagement and problem-solving abilities. Studies from leading universities highlight both potential benefits and risks of AI integration in education and work.
MIT study reveals brain activity changes
Researchers at the Massachusetts Institute of Technology found that students using ChatGPT to write essays exhibited lower brain activity in regions associated with cognitive processing. The study, which monitored 54 participants via electroencephalography (EEG), showed these students also struggled to recall details from their own work compared to peers who wrote without AI assistance.
Participants used AI for tasks such as summarizing essay questions, sourcing references, and refining grammar. While some reported frustration with AI's ability to generate original ideas, the tools were widely used for structural and stylistic improvements.
Overconfidence in AI linked to diminished critical thinking
A joint study by Carnegie Mellon University and Microsoft surveyed 319 white-collar professionals who used AI tools like Copilot weekly. The research analyzed 900 task examples, including data analysis and rule compliance checks, and found that higher trust in AI's capabilities correlated with reduced critical engagement.
"While generative AI can boost efficiency, it may lead to long-term overreliance and erode independent problem-solving skills," the researchers noted.
UK students report mixed impacts on learning
An Oxford University Press survey of UK schoolchildren revealed that 60% felt AI had negatively affected their academic skills. However, 90% also credited AI with helping develop at least one skill, such as problem-solving, creativity, or revision techniques.
Dr. Alexandra Tomescu, a generative AI specialist at OUP, described the findings as a "nuanced picture." She emphasized the need for clearer guidance on responsible AI use in education, noting that many students seek structured support.
Experts call for caution and further research
Professor Wayne Holmes of University College London warned that the lack of large-scale, independent research on AI's educational impact raises concerns. He cited studies on cognitive atrophy-where skills deteriorate after prolonged AI use-as evidence of potential risks.
Holmes pointed to a Harvard Medical School study showing AI-assisted radiologists experienced mixed outcomes, with some clinicians' performance improving while others declined. The authors called for deeper investigation into human-AI interaction to maximize benefits and minimize harm.
"Students may produce better outputs with AI, but their actual learning could suffer," Holmes said.
AI developers advocate for targeted use
Jayna Devani, who leads international education initiatives at OpenAI, acknowledged the debate but argued that AI should function as a tutor rather than a shortcut. She highlighted ChatGPT's "study mode," which allows students to engage in interactive problem-solving rather than receiving direct answers.
Devani provided an example of a student working late on a difficult assignment, noting that AI could break down complex questions into manageable parts. "The potential is there to accelerate learning when used intentionally," she said.
Holmes countered that students must understand AI's limitations, including its reasoning processes and data handling practices. He stressed the importance of verifying AI outputs and viewing the technology as more than a simple tool.
"AI is not just the latest iteration of the calculator. Its implications are far-reaching," Holmes said.
Balancing innovation and skill development
As AI adoption grows, experts urge a balanced approach that leverages its benefits while mitigating risks to cognitive development. OpenAI has released 100 educational prompts to help students use ChatGPT effectively, but critics argue more research and oversight are needed before widespread integration into learning environments.
