Generative AI’s power sparks fears of dumbing humans down
• Studies suggest memory, decision-making, critical thinking are most at risk • Experts say artificial intelligence removes ‘learning opportunities’ • Studies suggest AI boosts short-term gains, but weakens long-term learning • Long-term impact on human brain...
PARIS — Generative artificial intelligence is making everyday tasks faster and easier, but a growing body of research is raising concerns that the same convenience could weaken the very mental skills people rely on most.
Chatbots such as ChatGPT, Claude and Gemini can draft emails, translate text, suggest travel plans or help write computer code with only a short prompt. Their speed and versatility have made them useful in classrooms, offices and personal life alike. Yet researchers and educators are increasingly asking whether handing over too many thinking tasks to AI could leave memory, decision-making and critical thinking underused.
A recent American-British study involving 1,222 people, still awaiting peer review, found a mixed result: participants who used AI tools performed better in the short term on arithmetic and reading-comprehension exercises, but their results worsened over time and their willingness to keep trying declined when the tools were not available. The researchers said that was especially troubling because persistence is closely tied to learning and skill-building.
Grace Liu, a Carnegie Mellon University doctoral student and the study’s lead author, said AI can strip away valuable chances to learn. She argued that the technology is unusual because it is not limited to a single task, but can be used across many kinds of reasoning and intellectual work.
That broad reach sets generative AI apart from earlier digital aids. Calculators, for example, could speed up arithmetic, but they did not replace the human reasoning needed to work through a problem. With generative AI, by contrast, users can be tempted to outsource not only the answer but also the process of getting there.
Other research has pointed in a similar direction. A 2025 MIT study that drew widespread attention reported that students using generative AI to write essays showed weaker critical-thinking abilities. Scholars have described the pattern as “cognitive offloading,” or in stronger terms, “cognitive surrender,” as people increasingly let machines do the mental heavy lifting.
Johann Chevalere, a researcher in social and cognitive psychology at France’s CNRS, said humans naturally try to conserve energy and often choose shortcuts that get them to the core of a task more quickly. He said generative AI may intensify that tendency, and warned that if certain skills are never exercised, the brain may stop maintaining the connections involved in them.
In response to such concerns, developers have started adding features meant to keep users engaged rather than passive. Some AI tools now include “Socratic” or guided-learning modes that offer hints instead of direct answers. OpenAI’s ChatGPT has a study mode, while Google’s Gemini offers guided learning. Microsoft said its Copilot models warn users that mistakes are possible and remind them to verify results. The company said the risk of excessive cognitive offloading is real, especially when AI is used to automate tasks that also help people build skills.
Researchers say the long-term effect of generative AI on the human brain remains unclear because large-scale, long-term studies are still lacking. For now, the debate centers on a trade-off that is becoming more visible across schools, workplaces and daily life: short-term convenience versus the possibility of long-term weakening of human cognition.
This report is based on a monitored public feed and attributed to Dawn Tech.
Source: Dawn Tech - https://www.dawn.com/news/2015756/generative-ais-power-sparks-fears-of-dumbing-humans-down
