As artificial intelligence (AI) rapidly replaces routine and repetitive tasks, concern is growing that users could lose critical thinking and judgment. On April 12, Business Insider compiled ways to use AI efficiently while maintaining thinking skills, based on advice from corporate executives, professors and neuroscientists.

The key is to use AI not as a tool that works and thinks in your place, but as a supporting means to expand and verify thinking. Majid Fotuhi (마지드 포투히), a professor at Johns Hopkins University, said the more people depend excessively on technology, the weaker critical thinking skills can become. He added that using AI appropriately can stimulate the brain to process and analyse more information.

Anxiety in the workplace is also significant. In a survey Workday conducted last year of 2,950 people, about half of respondents said they were concerned AI agents could weaken critical thinking. Cisco executive vice president Anurag Dhingra (아누라그 딩그라) said the long-standing question is being raised again: "Are we relying too much on AI, and are we becoming dumber because of it?"

The first principle experts commonly offered is to write the draft yourself. Gita Rattan (기타 라잔), who led the introduction of AI at PwC, said he uses AI to verify numbers, extract unstructured data or challenge his work, but writes the first draft himself. Joe Depa (조 디파), EY's global chief innovation officer, said he tells team members to organise their thoughts themselves, as if writing an email, before using a large language model (LLM). He said polishing sentences afterward with tools such as Copilot can help find problems or raise additional questions, pushing thinking further and making work far more productive.

The second method is to ask AI for counterarguments. Vivian Ming (비비언 밍), chief scientist at the Possibility Institute, advised asking AI what you missed and what objections could be made to your argument. He described this as "productive friction".

Jacob Sherson (제이컵 셔슨) of Aarhus University in Denmark also stressed that humans must firmly hold the initiative in the thinking process. He presented the "Frame-Explore-Refine-Commit" (FERC) framework and said people should first define the problem, use AI to explore options, then compare and revise them, with the final decision made by a person. He added: "If you review only one output, that is not evaluation but acceptance."

The third is to make a habit of deliberately repeating difficult tasks. Gloria Mark (글로리아 마크), a professor at the University of California, advised building daily habits that force deeper thinking, such as long-form reading or online lectures that require sustained concentration. She said people should avoid the trap of taking the "easiest path" by letting AI do everything, and should keep doing intellectual work while maintaining a state of immersion. Aniket Kittur (아니켓 키투르), a professor at Carnegie Mellon University, also said the harder the thinking process, the more you gain, and that tasks that feel too easy are unlikely to build real capability.

The fourth is to keep training the brain with new stimuli. Michael Merzenich (마이클 머제니치), described as a pioneer of research into brain plasticity, said, "The brain needs exercise to stay sharp." When people solve problems on their own, they go through processes of reasoning, connecting and recalling important information, but those efforts can easily drop away when AI gives instant answers, he said. Fotuhi said it also helps to start a new hobby or check memory and attention by memorising information such as names or card numbers.

The fifth is a verification step of explaining AI-generated results in your own words. Sol Rashidi (솔 라시디), a former executive at IBM, Amazon Web Services (AWS) and Estee Lauder, stressed that people should not copy and paste AI-generated answers as they are, because results may not be accurate. Ming said many users fall into an "illusion of competence," mistaking AI outputs for their own skill. He advised explaining the logic of the result, without looking at the screen, to a colleague, yourself or even a pet. "Not summarising, but teaching," he said, adding that if you cannot explain it, it may be closer to sophisticated copying than to your own thinking.

It was also stressed that while the spread of AI clearly boosts work efficiency, people should not hand over the final responsibility for thinking. The shared conclusion of the advice was that maintaining the process of writing yourself, seeking rebuttals and verifying results is necessary for AI to function not as a tool that erodes capability but as one that strengthens judgment.