Olga Rolenko/Getty Images
- If you're concerned you're outsourcing too much critical thinking to AI, you're not alone.
- We asked tech executives, professors, and neurologists what can be done to stay sharp.
- They suggest gaining deep expertise, creating a brain routine, and writing first drafts without AI.
It's getting harder to resist the temptation to pull up a chatbot and switch off your brain for simple tasks, like replying to emails.
But that convenience comes with a downside: If you stop engaging in tasks that require critical thinking, creativity, and judgment, those mental muscles can weaken over time.
As AI becomes increasingly embedded in everyday work, some researchers have found that a heavy reliance on AI is quietly deskilling workers.
It's a concern for average employees, too. Nearly half of 2,950 workers surveyed by Workday last year worried AI agents would lead to a decline in critical thinking.
Anurag Dhingra, Cisco's senior vice president and general manager of enterprise connectivity and collaboration, told Business Insider in October that the concern comes with every major technological revolution.
It's the age-old question, Dhingra said: "Are we getting too dependent — and does that mean we're getting dumber as a result?"
The short answer? Not necessarily.
Dr. Majid Fotuhi, a Johns Hopkins professor who studies neuroplasticity and Alzheimer's prevention, told Business Insider that passively relying on technology can lead to a decline in our ability to think critically. However, if used correctly, AI can force our brains to process and analyze more information, which is good.
We asked executives, professors, and neurologists about how AI users can stay sharp:
1. Go deep on a topic
Dhingra said that developing a deep understanding of a topic can help keep critical thinking skills fresh. For example, he said he invests a lot of time in understanding how AI models are built and function, using a range of models as tools to help him digest information.
Gloria Mark, a chancellor's professor of informatics at the University of California, Irvine, suggests adopting everyday habits that encourage deep thinking, like reading challenging long-form material or taking online courses that require sustained focus.
"It's more about lifelong learning — returning to what we did as students to build our capacity," she said.
She cautioned against "the trap of taking the path of least resistance and letting AI do all the work."
"Stay in the loop by doing intellectual work."
2. Write your own first draft
There's value in getting your own ideas down first.
Geetha Rajan, a strategy leader who previously headed AI adoption at PwC, told Business Insider that she always writes her own first draft, even if she later uses AI to verify numbers, extract unstructured data, and challenge her work.
Joe Depa, global chief innovation officer at EY, told Business Insider that he similarly tells his teams to write content as they would in an email before turning to an LLM.
"Then ask Copilot, or whatever AI tool you're using to help modify," Depa said, adding that sometimes it will find some issues or ask a follow-up question that challenges your thinking. He said it's "much more productive" to use AI for enhancement.
Vivienne Ming, chief scientist at the Possibility Institute and founder of Socos Labs, suggests using AI to challenge your thinking.
"Here's my argument — what am I missing? What's the strongest case against this?" she said, describing the approach as "productive friction."
3. Find ways to challenge your brain
Michael Merzenich, a professor emeritus at the University of California, San Francisco Medical School and a pioneer in brain plasticity research, told Business Insider that the brain "needs exercise" in order to stay sharp.
Merzenich, who is also the founder and chief scientific officer of Posit Science Corporation, said that when humans solve problems, they engage their brains in reasoning, make connections, and recall important information. When people use AI to get quick answers, they're getting it without effort — and not coming to any conclusion on their own.
Johns Hopkins' Fotuhi told Business Insider that picking up a new hobby can serve as a daily brain routine. He also said he makes a point of testing his memory and attention span whenever possible, such as by memorizing names or credit card numbers.
Aniket Kittur, professor at Carnegie Mellon University's Human-Computer Interaction Institute, said AI users should do things that feel hard.
"The harder you think, the more you benefit," he said, adding that if something feels effortless, you're probably not building real skill.
He recommends reading material that challenges your thinking, watching complex shows that require you to follow multiple threads, and taking a new route home without relying on GPS.
4. Think before you prompt
Mike Wynn, Bank of America's Academy executive for AI capabilities and enterprise learning products, told Business Insider that AI can enhance strategic thinking. However, quickly prompting an LLM and relying on the output without validating it undermines that benefit.
He said people often use AI like a search engine, and AI will find answers because it's designed to do so. However, Wynn said users should think about why they need a summary and how it'll be used.
Jacob Sherson, a professor at Denmark's Aarhus University and founder of the Center for Hybrid Intelligence, similarly advised that workers stay firmly in control of the thinking process.
His research outlines a framework called FERC — short for Frame, Explore, Refine, Commit — designed to ensure humans remain authors rather than passive consumers of AI output. In practice, that means slowing down at the start before you use AI to generate multiple options, compare them, and actively critique them.
"If you only review one output, you are not evaluating — you are accepting," Sherson said.
5. Don't accept every output
Sol Rashidi, a former tech executive at IBM, AWS, and Estée Lauder, previously told Business Insider that workers need to develop discernment muscles. She said a large percentage of content is AI-generated and that's being used to retrain models.
With that in mind, Rashidi said you can ask AI for a response, but workers shouldn't copy and paste that output because it may not be accurate.
Ming said many users fall into what she calls a "competence illusion," where AI-generated work feels like their own, even if they couldn't produce it from scratch.
She recommends explaining the reasoning behind anything AI helped produce out loud — to a colleague, to yourself, or even your dog — without looking at the screen.
"You're not summarizing it; you are teaching it," she said. "If you can't, you haven't done any thinking; you've just done some very sophisticated copying."
Are you experiencing deskilling at work? Contact this reporter via email at tspirlet@businessinsider.com or Signal at thibaultspirlet.40. Use a personal email address, a nonwork WiFi network, and a nonwork device; here's our guide to sharing information securely.