Do you want AI to make you dumber or smarter?
A growing body of research suggests that people who offload their cognitive and creative tasks to AI tools are paying a dear price. Like any muscle in your body, if you don’t actively use your mind it will atrophy.
In a 2025 study, researchers at the MIT Media Lab asked participants to wear EEG headsets to monitor brain activity. They were asked to write essays with and without ChatGPT. AI users showed significantly lower brain activity—particularly in regions linked to memory and reasoning—compared to those who wrote manually. A whopping 83% of AI users could not recall a single quote from their own essays shortly after “writing” them. While this was a small study that has not yet been peer reviewed, it largely supports what other studies have found.
A 2024 Swiss study published in the journal Societies, for instance, found “a significant negative correlation between the frequent use of AI tools and critical thinking.”
When you are busy, it’s easy to leave the thinking to the LLMs. And that’s the beginning of an expertise death spiral. When you outsource your thinking to AI, you are only recycling other people’s ideas.
This reliance on LLMs has a cost on society, too.
We can just look around us and see how generative AI has taken over so much news, sports and social interest reporting. These pieces may be technically perfect—neatly laying out the facts in clear, grammatical English—but they lack the insight and soul that a true expert can bring to an issue. And it can lead to some pretty bad outcomes, too.
Does all this mean experts should avoid AI? Not at all. In fact, if you use it right it can make your expertise even sharper.
The job of an expert is not simply to lay out the facts. The expert’s job is to put them in context, to interpret them and to solve the problems they expose.
This kind of thinking is hard. If you don’t think deeply about a problem, it’s possible to misinterpret the facts and complicate the solution.
This is where AI can be truly valuable—but only after you’ve done the hard thinking. Feed your favorite chatbot like ChatGPT, Claude or Gemini all your notes, research, thinking, conclusions, recommendations, article draft—whatever you are working on and the inputs that went into it.* Then start asking some tough questions.
Here are five specific ways you can use AI to strengthen your expertise:
- Avoid embarrassing mistakes. Ask your chatbot to check your document for spelling and grammar. Ask it to check that you have correctly cited figures or calculated percentages. Ask it to check that your internal logic is sound. Actually tell it to not hallucinate.
- Challenge your assumptions. Ask it to examine your underlying assumptions and find every reason they are wrong or misguided. You may not agree with everything the AI identifies, but this exercise can be invaluable in uncovering gaps and areas you hadn’t thought about.
- Challenge your argument. Ask it to make the strongest possible case against your argument or recommendations. Use the output to strengthen your reasoning. You can even use AI to help you figure out how.
- Uncover alternative solutions. Ask the chatbot the three best solutions to the problem at hand and to explain its reasoning. Often, this approach uncovers at least an idea or two you hadn’t considered before.
- Predict what could go wrong. Buyers rely on experts not only to fix problems but to reduce risk. Ask the chatbot to assume the proposed solution was ultimately a complete failure and to do a post-mortem on what might have gone wrong. Here you turn AI’s proclivity to hallucinate into a brilliant asset.
If you are still trying to wrap your head around AI and how it will transform your work, join the club. Depending on whom you ask, the technology is either evolving faster than they imagined or more slowly than expected. But one thing is certain: AI is here to stay. And you don’t have to jump into vibe coding or agentic AI (but more power to you if you do) to start enjoying the power of these modern tools. You can use them today to improve the quality of your work. So what are you waiting for?
Just don’t let them do the thinking for you.
*If you are working with sensitive or confidential client information, be sure you are working with a closed enterprise version of the chatbot that won’t use your data to train the LLM.
