Don’t Let AI Do All the Thinking: Using Large Language Models (LLMs) Intentionally

Introduction: AI, the Great Enabler or Enfeebler?

Generative AI has become the intern we always dreamed of: efficient, fast, and increasingly competent. But what if this helpful assistant is also quietly atrophying our minds?

A provocative new study from the MIT Media Lab suggests that relying too heavily on ChatGPT might actually dull our brains. The research raises a critical question: Are we outsourcing not just our work, but our thinking? In a world where AI tools are becoming ubiquitous, the challenge is no longer access. It’s an intention. Do we use these tools to become wiser, or simply to do less thinking?

Your Brain on ChatGPT: The MIT Study

MIT Media Lab’s preprint study, “Your Brain on ChatGPT,” placed participants into three groups: those using ChatGPT, a search engine, or no external tool. Over four months, they wrote essays under these conditions while their brain activity was monitored via EEG.

The results were striking:

• Participants using ChatGPT showed weaker neural connectivity, indicating lower mental engagement.
• Even after switching to unaided writing, previous ChatGPT users had lingering reductions in recall and cognitive effort.
• Those using their own brains or search engines showed stronger memory and essay ownership.

In essence, AI didn’t just write for them, it thought for them. And that came at a cognitive cost.

The takeaway? How we use AI matters immensely.

Using AI to Get Smarter, Not Just Faster

The key to benefiting from AI lies in transforming it from a vending machine of answers into a tutor, coach, and thought partner.

1. Ask It to Teach, Not Tell

Prompt ChatGPT or other LLMs to walk you through ideas, not just deliver conclusions. For example: “Explain quantum tunneling, then quiz me on it.”

2. Engage in a Socratic Dialogue

Good learning is a conversation. Push the model to challenge your assumptions or present counterarguments.

3. Scaffold Your Thinking

Use AI to break down big tasks. Don’t ask it to write something you need. Ask for help outlining, brainstorming, or improving a draft you wrote.

4. Test and Reflect

Treat outputs as hypotheses. Validate them. Explore alternatives. Use LLMs to simulate the mental friction that builds real understanding.

Tools That Make You Think, Not Just Produce

Several AI tools are designed to enhance human learning rather than bypass it.

• ChatGPT’s Study Mode: OpenAI’s new Study Mode helps learners engage actively by asking questions, prompting quizzes, and coaching users through concepts step-by-step. It’s especially strong in science, math, and tech topics, turning ChatGPT into more of a tutor than a typist.
• Gemini’s Guided Learning Mode: Google’s Gemini offers interactive lessons that mix visuals, quizzes, and tailored feedback. It guides users through learning journeys, from cheesemaking to coding, with embedded resources and checkpoints.
• Khanmigo: Khan Academy’s GPT-powered tutor works with both students and teachers. It scaffolds questions, helps plan lessons, and tracks comprehension without giving away answers too easily.
• TutorAI: This tool offers a sleek interface, adaptive content, and progress tracking. Though accuracy isn’t perfect, it’s a smart starting point for structured self-education.

Conclusion: The Choice Is Ours

Whether AI becomes an amplifier of intellect or a crutch for cognition depends entirely on us. Used passively, LLMs can lull us into intellectual laziness. Used intentionally, they can accelerate learning, deepen understanding, and elevate our thinking.

At Admind, we help businesses and teams use AI tools wisely and creatively. If you’re curious about how to integrate LLMs into your workflows, please reach out. Let’s build a smarter business together.