Are You Thinking With or Through AI? A Structured Approach to Using Generative AI

Situation

Generative AI has become an indispensable partner in our daily work writing, research, analysis, and decision-making. Its fluency and speed give the impression that it’s an extension of our thinking.

Yet, what feels like augmentation can quickly become delegation. Research from Michael Gerlich (2025) has shown that when people use AI without guidance, they often offload cognitive effort instead of engaging with it. The result: more polished outputs, but shallower reasoning.

The challenge isn’t that AI thinks for us, but that we stop thinking with it.

Complication

When interactions with AI are unstructured, the human brain naturally defaults to cognitive ease and accepts fluent, coherent information without questioning it. This effect compounds several risks:

• Automation bias: over-trusting AI outputs because they sound authoritative.
• Confirmation bias: framing prompts in ways that reinforce what we already believe.
• Loss of authorship: mistaking AI’s synthesis for our own reasoning.

Unchecked cognitive offloading to AI erodes critical thinking, turning the user from an active analyst into a passive recipient.

Resolution

To counter this, Gerlich’s research proposes a structured prompting protocol, a practical five-step method that transforms AI from a thinking shortcut into a thinking partner. Each step deliberately reintroduces human reflection, evaluation, and ownership into the process.

1. Initial Reflection — Think Before You Ask

Before opening the chat, articulate your own perspective. Write down what you believe, what you don’t know, and what you expect to learn. This primes your reasoning network and prevents immediate anchoring on AI’s first answer.

Example: Instead of starting with “Summarize trends in healthcare AI,” note your own hypothesis: “I think most healthcare AI innovations cluster around imaging and diagnostics. I want to test if that’s true.”

2. Targeted Research Use — Ask for Data, Not Decisions

Use AI to gather information, not to generate arguments. Frame prompts as factual or exploratory: “What data supports…?” or “List recent studies about…,” rather than “Why is this a good idea?”

This step keeps AI in the role of research assistant, not reasoning proxy.

3. Argument Construction — Synthesize Independently

Once you have the information, step away from the model. Construct your own reasoning, connecting evidence to your hypothesis. Integrate facts but write the logic yourself.

This reinforces ownership of thought and distinguishes thinking with AI from thinking through AI.

4. Critical Review — Invite Constructive Friction

Now, re-engage AI as a challenger, not a collaborator. Ask it to test your reasoning:

• “What counterarguments might weaken this position?”
• “What assumptions am I making?”
• “What evidence could disprove this conclusion?”

This stage activates metacognition (thinking about your thinking) and surfaces blind spots that thinking through AI would overlook.

5. Final Reflection and Revision — Own the Conclusion

Review both your reasoning and the AI’s critique. Refine your argument but maintain authorship. Ask yourself: Would I stand behind this conclusion if AI disappeared? That question reclaims your ownership of what you know and why you know it.

Benefit

This structured prompting method turns AI use into thinking with AI rather than thinking through AI. It:

• Preserves critical reasoning: You remain the active thinker, rather than delegating thinking to AI.
• Mitigates bias: Each stage interrupts automatic thinking and exposes assumptions that can distort reasoning.
• Builds intellectual confidence: You understand not just what the AI said, but why you agree or disagree.
• Improves outcomes: Studies show guided users produce arguments that are more coherent, balanced, and original than unguided users.

Closing Thought

AI’s true value lies not in how fast it can answer, but in how deeply it can help us think. By following this structured approach, you replace passive consumption with deliberate inquiry. You don’t just use AI; you engage it to enhance your own reasoning.

Are you thinking with AI or through AI?

Share Your Thoughts

The challenge isn’t that AI thinks for us, but that we stop thinking with it.

Gerlich’s research suggests that structure matters and how we prompt determines whether AI strengthens or substitutes our reasoning. But what about your own experience?

• Do you think Gerlich’s framework could help you think more critically with AI?
• Where do you notice yourself relying on AI for answers instead of insight?
• What habits or structures help you stay the thinker, not the delegate?

Try applying the five-step structured prompting method in your next analysis, strategy session, or writing task and see what changes in your thinking.

You can explore the full study of the framework here: Gerlich, M. (2025). “From Offloading to Engagement: An Experimental Study on Structured Prompting and Critical Reasoning with Generative AI.” Data, 10(11), 172. https://www.mdpi.com/2306-5729/10/11/172