AI Is Narrowing Your Judgment — Right When You Need It Widest
You started using AI for the fast stuff — the first draft, the outline, the brief nobody wanted to write. It worked, so you gave it more. Now it drafts the positioning, sizes the segments, suggests the channels, and hands you three options with a confident recommendation on top. So here’s the question worth sitting with: does AI weaken decision making — yours — in the exact moment you’re leaning on it hardest?
The uncomfortable answer is that it can, and it does it so smoothly you won’t feel it happen. The output looks like better thinking. Underneath, it’s often narrower thinking, arrived at sooner.
The mechanism: it narrows before you’ve explored
Good strategy is a choice made after you’ve held the field open long enough to see the non-obvious move. AI is built to do the reverse. Ask it for a marketing strategy and it converges — fast — on the modal answer, the thing most companies in your situation would plausibly do. Left unmanaged, that’s a consensus machine with excellent grammar. And the average of what everyone would do is, by definition, undifferentiated.
The research: the muscle you stop using gets weaker
In 2025, Michael Gerlich studied 666 participants across ages and occupations, measuring how heavily they leaned on AI tools against how they scored on critical thinking. Heavier AI users showed weaker critical-thinking performance. The mechanism he identified is cognitive offloading: when a tool reliably does the reasoning for you, you stop doing the reasoning yourself, and the capacity quietly atrophies.
The danger isn’t that AI gives you a bad answer you’ll catch. It’s that it gives you a plausible answer you’ll wave through — and each time you wave one through without wrestling it, you get a little less able to tell the plausible from the right.
Use AI to widen. Protect the human for the narrowing.
AI is superb at expansion: twenty headline directions, stress-testing your positioning, summarizing what competitors are saying, playing devil’s advocate. Point it at divergence and it earns its keep. The danger is only when you let it do the converging — the actual decision about who you’re for, what you’ll say that no one else will, and which bet you’re not making. That call is the marketing.
Four guardrails for a lean team
- Form your own answer before you read the machine’s. One line, thirty seconds, written down first.
- Use AI to widen the options, never to pick. Ask for eight genuinely different approaches, including a contrarian one. Then you narrow, out loud, with reasons.
- Make it argue against itself. After any recommendation, ask for the strongest case that it’s wrong.
- Protect the irreversible and the differentiating. Speed on the reversible; judgment on the rest.
The quiet symptom to watch for: if you read your own positioning back and it would fit three competitors with the logo swapped, the machine has been doing your narrowing for you. Used well, AI clears the busywork so you have more room for the hard, differentiating judgment. Used passively, it does your thinking a little more each week until the thinking that’s left is everyone else’s.
Related reading: The Sameness Machine · The Reps You’re Handing to the Robot
Sources: Gerlich, M. (2025). “AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking.” Societies 15(1), 6.
