The Biggest Risk of AI May Be That Humans Stop Thinking for Themselves

The biggest risk of AI may not be that machines start thinking like humans. It may be that humans stop thinking for themselves.

I recently read a fascinating CIO article, When AI explains its decision, humans may stop thinking independently, about research into how people respond to AI recommendations. The surprising part wasn’t that AI sometimes gets things wrong. We know that. It was what happened when AI gave people a convincing explanation for its recommendation.

The researchers studied 228 experienced evaluators reviewing innovation proposals. Overall, evaluators accepted LLM recommendations 67% of the time. But the more interesting finding was what happened when AI explained why it made a recommendation. A convincing narrative could actually reduce independent human verification and suppress what the researchers called “productive overrides” — those moments when a human looks at what AI is recommending and decides to challenge it.

In other words, AI didn’t necessarily become more correct. It became more persuasive.

That should get our attention.

As AI becomes better at writing, reasoning, analyzing, and explaining itself, we’re going to need to become better at something deeply human: disagreeing.

That moment when you read an answer and think, Wait. I don’t agree. That matters. Your experience matters. Your judgment matters. Your intuition matters. Your empathy matters. Even those butterflies in your stomach telling you something isn’t quite right matter.

I keep coming back to a question that has become central to my writing and my Human First work: How do we be human again in a digital world?

Maybe part of the answer is refusing to outsource our thinking.

When a Good Story Becomes the Answer

This research struck me because generative AI is exceptionally good at telling a convincing story. Give it a conclusion and it can construct a logical explanation around it. It can organize the facts, present supporting evidence, anticipate objections, and deliver everything in language that sounds confident and authoritative.

That is an extraordinary capability. It is also something we need to understand.

We naturally associate a well-reasoned explanation with expertise. When something is clearly written and logically presented, we’re inclined to believe whoever — or whatever — produced it understands the subject.

But fluency isn’t the same as truth. Confidence isn’t the same as accuracy. And a convincing explanation isn’t necessarily a correct explanation.

AI can sell us a convincing story.

The responsibility for deciding whether that story makes sense still belongs to us.

Don’t Outsource the Butterflies in Your Stomach

We have spent decades trying to make technology easier for humans to use. Now we may need to spend more time making sure humans don’t become too easy for technology to persuade.

I’ve written and spoken before about gut feeling and those butterflies in your stomach. We sometimes treat intuition as something inferior to data, but experience creates intuition. You’ve been in that meeting before. You’ve watched a customer react that way before. You’ve seen a project showing those warning signs. You’ve hired someone who didn’t look perfect on paper, but something in the conversation told you they were exactly the right person.

Not everything we know can immediately be put into a prompt.

That doesn’t mean we ignore data. It means we don’t ignore ourselves either.

AI can analyze the information. AI can identify patterns. AI can show us things we missed. AI can challenge our assumptions and help us move faster. Those capabilities can make us significantly better at what we do.

But AI should expand human thinking, not replace it.

Before Asking AI, Ask Yourself

There is a simple habit I think becomes increasingly important as AI becomes part of almost everything we do.

Before asking AI what it thinks, ask yourself:

What do I think?

Form an opinion. Consider the problem. Use your experience. Think about what doesn’t feel right. Ask yourself what information might be missing.

Then ask AI.

Now something important changes.

Instead of asking AI to give you the answer, you’re asking it to challenge your thinking.

Show me what I’m missing. Argue the other side. What assumptions am I making? What evidence contradicts my conclusion? Where could I be wrong?

That is a very different relationship with technology.

The machine becomes a tool for expanding human thought instead of a substitute for it.

Speed Without Judgment

We’re going to make decisions faster because of AI. We’re going to analyze more information, produce more ideas, automate more work, and get recommendations in seconds that once took teams hours or days to develop.

That’s powerful.

But I’ve written before that speed without direction is just added noise. The same principle applies here. Speed without judgment can simply allow us to make the wrong decision faster.

Think about what happens inside an organization when every employee has an AI assistant providing recommendations accompanied by articulate explanations. The AI sounds intelligent. The reasoning appears logical. The answer arrives instantly.

Who still raises their hand and says, I disagree?

That isn’t just an AI problem. It’s a leadership problem.

We need cultures where questioning AI isn’t viewed as slowing things down. We need leaders who expect people to challenge the recommendation. We need systems that invite disagreement instead of simply presenting an answer with a beautifully written explanation underneath it.

Human judgment cannot become the final checkbox after AI has already made the decision.

It needs to remain part of the decision itself.

Be Human First

The future shouldn’t be about teaching machines to become more human while humans slowly become more like machines.

That’s the part of this conversation that matters most to me.

Use AI. Learn from it. Let it search thousands of documents, analyze millions of data points, identify patterns, summarize information, automate repetitive work, and challenge assumptions.

But don’t surrender the things that make you human in the process.

Keep your judgment. Keep your curiosity. Keep your intuition. Keep your empathy. Keep your ability to disagree. Keep your voice.

The better AI becomes at thinking with us, the more important it becomes that we continue thinking for ourselves.

Maybe that’s part of how we become human again in a digital world.

Be Human First.