Harry's Teleprompter Fail Shows the Hidden Cost of AI Reliance

Harry's Teleprompter Failure: A Stark Irony

There was a stark irony on stage at the Clinton Global Initiative in New York this week. Prince Harry was midway through an impassioned warning about the dangers AI poses to children when his teleprompter went dark. He stood in silence for several long seconds, said “And that’s it,” and walked off.

When he returned about 40 minutes later, he joked that AI had gotten into his speech.

I felt for him. Most of us have had a moment when the words we were counting on simply vanished. But his stumble holds a warning for all of us. When we lean too heavily on intelligence outside ourselves — a speechwriter, a screen, a chatbot — we’re left exposed the moment it fails us.

Harry’s teleprompter fail is a version of the same trade. When the words were never built in your own head, there’s nothing to fall back on when the screen goes dark.

Why We Outsource Thinking—and What It Costs

Behavioral science tells us humans are cognitive misers. Given the chance to spare our brains effort, we take it. That’s always been true. What’s changed is that GenAI now offers that shortcut instantly, for free, on almost everything we do.

The catch? The less we use our cognitive muscles, the weaker they get. And expertise offers far less protection than we’d like to think. One study of highly experienced doctors found that after AI detection tools were introduced, their ability to spot precancerous growths on their own dropped from 28.4% to 22.4% within just three months. That’s a 20% decline in a highly developed skill built over many years.

Research from Boston Consulting Group found that the skills leaders see as most vital to long-term performance are the very ones most at risk from AI. Topping the list are judgment and decision-making, and problem understanding and framing. These are exactly the skills we need to decide when to use AI at all. So our reliance feeds on itself.

The Evidence: Skills Decline When AI Takes Over

This isn’t hypothetical. A study of highly experienced doctors found that after AI detection tools were introduced, their ability to spot precancerous growths on their own dropped from 28.4% to 22.4% within just three months. That’s a 20% decline in a highly developed skill built over many years.

Research from Boston Consulting Group found that the skills leaders see as most vital to long-term performance are the very ones most at risk from AI. Topping the list are judgment and decision-making, and problem understanding and framing. These are exactly the skills we need to decide when to use AI at all. So our reliance feeds on itself.

The Trade-Off We Must Make

Reliance feeds on itself. Once a task takes 15 seconds with AI, deciding to return to a way that requires more cognitive effort is hard—our brains push back and convince us the faster path is the more prudent one. That is why thinking for ourselves increasingly requires a conscious and courageous decision to choose discomfort over convenience, again and again.

The stakes are simple: neurons that fire together wire together, and those left idle don’t just rest—they start to go into a coma, and the longer they lie dormant, the harder they are to bring back to life.

Harry’s teleprompter fail is a version of the same trade. When the words were never built in your own head, there’s nothing to fall back on when the screen goes dark. We should take our own lesson from his moment of embarrassment and decide to be more conscious about outsourcing the thinking we cannot afford to lose.