Sir Ken Robinson once stood on a stage somewhere in the early 2010s and told an audience something genuinely strange.
He said he’d been speaking to somebody recently from one of the main computer companies, who’d told him that the most powerful computer on Earth at the moment had the processing power of the brain of a grasshopper.
He paused, smiled at the audience, and said: “I don’t know if that’s true. But it sounds great, doesn’t it? You should trot this out at your next dinner party and then look confident. That’s the key to it.”
The room laughed.
He kept going.
The engineer had told him, Robinson said, that in the not too distant future, the most powerful computers would have the processing power of a six-month-old baby. And that when that happened, it would be a big shift — because the computers would become capable of learning. They’d rewrite their own operating systems in the light of their experience.
“Well, that’s Skynet, isn’t it? That is Terminator 3.”
The room laughed again.
Then Robinson delivered the line that matters, the one I haven’t been able to shake for years.
He said that, according to people like Ray Kurzweil, in the not too distant future, for $1,000 you’d be able to buy a laptop computer with the same processing power as an adult human brain.
And then he asked the question almost nobody else in the AI conversation was asking.
“So how’s that going to feel?”
Not what’s the strategy. Not what’s the policy. Not what’s the business case.
How’s that going to feel?
“In the not too distant future, maybe in your working lives, you may find yourself sitting in front of a computer, a laptop, that’s as intelligent as you are. Not as attractive. Not as in demand socially. But as smart as you are. You give this thing an instruction and it hesitates. It says, ‘Well, I don’t know. Have you thought this through? I’m not sure you have.'”
That’s the line. That’s the entire argument compressed into a joke.
And I want to write about it today because the world Robinson was warning about isn’t a hypothetical anymore. It arrived. And he isn’t here to ask his own question of the people now living in his prediction.
What the data actually shows
Let me give you the verifiable picture, because Robinson’s framing turns out to be unusually accurate.
The Kurzweil prediction he was citing was real. Ray Kurzweil — the inventor, futurist, and now Google’s Principal Researcher and AI Visionary — argued in books going back to 1999 that the processing power available for $1,000 doubles roughly every year, following a kind of accelerated Moore’s Law.
His specific prediction: $1,000 of computing power would equal the processing power of one human brain — which he estimated at roughly 10^16 (about 20 million billion) calculations per second — around the year 2023.
The actual data point, from Kurzweil’s 2024 book The Singularity Is Nearer: in 2023, $1,000 of computing power could perform up to 130 trillion calculations per second.
That’s 10^14, not 10^16. So Kurzweil was about two orders of magnitude off on the cost-per-flop curve — though he argues, plausibly, that we’re now within striking distance of the threshold given the trajectory.
But the headline-grabbing benchmark isn’t really the point. The point is that Robinson, talking from a stage at least a decade ago, was warning a non-technical audience about a curve that turned out to be directionally right.
The grasshopper-brain computers he was talking about have, in fact, become baby-brain computers. The baby-brain computers have become something that, when you ask them for a strategic memo, will produce something better than the average MBA’s first draft in 90 seconds.
And the $1,000 laptop with adult-human-brain processing power? Not quite yet by every measure. But close enough that the experience Robinson was describing — sitting in front of a machine as smart as you are, not as attractive, not as in demand socially, but as smart as you — is no longer hypothetical. It’s the experience anyone working with frontier AI in 2026 has had.
Robinson didn’t live to see this
There’s a detail about this story that I think is the heart of it.
Sir Ken Robinson died on August 21, 2020, of cancer, at the age of 70.
He was warning us about this moment, on stages around the world, for at least a decade before he died. He framed it as a joke about dinner parties because that’s what made it land with audiences. The joke was the wrapping. The warning was the gift.
But he didn’t live to see whether his framing was correct.
He died before GPT-3. Before GPT-4. Before Claude. Before Gemini. Before the moment when senior executives around the world started quietly opening up an AI tool on their desk and asking it questions, and feeling something complicated when the answers came back.
We’re the ones who get to find out if Robinson was right.
And on the evidence so far, he was. Not on the technical specifics — Kurzweil’s curve and timing are imperfect. But on the human specifics. On the question almost nobody else was asking.
How does this feel?
The question we keep avoiding
In every senior leadership conversation about AI I’ve been part of in the last eighteen months, there is a moment that almost always gets skipped.
The strategy conversation gets had. The vendor conversation gets had. The governance conversation gets had. The talent conversation gets had.
The one Robinson was actually asking — how does this feel? — gets quietly skipped.
It gets skipped because it’s uncomfortable. Because it doesn’t have a tidy deliverable. Because it doesn’t translate into a slide. Because admitting “the tool on my desk is, by any honest measure, as smart as I am about most of what I spend my day doing” is a sentence most senior leaders have spent thirty years building careers that wouldn’t require them to say.
But Robinson’s question is the only one that actually matters for what comes next. Because the experience of working with a peer-level AI is what reshapes the leader. Not the strategy memo about it. Not the vendor pitch. The actual moment of giving it an instruction and watching it think.
Robinson described the experience precisely. “You give this thing an instruction and it hesitates. It says, ‘Well, I don’t know. Have you thought this through?'”
I think about this every time I watch a senior executive use a frontier AI for the first time on something that actually matters to them. There is always a moment — sometimes thirty seconds in, sometimes three minutes — where their face changes. They’ve stopped treating the tool as an interesting demo. They’ve started treating it as someone in the room.
That’s the moment Robinson was warning about.
And what happens next, for that executive, determines almost everything about how they’ll lead in the next decade.
What separates the leaders who handle this well
Some leaders, after that moment, become better. Curiouser. More humble. They start using the tool seriously. They start asking it harder questions. They start letting their thinking get sharpened by it. Their work improves visibly.
Other leaders, after that moment, retreat. They go quiet about AI internally. They delegate it. They start posting LinkedIn updates about “AI transformation” without using the tools themselves. They build distance between themselves and the technology, because the proximity is too uncomfortable.
The difference between these two responses, in my observation, has almost nothing to do with intelligence, age, or technical background. It has everything to do with whether the leader is willing to sit with Robinson’s question honestly.
What is my value, when this exists?
The leaders who can sit with that question — really sit with it, not deflect it — come out the other side different. They’ve usually concluded something like: my value isn’t in producing the first draft. It’s in knowing which draft to produce. It’s in the taste, the judgment, the human relationships, the ethical bottom-lines, the strategic intuition that comes from twenty years of being in rooms. It’s in the things that the machine genuinely can’t do — not yet, and possibly not ever — even if it can do a startling amount of what the rest of my job consists of.
The leaders who refuse to sit with the question can’t update their conception of their own value. So they default to defending the old conception, which the machine is rapidly undermining. They look brittle. They start sounding like the kind of executive who, in 2026, is still describing AI as “interesting but not really applicable to my industry.” We all know that executive. We can all see what’s going to happen to their career.
The Robinson question is the entry ticket to the next decade of leadership. The leaders who pay it walk through. The leaders who try to skip the line stay stuck on the wrong side.
Two specific exercises
If you want to actually do what Robinson was suggesting — sit with the question rather than deflect it — here are two specific exercises.
The first. Open the AI tool you use most. Give it something hard from your real work this week. Not a demo prompt. Not a “let’s see what it can do” question. Something you’d genuinely been planning to spend two hours on. Let it produce its answer.
Read the answer slowly. Don’t look for flaws. Don’t grade it. Just read it.
Then ask yourself, not your team: where, specifically, would my version have been better? Where, specifically, would it have been worse? Where, specifically, would it have taken me longer to produce something this good?
The honest answer to those questions is the beginning of an updated understanding of your value in 2026.
The second. Have one conversation this week with a peer-level senior leader, openly, about the Robinson question. Not about AI strategy. About the feeling of using AI. About what it’s like, as someone who’s spent decades in your domain, to watch a machine produce competent work in your area in seconds.
The conversation will be awkward. It will probably be the most useful conversation about AI you’ve had this year. Because almost no one is having it, and almost everyone needs to.
The closing thought
Robinson was a particular kind of speaker. He used jokes to deliver hard truths. The Skynet bit, the Terminator 3 bit, the dinner party advice about looking confident — these were the wrapper. The actual cargo of the talk was a question almost nobody else was asking, posed to people who weren’t going to ask it of themselves.
The technological innovations, he said, are moving faster than ever and becoming more encompassing.
He said that last sentence in the early 2010s. Roughly fifteen years ago. The acceleration he was warning about has, if anything, exceeded the pace he described.
He died in 2020. He didn’t get to see the world he was warning about arrive.
We did.
The question he left us — how’s that going to feel? — is the one that, more than any other, will sort the leaders who define the next decade from the ones the next decade leaves behind.
It’s a question you can answer.
But you have to actually sit with it. You can’t strategize your way around it. You can’t delegate it. You can’t get a consulting firm to do it for you.
You have to open the tool. You have to give it something real. You have to feel what arrives.
That feeling is the start of every honest conversation about AI worth having.
And it’s the question Robinson asked us when we still had time to prepare.
We didn’t, mostly. We laughed at the dinner party joke and moved on.
The good news is, it’s not too late to actually answer him.
The world has changed. The leaders who notice will be the ones the next decade is built around.
