Ten years ago, streaming a movie over Starlink at 30,000 feet would have sounded like fantasy. We are about to be just as wrong about AI, and the people who get the next decade right will not be the ones you expect.
We are reliably bad at imagining ten years out, and we miss in one direction: we underestimate. AI is the next thing we are underestimating. The instinct in most rooms is that the young will lead and everyone else will catch up. That instinct is wrong twice over, and the answer to why is older than any of the tools.
A few weeks ago, I sat at a dinner with regional leaders from government, business, and consulting. Someone said the line that stopped the room: getting on a plane, connecting to Starlink at 30,000 feet, and streaming a movie on your phone is now completely ordinary. Ten years ago, it would have sounded like fantasy.
That line points at something we keep getting wrong. We are bad at projecting ourselves ten years forward, and the error has a direction. We do not overshoot and imagine flying cars that never arrive. We undershoot. The absurd becomes the thing we do without thinking, and we forget we ever found it strange. Ask anyone in 2014 whether they would stream video from an economy seat above the Pacific, and they would have laughed.
The conversation turned, as these conversations now do, to AI. And here is the uncomfortable part: if we have always underestimated the next ten years, we are almost certainly underestimating this one. Not in the breathless way the hype merchants mean, but in the quieter way the Starlink line captures, where the extraordinary becomes ordinary faster than our judgment can recalibrate.
So the question is not whether the next decade will be strange. It will. The question is who navigates strangeness well. The room, including myself, arrived with an answer almost everyone shares: the young will lead, and the rest of us will catch up.
Digging into it deeper since the dinner, I came to the following conclusion: That answer is wrong twice over.
It is wrong about the facts first. The reflex says digital natives are the AI natives, but inside companies, the pattern inverts. A 2025 London School of Economics study found that while personal AI use is highest among the youngest workers, workplace adoption climbs with seniority and pay. The most experienced people are often using AI at work faster than their early-career colleagues, who everyone assumed were already fluent. The fluency on a teenager’s phone does not automatically translate into business leverage. Seniority brings the budget, the latitude to experiment, and the judgment about what is worth pointing the tool at. That last part is the one no app teaches.
It is wrong on the framing second, and this is the deeper error. The whole premise, one group leads and the other catches up, treats a partnership as a race.
Consider what each person actually brings. The younger one arrives with fewer inherited assumptions, greater openness to AI-native ways of working, and greater comfort with uncertainty. What they lack is judgment about what good looks like, what is worth building, what will land in the real world. The more experienced one brings exactly that judgment, the pattern recognition, the instinct learned over decades, and a few scars for what is worth doing. What they sometimes lack is the willingness to pick up a tool that makes their hard-won way of working feel slow.
The blank canvas meets the seasoned eye. Neither one wins. That is the whole point.
This is a tension, not a problem to solve, and tensions punish you for choosing a side. Lean only on the seasoned eye, and you ship the same product you shipped five years ago, now with an AI label that changes nothing the customer feels. The judgment is intact, and the canvas stays blank. Lean only on the blank canvas, and you run 200 experiments a quarter without being able to name three assumptions you actually killed. The canvas fills with confident, well-formatted noise, and the judgment never arrives. Each pole, left alone, decays into the other’s failure mode.
What both poles together serve is the only thing that survives the decade: value the customer can feel. Speed is nearly free now; everyone is getting faster, so faster stops being an advantage. Accuracy still needs a human because the model will hand you a fluent, well-cited, confidently wrong answer, and the cost of trusting it without judgment gets paid later in retracted work and shipped bugs. And the read on whether something will actually matter to a real person who does not live in the model at all. That is the seasoned eye’s contribution, and nothing in the tool is trying to replicate it.
None of this is new. Curiosity is what lets the seasoned eye look at the new canvas without dismissing it. Openness is what lets the blank canvas accept an edit from a hand that looks slower. Systems thinking, first principles, constant learning: these are not requirements AI introduced. They are the requirements that every generation of leaders has lived by, now being run at a pace none of us has experienced. The tools changed. The fundamentals did not.
Which points at the move worth making this quarter?
Stop designing your AI rollout as a one-way class. The reverse-mentoring program where the youngest teach the most senior how to prompt is half a picture, and the half that flatters the assumption we just dismantled. Build the other half. Put senior judgment and junior fluency together as a single working unit, with one shared outcome on the wall, not a teacher and a student facing in opposite directions. The senior brings the question worth answering and the read on when the output is quietly wrong. The junior is willing to try 10 approaches in the time it would take the senior to try 1. Neither is the mentor. The pairing is the unit.
And the move is fractal. It works with one senior individual contributor and one early-career hire on a single deliverable. It works with a tenured general manager and a younger product leader co-owning a new line. It works with the board’s sharpest pattern reader, paired with the company’s most AI-native builder, on the AI strategy itself. It even works at the scale where the dinner began: the regions that pair the people who built the last economy with the people who will build the next one are the ones that compound. Same shape, every altitude.
Before your next leadership review, three questions. Who in your most senior ranks is genuinely fluent with AI? If the honest answer is nobody, your reverse-mentoring program is solving a problem you do not have. Who in your junior ranks has the judgment to catch the AI when it is confidently wrong? If nobody, your productivity numbers are partly fiction. And where are these two already working side by side, by accident or design, and what are they making together? Whatever it is, it is the clearest preview you have of the whole company over the next two years.
The Starlink line was meant to be about how strange the present has already become. It was really about how strange the next ten years will be, and how badly we will misjudge them if we get the pairing wrong. The blank canvas needs the seasoned eye. The seasoned eye needs the blank canvas. The room that puts them at the same table, facing the same outcome, is the room that gets the decade right.
Originally published at orion.beehiiv.com.
Sources
- Protiviti and London School of Economics, Generations in the Workplace: AI Adoption and Generational Diversity, 2025. https://www.protiviti.com/uk-en/survey/lse-generations-survey Accessed 2026-05-31.
- Megan Leonhardt, AI’s Generation Gap, CFO Brew, October 30, 2025. https://www.cfobrew.com/stories/2025/10/30/ai-s-generation-gap Accessed 2026-05-31.