I was at Culturati last spring — an annual conference on culture and leadership held in Austin. During one of the keynotes, three speakers were debating the usefulness of AI regulation. Sharp exchange. Good tension. At the end, one of the moderators turned to the room and asked a question that cut through everything that came before it: "Haven't we gone through these waves before? What did we learn?"
The room went quiet in the particular way rooms go quiet when nobody has a prepared answer.
Afterward I found her and mentioned I'd been thinking about exactly that question. I told her about something Bill Gates wrote in 1996, when the internet was already forming around him: "We always overestimate the change that will occur in the next two years and underestimate the change that will occur in the next ten." She looked directly at me and said, "Please write that, because people need to know."
So here it is.
Before electricity, a factory floor made a particular kind of sense. Every machine — every loom, press, lathe, and drill — ran off a single central steam engine. Overhead, iron line shafts ran the length of the building. Leather belts dropped from those shafts to each piece of equipment below. The floor was organized entirely around proximity to the central drive: heaviest machines closest, lighter ones further out, workers arranged around the logic of the belt. It was coherent. It was also completely determined by the constraints of steam.
In the 1880s, electric motors began to appear. The first thing most factory owners did was replace the steam engine with an electric motor. They kept the line shafts. They kept the leather belts. They kept the floor layout. They captured almost none of the productivity gains electricity made possible.
It took a generation of factory managers — ones who had never organized a floor around a steam engine — to understand what electricity actually changed. Each machine could have its own motor. The floor didn't need to center on a single drive anymore. You could redesign how work moved through the space entirely. The economist Paul David called this the productivity paradox of electrification: the technology arrived in the 1880s, but the gains didn't show up in manufacturing data until the 1920s. Nearly forty years. Not because the technology was slow. Because the thinking about how work should be organized was.
The prediction problem isn't new, and it isn't random. It follows a pattern consistent enough across technology transitions that you'd think we'd have learned to account for it.
In 1987, with personal computers in 15% of American homes, a writer declared the revolution "in shambles" and predicted the whole thing was a fad. By 1997, penetration had doubled. By 2007, it had doubled again. What looked like stagnation was the gap between early adopters and everyone else — a gap that closes not when the technology improves, but when the human systems around it do.
The current AI data suggests the same gap is opening, on a faster clock. ChatGPT reached 100 million users in 60 days — faster than any consumer technology in recorded history. The Stanford 2026 AI Index reports that 88% of organizations now claim some form of AI adoption. Yet Gallup finds only 12% of employees say AI has transformed how work actually gets done in their organization.
Eighty-eight percent pulling the lever. Twelve percent redesigning around the motor.
That gap is not a technology problem. It is the electrification problem, in a suit.
In 1800, a Welsh textile manufacturer named Robert Owen took over a cotton mill in New Lanark, Scotland. Two thousand workers. Five hundred of them children, taken from poorhouses in Edinburgh and Glasgow.
Owen made different choices than his peers. He shortened working hours, built schools, and kept his workers on full pay during a four-month cotton embargo that shuttered other mills. He devised a daily feedback system — a small wooden block above each station, color-coded to reflect the previous day's performance, which he called the Silent Monitor. Visible, immediate, intended to motivate rather than punish. New Lanark became more productive and more profitable than comparable mills. Heads of state and social reformers traveled from across Europe to see how it worked.
Owen believed that if you changed the conditions around people, you changed what they were capable of. He was right about that. He was less right about who should decide what those conditions looked like — that part he kept largely to himself. His workers lived better lives because of his decisions. They had little say in making them. By the standards of his era he was extraordinary. By the standard we should hold ourselves to now, that's not enough.
His model — demonstrably more profitable than the alternative — was not adopted at scale. Not because the evidence was unclear. Because the incentives weren't aligned and no one was required to care.
This asymmetry — between who bears the costs of a technology transition and who captures its gains — is not a 19th century problem. It is the human pattern inside every transition on record. It is also, right now, a leadership choice.
The United States leads the world in AI private investment by a factor of twenty-three. It produces the most advanced models and hosts more than five thousand data centers. The Stanford 2026 AI Index ranks it twenty-fourth in population-level AI adoption.
Singapore, at nearly identical income per capita, ranks second — with an adoption rate more than double that of the United States. The difference is not the technology. Singapore uses the same tools available globally. The difference is that Singapore built the surrounding human system: governance frameworks, national literacy programs, organizational structures that made AI accessible and trustworthy enough for people to actually use.
The country that built the motors is not the country that redesigned the factory floor.
A few months ago I was in conversation with a CFO trying to figure out his company's AI posture. He told me he wasn't ready to move forward yet. I asked why.
"I feel like the executive team hasn't had any of the conversations we need to agree on a path forward."
He changed the agenda of an executive offsite three weeks out to include a two-hour conversation on AI in the business. After the first round of input from his team, it became six hours on day two.
What happened in that room mattered as much as any outcome it produced. People who had been carrying genuine excitement about what AI could unlock finally had somewhere to put it. People who had been quietly managing anxiety — about their team's readiness, about their own — finally had permission to say so. Both feelings had been present for months. Neither had been spoken out loud at the table where decisions get made. The six hours didn't resolve the tension between them. It acknowledged that both were real, and that the organization would need to hold them together rather than pretend one didn't exist.
The outcomes that emerged weren't about which tools to deploy. They were about how the leadership team wanted to lead through this for the next few years. His Chief Customer Officer used that space to do something she hadn't had room to do before: she gathered her direct reports and a broader team to imagine what their world would actually look like in early 2028. From that picture, they built a concrete plan — redesigning their structure, their talent, their processes, and the work itself to get there. Not a tool deployment plan. A vision of the future with an organizational path running toward it.
The conversation that made all of this possible is the one most executive teams are skipping — because there's always a more immediate pressure, a product to ship, a quarter to close. Owen didn't skip it. He asked a harder question about how work should be organized, and his people were better for it, and his business was stronger for it. He still kept too much of the deciding to himself. The leaders who get this right will go further than he did.
The technology will keep moving. The only question that remains is whether the people in your organization are shaping what it moves toward — or just moving with it.