AI Is Not Just Another Tech Trend. It’s a Paradigm Shift.
Every generation or so, a technology comes along that doesn’t just improve what exists. It rewrites the rules entirely.
Economists have a name for this kind of technology: a General Purpose Technology, or GPT. The concept was formalized by economists Timothy Bresnahan and Manuel Trajtenberg, who identified a small set of technologies sharing three defining characteristics: they spread across most sectors of the economy, they improve over time, and they make it easier to invent other things. The steam engine, the electric motor, and semiconductors are their canonical examples. These technologies don’t just do a job. They become the infrastructure from which entirely new categories of work get built.
AI fits that pattern.
The Waves That Came Before
The steam engine kicked off the first Industrial Revolution by creating the factory system. Production moved from scattered craftspeople to centralized manufacturing. The world reorganized around what the machine made possible.
Electrification did it again. It didn’t just replace gas lamps. It enabled mass production, powered entirely new industries, and built the infrastructure that multinational corporations would eventually run on. The same reorganizing logic applied to the digital era: personal computers automated data processing, the internet democratized access to information, and mobile extended everything further.
The pattern across all three is consistent. Each wave expanded what could be automated, who could participate in the economy, and how organizations had to be structured to keep up.
What Makes AI Different
AI continues that arc, but it targets something the previous waves did not: knowledge work and decision-making itself.
Where electricity powered machines and the internet connected people and information, AI automates cognitive tasks. It can process language, recognize patterns, generate content, make predictions, and support complex decisions at scale. A growing body of economic research, including work by Brynjolfsson, Agrawal, and others, positions AI as likely to meet the criteria of a General Purpose Technology, with adoption already spreading across sectors well beyond its origins in technology.
That cross-sector reach is one of the defining characteristics of a paradigm shift. Whether AI fully earns the GPT designation will be clearer in hindsight. What’s already clear is that it targets a different layer of work than any previous wave did, and that alone changes what the transition requires of organizations living through it.
The Payoff Won’t Be Immediate, and That’s Normal
One pattern that repeats across every major paradigm shift is the lag between arrival and impact. This is not a failure of the technology. It reflects the time required to reorganize around it.
Stanford economist Paul A. David documented this dynamic in a 1990 paper comparing electrification to the computer. The lightbulb arrived in 1879. By the turn of the century, electric motors still accounted for less than 5% of factory mechanical drive. It took until the 1920s for electrification to hit 50% diffusion across factories, and only then did productivity numbers start to reflect it. The delay wasn’t because the technology didn’t work. It was because factories had to be redesigned from the ground up. The old layout, built around a single central steam engine distributing mechanical power throughout the floor, didn’t translate. Once factories adopted individual electric motors for each piece of equipment, the productivity gains materialized. The technology had been real the whole time. The systems needed to unlock its value just hadn’t caught up yet.
IT followed the same trajectory. Spending on computers surged through the 1980s and 90s. Measured productivity, for most of that period, didn’t move. The gains came later, after organizations changed their processes, not just their tools.
What This Means for Leaders
History is consistent on one point: the organizations that gained the most from each paradigm shift were not the ones that adopted the technology earliest. They were the ones that rethought how work was structured in light of what the technology made possible.
AI will follow the same logic. Organizations that treat it primarily as a cost-reduction lever, automating existing processes without questioning whether those processes still make sense, will realize limited returns. The ones positioned to pull ahead are the ones that use AI to rethink how knowledge work gets done at a structural level.
That’s a harder ask than buying a tool. It’s also, historically, the only one that’s worked.
AI won’t pay off by automating existing processes. It pays off when organizations rethink how decisions get made, how teams are structured, and how work flows around value. The Building High-Performing Organizations workshop gives leaders a practical framework for doing exactly that.
