As AI’s promoter-in-chief, Altman recently spoke of his disappointment with the speed of AI’s advancement at an industry conference. According to The New York Times, he complained that there was “more resistance to ‘the diffusion, the absorption’ of AI into the culture and economy than he expected.” The Times also quoted Altman as saying “Looking at what’s possible, it does feel sort of surprisingly slow.”
And he’s not alone amongst AI titans. Nvidia CEO Jensen Huang is quoted in the same story as saying that AI skeptics are “scaring people from making the investments in AI” that would make it better.
Which is no surprise, given that Anthropic CEO Dario Amodei regularly publishes essays foreseeing AI burning down half of all white collar jobs in the next five years.
The executives are blaming others for this crisis in AI confidence. It’s the public’s fault. It’s the market’s fault. It’s the critics’ fault. But the problem is more fundamental: AI’s leading companies have run afoul of a market development principle called the “adjacent possible.”
That principle asserts that innovations only truly catch on when two factors connect: One, the new thing works reliably, and two, people understand why they need it. Simply creating a cool new technology is never enough; fail to bring the public along, and you wind up with either weak demand (think Segway) or a backlash (like with nuclear power in the 1980s).
While demand for AI isn’t weak, it’s weaker than its proponents think it should be. And at the same time, a backlash against AI has been brewing over the technology’s potential impacts.
The concept of the adjacent possible was popularized in Steven Johnson’s 2010 book, Where Good Ideas Come From: The Natural History of Innovation. Historical patterns precede an explosive moment when an innovation – the pencil, the flush toilet, batteries, the smartphone – catches on and changes the way we work or live.
“Possible” technologies already exist, work well, and have been adopted by consumers and businesses. “Not-yet-possible” technologies are untested, unreliable and not yet well-understood by their target market.
Today, for instance, mass-market electric cars land in the possible. Flying cars in every driveway land in the not-yet-possible.
The adjacent possible is a thin band between those two zones. Innovations change the world when they land there, stretching boundaries and changing habits — but not so much that the technology keeps glitching or makes us feel uneasy. When an innovation hits this sweet spot, the result is user delight and new consumption patterns that create popular enthusiasm and rapid, broad adoption.
For example, when the Wright brothers first flew in 1903, all the mechanics and theories necessary — from the piston engine to wing aerodynamics — already existed. The Wrights just had to push the technology a bit further by putting the right parts together and adding some key insights of their own.
And by that point, inventors had been trying for years to fly, so the public was ready to believe a machine could deliver on the promise. Twenty years earlier, powered flight was science fiction to most people. But news of the first successful flight at Kitty Hawk was reported breathlessly by newspapers around the country, and airplanes were soon embraced by an excited public.
Which circles back to AI. While artificial intelligence has been around for decades, for much of the population it seemed to suddenly slam into their lives with the introduction of OpenAI’s ChatGPT in late 2022. AI has since advanced faster than any technology most of us have experienced.
We’re being told over and over by the tech crowd that AI is going to change everything – the way we work, our careers, our art, our politics – and might even come to control us.
It’s too much too fast. The mass market can grasp that AI is better than search. The adjacent possible would tell us that it’s a leap we can make from where we were to where we’re going.
But telling us that we should already have a team of AI agents doing half our jobs and making us ten times more productive – or alternatively that we’ll all be unemployed in the near future — is just too far of a leap. Further, it’s a leap attached to a threat.
And Altman and Huang and other AI industry leaders wonder why AI adoption is lagging their expectations?
AI companies need a strong dose of adjacent possible medicine right now. The technology may be moving at breakneck speed, but the general public is not. In tech product planning it’s always better to hit the sweet spot now while building toward a future that may take time to digest.
So AI leaders might consider dialing back the revolution and instead focus on turning out products and services today that push us into new territory at a human pace. Map out a journey into the future for us that we can buy into without feeling threatened – or risk more pushback from the public and, eventually, policymakers.
The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.
