Ever since frontier models made their way into business we’ve been awestruck and impatient. New releases seem able to do more than any human or collection of humans can do. Companies struggle to keep up. That struggle has a name: capability overhang, or the gap between what AI can do and the uses companies put it to. We are impatient to close it. But should we be?
Capability overhang turns our impatience inwards, to how companies utilize AI, away from the outside world of customers and markets. If only companies adapted quicker, we’d get there sooner. But companies have always suffered from capability overhang, or having more internal capability than they know what to do with.
Many companies have a talent “capability overhang.” Their most entrepreneurial talent leaves to start their own companies. Company culture constrains, rather than enables, risk-taking. If only they could clear the way for their most entrepreneurial people to succeed, imagine the markets they could serve.
Some companies have a data “capability overhang.” They have more data than they know what to do with. Their existing products constrain their ability to build new businesses. If only they could bridge the gap between the data they have and the businesses they can build from it, imagine the markets they could serve.
Other companies have an invention “capability overhang.” They invent technologies they never commercialize, falling behind competitors who do. Their customer relationships constrain what they can take to market. If only they could overcome the gap between the lab and the field, imagine the new markets they could serve.
When companies fail to act on every opportunity before them, they are, in short, behaving like companies. AI “capability overhang” falls into a familiar category. On its own, it doesn’t tell us much about what companies should do when confronted with powerful new technologies. What would?
The gap between what customers have always wanted to do, but couldn’t. The real capability overhang comes from the market, from the experience customers wish they had, but don’t.
People always want faster access to their money. Paper checks used to make us all head to the bank to deposit them. But banks are only open at certain times. ATMs changed this, allowing people to deposit checks after hours. The camera-equipped smartphone went further, allowing people to deposit checks from wherever they are. Technology continued to deliver on faster access to money, closing the gap between what customers want and what they could do.
Every company has a list of things it knows its customers want, but it can’t do. Many of those constraints make sense. Companies have limited resources and focus matters. But when we are honest with ourselves, we know there’s much that customers want from us that we should be able to deliver, but haven’t. AI changes that equation. It expands what every company can do and makes it possible to address a broader range of customer wants and problems.
Across industry, government, healthcare, and education, there are so many things people want but do not have today where AI can make a difference. We need to shift our awe and impatience toward people and their problems. To stand in awe of the person who manages to navigate the complexity of modern life. To be impatient to give that person back time, agency, and peace of mind.
Technologies spread when they enable everyday people to do what they could not do before. The person who couldn’t find a taxi in the rain now has options. The person who’s tried every weight-loss trick and never succeeded now has options. The person who wanted to share their story with the world but could never get the attention of the gatekeepers now has options. By narrowing AI’s focus to how companies work, rather than the customers they serve, we build for the wrong outcomes.
The capability overhang has always lived outside the company’s four walls. We should be impatient to close it.
