September 14, 2026

Three Futures, and Why We Cannot Yet Say Which One We Are In

by Ewen Emmerson

in Insights Research

Most of what gets written about artificial intelligence and the economy falls into one of two camps. Either it is a modest technology that will, in time, add a bit to growth the way earlier waves of computing did. Or it is something far larger, a genuine transformation in how economies produce and distribute wealth. The gap between these two views is enormous, and most commentary simply picks a side.

A recent working paper, by the economists Anton Korinek, Charles I. Jones, Szymon Sacher, Tess Cotter and Peter McCrory, takes a more useful approach. Rather than arguing for one outcome, it builds a single framework that can describe several possible futures at once, and shows precisely what has to be true of each one for it to happen.

The framework works like this. A small number of measurable things, how much of the economy’s work AI can actually do, how widely it gets used once it can, how much more productive it makes the tasks it touches, and how much of that work is automated outright versus simply assisted, together determine what happens to growth, wages, and jobs. Change those inputs, and the model traces through what follows for the whole economy.

Three illustrative paths emerge. In the first, AI remains a fairly ordinary technology over the next several years, adding a modest amount to growth, similar in scale to previous waves of automation, and the labour market barely notices. In the second, AI becomes something considerably more significant, comparable in scale to the arrival of the internet, meaningfully lifting growth but also shifting a noticeable share of national income away from wages and towards the owners of capital, with real, if manageable, disruption to certain kinds of jobs. In the third, more extreme path, the change is transformative: growth accelerates to a pace with no real historical precedent, and a very large share of a particular category of work, the kind done at a desk rather than with one’s hands, is displaced faster than the economy can readily absorb it.

Here is the detail we found most striking, and most useful. The researchers also asked a large sample of ordinary people what they personally expected from AI over the next several years, then ran those individual expectations through the same model. The typical answer landed closest to that middle scenario, the “considerably more significant than a normal technology, but not fully transformative” case. In other words, most people are not dismissing this as hype, nor are they expecting economic upheaval on an unprecedented scale. They expect something real, and disruptive to some, but broadly navigable, a workplace and economic shift larger than most in living memory, but well short of a rupture.

One further, honest finding from the research is worth sharing, because it is a useful discipline for anyone trying to think clearly about investing through this period. The three paths look almost identical for the next year or so. It is only from around 2027 onwards that they properly diverge. This means the news of any given month, a strong company result here, a disappointing one there, tells us very little about which of these futures the world is actually heading towards. The temptation to read too much into any single data point is understandable, but the research itself suggests genuine patience is warranted before drawing firm conclusions.

There is also a distributional question worth being honest about, since it bears on how wealth itself is held. In every scenario except the mildest, a meaningful share of what would otherwise have gone to wages instead flows to the owners of productive capital, businesses, equipment, and the investments that fund them, rather than to the people employed by them. For a portfolio built around owning a diversified slice of productive assets, this is one of the reasons broad, long-term equity ownership has historically been a sound way to participate in economic progress, whatever form that progress takes. It is also, we think, a reason for genuine care: an economy where the gains are shared this unevenly is not, in the researchers’ own words, something growth resolves by itself. It becomes a question of policy and of time, and it is one worth watching closely, not because it changes what should be held today, but because it may shape the investment environment for years to come.

We share this not to alarm, nor to suggest we know which of these futures will unfold, because, honestly, nobody yet does. We share it because we think being genuinely informed, about the full range of plausible outcomes rather than just the loudest ones, is part of what good investment thinking looks like. Perspective, not prediction, remains our approach, and we think it applies especially well here.

To find out more about how Brighton Capital Management thinks about long-term themes like this one, please get in touch.

This article reflects our own commentary and is provided for general information only. It does not constitute investment advice or a personal recommendation, and should not be relied upon as such. The value of investments can fall as well as rise, and past performance is not a guide to future returns.