The Choice AI Is Asking Every Business to Make

In October 2024, Dario Amodei wrote Machines of Loving Grace. It described a world where AI compressed a century of progress in health, economics, and human welfare into a decade. Where personalised drug discovery became cheap enough for broad deployment. Where the developing world caught up with the developed. Where GDP growth in the poorest countries reached rates that would have seemed impossible a generation earlier. In his words, anything less than that broad distribution of gains would be a terrible moral failure.

In January 2026, he wrote The Adolescence of Technology. Same author, same organisation, same belief in the capability. A fundamentally different subject. This essay mapped the risks: economic disruption, concentrated power, the possibility that the most important technological shift of our lifetime widens the gap between those who capture its value and those who absorb its cost. He called it a rite of passage. Turbulent and inevitable.

Two possible outcomes of the same technology. Not a utopian dream and a dystopian nightmare. Something more honest: a choice. And a choice that is shaped, above all, by how the technology is adopted.

This pattern has a name in economic history. During the Industrial Revolution, Robert C. Allen documented a fifty-year period when output per worker climbed while wages stayed flat. He called it Engels' pause. The surplus created by new technology flowed to the owners of capital. Labour, despite producing more than ever, saw almost none of it. The pause ended when capital accumulation caught up, when skills evolved, when institutions intervened. But the correction took two generations.

AI carries the same structural risk. The technology creates surplus. Whether that surplus is shared or extracted depends on the decisions made around it.

Technology has always been an equaliser of sorts. Clean water, antibiotics, the mobile phone. Each lifted the floor. But the gap between the best available version of progress and the most basic remains vast. The frontier of healthcare and the baseline of healthcare are not the same thing, even though both exist because of the same underlying science. Every major technological shift has the power to narrow that distance or stretch it further.

AI is unique because it is not just a tool that performs a task. It is, for the first time, an intelligence that can be directed at the problem of distribution itself. Amodei's Machines of Loving Grace makes this case with specificity: epidemiological modelling that outperforms human planners, a second Green Revolution in agriculture, health outcomes in the developing world reaching developed-world levels in a decade rather than a century.

None of that is automatic. All of it depends on adoption. By governments and research institutions, yes. But also by the businesses that make up the economic fabric of every country. Because it is inside businesses that the translation from capability to value actually happens. A new treatment is developed by a lab. It reaches a patient because a supply chain works. Prices fall because manufacturing adapts. Employment holds because firms redesigned around the technology rather than simply cutting headcount.

When businesses adopt AI without understanding, capital flows upward and disruption flows downward. When they adopt it with honest economics, genuine measurement, and their people included in the redesign, they contribute to the outcome where the gains spread.

We built this practice because we believe getting the AI decision right at the level of individual businesses is one of the most useful things an advisory firm can do. Not because any single engagement tips the balance. Because they accumulate. Because the version of the future where AI serves a broader purpose is built from millions of decisions that got it right.

That is what we are working toward. One business, one honest study, one set of numbers that moved, at a time.