Imagine that artificial intelligence delivers much of what its advocates promise: Businesses become more productive. Routine administration takes minutes instead of hours. Scientists discover promising drugs and materials faster. Energy networks become easier to manage. A small company gains analytical capabilities that once required an entire department. By familiar economic measures, there would be much to celebrate.
Now widen the picture. Are people becoming more capable of shaping their lives? Are the gains reaching those who need them? Are natural systems becoming healthier? And as machines become better at writing, analyzing, remembering, and recommending, are they strengthening human capability or encouraging us to exercise less of it? These are economic questions too. They point towards what may become the next frontier of economic progression: an architecture that becomes better at generating more of what matters, by recognizing that social, economic, ecological, and technological progress are deeply intertwined.
One economy, four connected systems
We have become accustomed to dividing reality into manageable pieces. Economic ministries deal with growth. Social ministries address health, education, and inequality. Environmental agencies focus on climate, biodiversity, and pollution. Technology policy concentrates on innovation and competitiveness. The divisions make administrative sense. Life itself is less cooperative.
Take a data centre. Economically, it represents investment and productive capacity. Technologically, it provides the computing power behind AI. Socially, its significance depends on what that power enables, which jobs it changes and who benefits. Ecologically, the same facility needs electricity, water, equipment, minerals and land. One investment moves through four systems at once. The numbers make this tangible. Current energy projections suggest that the world's data centres could consume around 950 terawatt-hours of electricity annually by 2030, almost twice their 2025 consumption. That is slightly more electricity than the whole of Japan uses today. Data centres would still represent only around 3 per cent of projected global electricity demand, so the point is not that AI will consume the world's power. What matters is the speed of expansion and the pressure concentrated demand can place on particular grids.
This is happening on a planet already under strain. The 2026 Planetary Health Check finds seven of nine planetary boundaries beyond their safe operating space. In everyday language, several of the natural systems that have provided relatively stable conditions for human societies are being pushed beyond scientifically assessed limits.
AI could help. It can balance electricity networks, improve industrial efficiency, reduce waste, accelerate materials discovery and monitor forests, oceans and agriculture. Technology can increase ecological pressure and help reduce it; widen access to opportunity and concentrate economic power; relieve people of tedious work and remove experiences through which expertise develops. The effects are connected. That is why improving one part of the system while transferring costs to another resembles moving an expense between columns of a household budget and declaring ourselves richer.
Where does the productivity go?
AI makes this question urgent because productivity gains are already appearing. A large workplace experiment found that regular users of generative AI spent about two fewer hours each week on email and reduced work outside normal hours. AI, in other words, created time.
What happens to that time?
An hour saved can become another hour of production. It can also become learning, care, creativity, better customer interaction or time returned to an employee. Higher productivity can translate into lower prices, larger profits, higher wages, greater investment in social projects or better public services. Which way it goes depends on economic systems, and the human choices which they incentivice. Which links to another asset that our economic statistics capture poorly: human agency — our capacity to understand, judge, choose and act.
AI can strengthen that capacity and encourage us to exercise it less. GPS offers a familiar analogy: Navigation has become wonderfully efficient, while it is perfectly possible to travel repeatedly through a city without developing much understanding of its geography. Apply the same principle to writing, remembering, and making decisions, and the stakes become higher. In a 2026 randomized experiment AI substantially narrowed performance differences between people with different educational backgrounds while it was available. When participants later worked without AI, the gains did not translate uniformly into independent capability. How people engaged with the technology mattered. The economic challenge is consequently more sophisticated than deciding whether humans should delegate to machines.
Making progress work harder for people and planet
This is where ProSocial AI comes in. It's not about creating machines that are friendly or altruistic but developing and deploying AI with a wider understanding of value: using technological capability to improve outcomes across people, prosperity and planet rather than optimizing one measure while ignoring effects elsewhere.
The applications can be thoroughly practical. An agricultural system can help a farmer improve yields while reducing unnecessary water and fertilizer. AI can remove paperwork from a nurse's day and return time to patients. A logistics system can reduce costs and waste. A teacher can identify where a student needs support while ensuring that the student still does the thinking required to learn.
The common thread is better alignment.
This leads towards Hybrid Intelligence, a deliberate combination of natural and artificial intelligences such that their strengths complement one another. Humans contribute lived experience, values, responsibility, social understanding and judgment. Artificial systems contribute speed, scale and the ability to process quantities of information no person could reasonably manage. Both operate within natural systems on which every economy depends.
The ultimate destination is Hybrid Flourishing—a condition in which technological progress expands the capacity of people and planet to thrive, giving every living being a fair chance to fulfill its inherent potential. It grows from the inside out through human values, choices and agency, while being nurtured from the outside in by institutions, markets and technologies that support those choices. We can't abandon growth; billions of people still need better housing, healthcare, education, infrastructure and economic security. Material progress remains indispensable. The task is to ask more of it.
=> Purpose. People. Prosperity. Planet. For two centuries, economic development has become progressively better at producing more stuff, while valuing humans and nature less. Injected in the current economic architecture, AI will accelerate that "achievement". There is an alternative: creating prosperity in ways that strengthen the people, societies and living systems from which prosperity comes.
We know how to make machines more productive. The opportunity of a lifetime is making progress more productive for life.