Some thoughts on AI

Just as the laptop made computation widely personal, generative AI is making certain forms of intelligence radically cheap and widely accessible. This will make software far more abundant and deeply customised to individual users. But it will also erode skill barriers, intensify dependence, increase pressure for surveillance, and shift advantage away from those who can merely execute toward those who can judge, direct and verify.

Generative AI does not simply automate answers. It compresses the distance between observation and decision, and ultimately between human desire and its fulfilment.

Historically, technology has travelled through a long chain before becoming commercially accessible at scale. Research produces new knowledge. That knowledge is translated into a proof of concept. Companies turn the proof of concept into a product, build the infrastructure required to deploy it and distribute it at scale. Only then does it reach the individual user.

Every stage of this chain has acted as an inhibitor. Research requires years of specialised study and a depth of understanding sufficient to push the frontier of human knowledge, even incrementally. Commercialisation requires capital, technical capability and organizational coordination. Deployment requires infrastructure, distribution and operational execution. Usage itself requires the user to possess the time, money or expertise necessary to benefit from the technology.

Generative AI is beginning to weaken these barriers at every stage.

Today, it allows individuals to write, analyse, design and build software without mastering each underlying discipline. Tomorrow, it may help originate hypotheses, design experiments, interpret their results and translate new discoveries directly into usable products. If AI eventually becomes capable of producing fundamentally novel knowledge, pushing forward what we understand about ourselves and the universe, innovation may no longer depend primarily on a small number of exceptionally trained people working through large institutions.

In that sense, AI could democratize not merely access to knowledge, but the production and application of knowledge itself.

The consequences would extend far beyond productivity. Much of modern society is organised around scarcity of capability. Education certifies it. Professions protect it. Companies coordinate and scale it. Labour markets price it. Status is often awarded according to its perceived difficulty. If intelligence, expertise and execution become abundant, many of the institutions built to organize scarce human capability may lose part of their economic and social function.

This does not necessarily mean that human beings become irrelevant. Human desires will remain distinct, diverse and often contradictory. But the chain separating desire from action may become dramatically shorter. An individual may increasingly be able to observe a problem, articulate an intention and direct an intelligent system to research, decide, design and execute a response.

That is liberating, but it is also destabilising. Skills through which people once developed identity and earned status may lose their value. Dependence on intelligent systems may weaken independent judgment. The systems that know us well enough to serve us personally may also become instruments of unprecedented surveillance and influence. Power may move away from those who possess expertise and toward those who control the models, data and infrastructure through which expertise is delivered.

As this process accelerates, the central question is no longer simply what AI will allow us to produce. It is what human life is for when production ceases to provide its organising logic.

If intelligence becomes abundant, what remains scarce? If execution becomes effortless, what makes an achievement meaningful? If our desires can be satisfied with progressively less effort, which desires should we choose to pursue? What happens to dignity, mastery and purpose when they can no longer be anchored as easily in economically useful work?

The danger is not necessarily a sudden collapse of society, but the gradual erosion of the assumptions on which society is built. We move closer each day to a world in which many of our existing measures of contribution, merit and progress no longer hold.

That makes the task ahead partly technological, but fundamentally philosophical. We must decide what should remain human even when it no longer needs to be. We must ask what is lost when effort, skill and discovery are compressed; which forms of dependence we are willing to accept; and what sources of meaning cannot be reduced to productivity.

AI may democratize innovation. It may also force us to confront a question that scarcity has allowed us to postpone: once we are capable of doing almost anything, what should we aspire to do—and what should we aspire to preserve?