Just Too Late

[The following piece is written by retired Professor John Gray.  Given the recent call by the leaders of the artificial intelligence industry for government to please help them control themselves and its development, it is extremely timely.—Robert C. Hinkley]

Photo by Igor Omileaev

By John Gray*

16 September 2026

Many of the most serious public problems associated with business seem to follow a similar pattern. A company develops a successful product, technology, or business model. Investment flows in. The activity expands and becomes economically important. Only later do broader social consequences become visible. By then large investments have been made, organisational routines have become established, and change becomes increasingly difficult.

Climate change, tobacco, aspects of financial innovation, social media, and now artificial intelligence all exhibit elements of this pattern. The problem is not simply that governments fail to act. Rather, important decisions are often made before governance systems fully understand their significance. Market structures become established first, and governance follows later.

This is increasingly important because modern democratic capitalism depends upon maintaining a workable relationship between economic power and public accountability. Markets generate innovation, wealth, and opportunity. Democratic institutions provide legitimacy, accountability, and mechanisms through which societies can influence decisions with significant public consequences.

Difficulties arise when economic power becomes increasingly concentrated while governance remains reactive. Corporations are not merely participants in markets. Large firms increasingly shape the environments within which markets operate. They influence information systems, technological standards, participation opportunities, patterns of investment, and increasingly the conditions under which citizens and organisations conduct their social and economic lives.

The result is not simply economic inequality. It is a growing inequality of influence. Decisions capable of affecting millions of people are often made inside organisations whose objectives, incentives, and governance structures are only indirectly connected to democratic oversight. Citizens experience the consequences of those decisions but frequently possess limited capacity to influence them.

Part of the explanation lies in the way institutions develop. Businesses make decisions continuously. They decide how technologies are designed, how markets operate, who participates, what information is available, and which forms of behaviour are encouraged or discouraged. Over time these decisions become embedded in products, systems, infrastructure, habits, and ultimately institutions. What begins as a business decision can gradually become part of the environment within which everyone else must operate.

Once this occurs, change becomes difficult. Investments have been made. Expectations have formed. Dependencies develop. Organisations adapt. Consumers adapt. Governments eventually respond, but they are responding to realities that have already become established.

This is a problem of governance lag.

Artificial intelligence provides a contemporary illustration of the issue. The significance of AI is not simply that it is a powerful technology. Its importance lies in the speed with which it can influence communication, decision-making, participation, and economic activity. Processes that once unfolded over decades may now occur within years or even months. Institutional formation accelerates while governance often remains comparatively slow.

Once governance lag is recognised, attention naturally shifts to the question of where governance should operate.

Most current approaches focus on external regulation. Governments develop laws. Regulators establish standards. Organisations produce reports. Risk management systems are introduced. Disclosure obligations are expanded. These measures may all have value, but they generally take effect after key decisions have already been made.

The Code for Corporate Citizenship, originated by Robert Hinkley of this blog, points toward a different approach.

Its central idea is simple. Directors should not pursue corporate advantage through decisions that they reasonably foresee will cause severe harm to important public interests.

The significance of this principle is not that it offers a complete answer to every governance problem. Rather, it attempts to move governance closer to the point at which consequential decisions are made.

Instead of asking how society should respond after harmful outcomes become visible, it asks what responsibilities directors should recognise while decisions are still being formed.

The Code does not require directors to maximise social welfare. It does not require them to act as regulators. It does not require them to balance every competing interest in society. Instead, it introduces a boundary condition. Some decisions should not be pursued, even when advantageous to the corporation, where severe harm to the public interest is reasonably foreseeable.

The broader issue extends well beyond artificial intelligence and beyond any single industry. If democratic capitalism is to retain legitimacy, governance must remain capable of influencing the exercise of corporate power before harmful institutional arrangements become entrenched. The Code for Corporate Citizenship is best understood as one illustration of how this might occur.

The central question is therefore not simply how society regulates harmful outcomes. It is whether governance can influence institution-forming decisions before they become institutional realities. The Code offers one possible answer to that challenge.

* John Gray is a retired Professor of Economic Sociology who has spent much of his academic career studying how institutions emerge, evolve, and shape social life. His interests span governance, ethics, organisational change, and the relationship between private power and public accountability.

In recent years his attention has turned to the governance challenges posed by social media, artificial intelligence, and other rapidly developing technologies. His contribution to this issue reflects a continuing interest in how societies can respond when institutional change outpaces the mechanisms designed to govern it.