AI, Humanity and the Power to Decide
Nobel Laureate Daron Acemoglu has raised a series of questions about Artificial Intelligence.
Each goes beyond the usual discussion about how powerful AI will become.
The questions are about intelligence.
Democracy.
Investment.
Inequality.
Health.
And now, decentralisation.
These are important questions.
But perhaps they lead us to one larger question:
Who should control the extraordinary productive power that Artificial Intelligence is creating?
And an even more fundamental one:
How can that productive power be used to improve the lives of everybody?
The First Question: Is AI Really Intelligent?
Acemoglu asks whether Artificial Intelligence is moving towards superintelligence or whether much of what we call AI is actually a form of “distorted intelligence”.
He questions whether systems designed to pursue specified goals autonomously should necessarily be described as intelligent in the broader human sense.
This distinction matters.
A system can become extraordinarily capable without possessing the breadth of human judgement.
It can process enormous quantities of information.
It can discover patterns beyond human capacity.
It can generate solutions at extraordinary speed.
It can control machines and systems.
It can increasingly act autonomously.
But capability is not the same as wisdom.
And this raises a more important question.
Who controls the capability?
The danger from AI does not depend entirely on whether we eventually call it superintelligence.
A sufficiently powerful system, placed in the wrong hands, can create enormous consequences even without possessing anything resembling human wisdom.
AI could go out of control.
That possibility cannot be treated as a distant philosophical question.
It is a question of governance.
It is a question of control.
And ultimately, it is a question of who has the authority to decide what AI is allowed to do.
The Second Question: Who Gets to Decide?
Acemoglu's second question is about democracy.
AI could transform employment, productivity, inequality, science, communication, social organisation and politics.
Yet the discussion is increasingly presented as something to be left to experts, technologists and the institutions controlling the technology.
That is not enough.
The consequences will be experienced by humanity as a whole.
Nearly six billion people live outside the United States, Europe and China.
They cannot simply become spectators to decisions that will determine the future of work, production and human society.
AI governance therefore cannot become the preserve of the countries or corporations that happen to possess the greatest computing power.
Nor should it become an extension of the interests of the richest nations.
The world needs an extremely strong international mechanism for controlling the most powerful and potentially dangerous AI capabilities.
But that mechanism must itself be genuinely accountable to humanity.
It should not simply be a committee of national governments protecting national interests.
It should involve extremely responsible public leaders, operating through a genuinely democratic international framework.
The objective should be simple.
AI must remain under human control.
And humanity must remain capable of controlling it.
At the national level, governments should take responsibility for building the productive capabilities required for their own development.
AI should not become another field in which countries of the Global South simply become dependent upon technologies controlled elsewhere.
The Third Question: Who Will Control AI?
This may be the most consequential question.
Trillions of dollars of private investment are now flowing into AI.
Stijn Van Nieuwerburgh estimates that the US AI infrastructure buildout alone could require about $10.3 trillion of investment between 2025 and 2032, an average of 3.63% of US GDP each year.
The scale is difficult to comprehend.
But the more important question is not how much will be invested.
It is:
Who will control what that investment creates?
Private investment naturally seeks a return.
But the consequences of AI ownership go far beyond financial returns.
The existing economic system has already produced an extraordinary concentration of wealth.
A tiny fraction of humanity controls an enormous share of wealth, while a very large majority continues to live with unnecessary economic distress.
Consider the global food industry.
The industry has already become highly concentrated, with a handful of corporations, the ABCD, exercising enormous influence across agricultural commodities, trading and food supply chains.
Now imagine applying the same concentration to Artificial Intelligence.
AI is expected to transform production itself.
It could control increasingly large parts of manufacturing, logistics, agriculture, research, healthcare, software, communications and many other activities.
It could influence science and technology.
It could influence defence.
It could influence the way societies function.
And ultimately, it could become the infrastructure through which much of humanity's production takes place.
If that capability is controlled by a handful of corporations, the concentration of economic power could become unprecedented.
And the security implications would be equally profound.
The question would no longer be simply who owns a profitable company.
It would be who controls a technology capable of influencing much of humanity's productive and technological system.
That is a level of concentration society should not leave to private commercial interests alone.
AI Cannot Become the ABCD of Everything
There is an important distinction here.
A private company can own a food business.
It can own factories.
It can develop technology.
It can build successful AI systems.
But it should not become capable of nearly monopolising the productive foundation of entire sectors of human civilisation.
We should not allow a few private corporations to become the ABCD of everything.
China provides an important example from another direction.
China has built enormous manufacturing capacity.
Factories.
Ports.
Railways.
Power systems.
Logistics networks.
Industrial ecosystems.
Technological capability.
That capacity has transformed the global economy and has affected manufacturing in many other countries.
It has also produced the current international debate about Chinese “overcapacity”.
Whatever one's view of that debate, there is an important lesson.
A country can build productive capacity on an extraordinary scale.
Other countries are not required to remain permanently dependent upon the productive capacity of others.
They can build their own.
The question is therefore not whether such capacity can be created.
It is whether nations choose to create it.
And the same principle applies to AI.
The Stakes Are Greater Than Financial Return
The investment required for AI will be enormous.
But the quantum of investment need not itself be the obstacle.
The resources required to build AI are real.
Electricity.
Semiconductors.
Computing infrastructure.
Data centres.
Engineers.
Scientists.
Robotics.
Communications networks.
Organisational capability.
These can be developed.
The more important question is how society chooses to organise and control them.
If AI becomes a central component of future production, governments should take the lead in building and controlling this strategic productive capacity.
Private enterprise can continue to innovate.
Markets can continue to function.
Private investment can continue to play its role.
But the foundational capability on which humanity's future production may depend should not become the private property of a tiny group.
Where strategically important capabilities have already been created through private investment, governments could, where appropriate and through lawful arrangements, bring them into public ownership or public control.
The objective would not be to eliminate private enterprise.
It would be to ensure that humanity does not become economically or strategically dependent upon a handful of private owners of AI.
The Fourth Question: Can AI Investment Also Improve Public Health?
Acemoglu asks whether some of the enormous investment flowing into AI could instead be directed towards public health, particularly where it could produce major improvements in health and longevity.
There is no reason why society should have to choose between the two.
AI investment can support fundamental research into diseases, drug discovery and, potentially, the eradication of diseases that continue to impose enormous suffering on humanity.
At the same time, society can organise comprehensive healthcare so that every individual has access to the care they need.
This is where the Resource Standard takes the discussion further.
Healthcare is not simply a question of discovering new medicines.
It is also about ensuring that people actually receive healthcare.
The Resource Standard is designed to connect available resources with unmet needs.
Healthcare resources can therefore be organised to reach people where they live.
And health is influenced by much more than medical treatment.
A child whose nutrition, healthcare and education are assured has a different life trajectory.
A mother whose maternity care is assured faces a different set of risks.
An elderly person who is assured of care does not face the insecurity of being abandoned.
A person who is not living under constant economic distress has an entirely different basis for maintaining health and well-being.
The objective is therefore comprehensive.
Universal healthcare.
Universal old-age support.
Universal child support.
Universal maternity support.
And no person left destitute.
The Resource Standard treats these not simply as welfare expenditure, but as part of the productive and social capacity of society.
A healthy child is productive capacity.
A healthy population is productive capacity.
A society in which people are free from unnecessary distress is a healthier and more productive society.
AI can contribute to the scientific frontier.
The Resource Standard can organise the delivery of healthcare and the wider conditions necessary for human well-being.
There is no need to choose between them.
The Fifth Question: Can AI Be Decentralised?
Acemoglu's fifth question goes to the heart of the promise of Artificial Intelligence.
Should we accept the increasingly centralised nature of AI?
Or is there a way in which its promise can be realised while data, capabilities and control are much more decentralised, both across domains and across people?
And would decentralised AI be more consistent with human agency and control?
The promise of AI is not simply greater computing power.
It is the possibility of extraordinary productive capability.
Artificial Intelligence, together with robotics, has the potential to produce virtually all the goods and services required by a modern society while dramatically reducing the amount of human labour required.
If that potential is realised, the central economic question changes.
It is no longer primarily about increasing production.
It is about ensuring that the abundance created by technology benefits everyone.
That is the promise.
Shared prosperity.
And this is where decentralisation becomes important.
If AI capabilities are widely accessible, across domains and across people, its productive power can become much more widely distributed.
It can help people produce.
It can help communities produce.
It can help enterprises become more productive.
It can help governments organise production and public services.
AI can therefore become a powerful instrument for expanding human capability.
But there is another side to the question.
AI itself must remain under human control.
The danger is not only that the benefits of AI could become concentrated.
AI could also become increasingly autonomous and potentially uncontrollable.
Its development therefore cannot be allowed to become unwieldy.
The objective cannot simply be to develop increasingly powerful AI.
It must be to develop increasingly powerful AI while retaining the ability to control it.
If serious danger signals emerge, humanity must retain the ability to stop it.
AI must remain under human control.
And humanity must remain capable of controlling it.
The Promise of AI Does Not Depend on AI Alone
There is another question that follows naturally.
If the ultimate objective is shared prosperity, must humanity depend entirely upon Artificial Intelligence to achieve it?
Not necessarily.
The same objective can be pursued through the continuous deployment of society's real resources, with Artificial Intelligence providing additional productive capability wherever it can contribute.
AI can support State-level manufacturing.
It can help coordinate Village Operational Systems.
It can support advanced farming.
Processing.
Animal husbandry.
Education.
Healthcare delivery.
Logistics.
And many other productive and social activities.
In this way, AI does not have to become the economic foundation upon which human survival depends.
It can become an extraordinary productive capability within a broader economic system.
This distinction matters.
Because an economy that can produce and provide for its people without depending upon AI can use AI without becoming dependent upon it.
And if AI ever becomes dangerous enough to require its use to be restricted or stopped, society should still be capable of meeting the needs of its people.
The Monetary System Offers a Warning
There is a deeper parallel here.
The world underwent a fundamental monetary transition, moving away from currencies constrained by gold or foreign-exchange reserves towards the modern fiat monetary system.
That transition was gradual, layered and institutionally uneven, with important operational changes occurring long before the formal collapse of Bretton Woods in 1971.
Yet the operational reality of the non-reserves-constrained monetary regime remained poorly understood.
The institutions inherited from earlier monetary systems continued to operate, and governments continued to approach public finance as though the earlier constraints still existed.
The result was not an absence of real resources.
It was a failure to recognise and deploy the productive capacity that the new monetary regime made possible.
Productive resources remained underutilised.
Needs remained unmet.
And enormous wealth and power became concentrated in a very small part of society.
The problem was not the productive capacity of society.
It was how that productive capacity was organised and deployed.
The Resource Standard emerged from this understanding.
Its central question is:
What real resources are available, and how can they be continuously deployed to improve human life?
It treats the problem as one of organisation and deployment rather than simply financial availability.
AI Must Not Repeat the Monetary System's Failure
Artificial Intelligence now presents another enormous expansion of productive capability.
Trillions of dollars are flowing into its development.
The productive possibilities are extraordinary.
But if that development is allowed to proceed without adequate control and without ensuring that its productive capability serves society as a whole, the result could be another extraordinary concentration of wealth and power.
Only this time, the consequences could be much greater.
The monetary system's failure produced an extraordinary concentration of economic power.
AI could produce a concentration of economic power together with technological power and potentially autonomous power.
That is why Acemoglu's question about decentralisation is so important.
The issue is not simply whether AI can become more powerful.
It is whether that power can remain consistent with human agency and control.
And whether the enormous productive capability created by AI can contribute to shared prosperity rather than becoming another source of concentrated wealth and power.
A Different Route to Shared Prosperity
The Resource Standard offers another route to the same objective.
It does not require humanity to wait for Artificial Intelligence to solve the problem of poverty.
It is designed around the continuous deployment of available real resources to meet society's needs.
The institutional architecture reaches every State, every district and every village.
The State provides direction.
District administration coordinates implementation.
The Village Operational System organises local resources, local production and local consumption.
Its purpose is simple:
To ensure that every willing person can participate productively and every family can enjoy a dignified standard of living.
Artificial Intelligence can strengthen this architecture.
It can provide additional productive capability.
It can improve coordination.
It can improve productivity.
It can support State-level manufacturing, agriculture, processing, healthcare, education and other activities.
But the underlying objective remains unchanged.
Continuous productive deployment.
And the objective of that deployment remains human well-being.
AI as an Enabler, Not a Prerequisite
This is an important distinction for the future of developing countries.
A country does not have to wait for Artificial Intelligence to become fully developed before it can begin improving the lives of its people.
People.
Land.
Materials.
Skills.
Infrastructure.
Technology.
Knowledge.
Productive capacity.
These already exist in every country in different quantities and forms.
The challenge is to organise them.
The Resource Standard was developed precisely around this principle.
It is not merely a monetary framework.
It is an organising framework.
Artificial Intelligence can become one of the most powerful tools available within that framework.
But it remains a tool.
It is not the foundation of human prosperity.
The Real Promise of Artificial Intelligence
Every major technological advance has expanded humanity's productive capability.
Steam engines.
Electricity.
Mass production.
Computers.
The Internet.
Yet none of them automatically eliminated poverty.
None automatically ensured universal prosperity.
The limitation was never technology.
It was the institutions governing how society deployed that technology.
Artificial Intelligence presents exactly the same challenge.
It may become the most powerful productive technology humanity has ever created.
It could produce enormous abundance.
But production alone is not enough.
The real objective is ensuring that the abundance created by technology benefits everyone.
Technology creates productive capacity.
Institutions determine who benefits.
That is why the question of decentralisation matters.
And that is why the question of control matters.
Humanity Must Remain Independent of AI
If AI becomes safe, controllable and enormously productive, humanity should use it.
If it can multiply agricultural production, use it.
If it can transform manufacturing, use it.
If it can improve healthcare and education, use it.
If it can coordinate complex production systems, use it.
If it can help communities and enterprises become vastly more productive, use it.
But humanity should never become economically dependent upon AI to such an extent that it can no longer live without it.
The objective is not to reject Artificial Intelligence.
It is to ensure that humanity remains capable of prospering with it, without it, or after it.
If AI remains safe and controllable, it can accelerate the transition towards abundance.
If AI becomes dangerous, it must be possible to restrict it.
If the danger ever becomes so great that AI has to be put to sleep, humanity should still be capable of feeding, housing, educating and caring for its people.
The plug on AI must never become the plug on human prosperity.
That is the deeper meaning of decentralisation.
It is not simply about distributing computing power.
It is about preserving human agency.
Preserving human control.
And preserving humanity's ability to prosper independently of any single technology.
The Choice Before Humanity
Artificial Intelligence may give humanity an extraordinary opportunity.
It may allow us to produce virtually all the goods and services required by a modern society while dramatically reducing the amount of human labour required.
But technology alone does not determine the outcome.
Humanity must decide how that capability is organised.
We can allow extraordinary productive capability to become increasingly concentrated.
Or we can ensure that its benefits are widely accessible.
We can allow AI to become indispensable to human survival.
Or we can use it while maintaining an economic system capable of meeting human needs independently.
We can allow its development to become unwieldy.
Or we can ensure that increasingly powerful AI remains under human control.
The Resource Standard offers a way to pursue shared prosperity without making humanity dependent upon Artificial Intelligence.
AI can multiply productive capability.
It can accelerate production.
It can improve coordination.
It can expand what humanity is capable of achieving.
But the purpose remains unchanged.
Every willing person should have the opportunity to contribute productively.
Every human being should have access to the necessities of a dignified life.
And the productive power of technology should ultimately serve humanity.
Technology should expand human possibility.
It must never become the master of humanity.
Comments
Post a Comment