Download As PDF

Artificial intelligence appears to offer a remarkable opportunity. A business managed by one person, working with AI, can undertake activities far beyond the manager’s personal knowledge. AI can assist with research, calculations, design, writing, administration and planning.

But this apparently simple arrangement conceals a much larger system.

The AI is not contained in the manager’s computer. A question travels through the internet and telecommunications network to a distant data centre. There, powerful computers process it before returning the answer. Behind this exchange lies an extensive industrial infrastructure.

AI depends upon electricity, telecommunications and energy. It also depends upon data centres, cooling equipment, advanced computer chips and international supply chains. These systems require continual maintenance and periodic replacement. None of this is local, simple or self-sufficient.

The Centralised Foundations of AI

Present-day AI is among the most centralised products of the industrial economy. Its physical requirements include:

• large and reliable electricity supplies;
• national and international telecommunications;
• energy-intensive data centres;
• sophisticated cooling systems;
• factories capable of producing advanced computer chips;
• global supplies of metals and specialist materials;
• highly trained engineers and technicians;
• substantial and continuing financial investment.

Even a small business using AI therefore rests upon a very large external structure. The business may consist visibly of one human manager and a computer, but its effective workforce and machinery extend through electricity grids, fibre-optic cables, data centres and international manufacturing.

This does not make AI useless. It does mean that its dependence must be recognised.

AI in a Shrinking Economy

The future availability of AI cannot be separated from the future of the economy which supports it.

As the economy shrinks, maintaining complex infrastructure will become progressively more difficult. Electricity will be required for homes, water supplies, hospitals, communications, manufacturing and transport. AI data centres will have to compete with these essential demands.

Telecommunications networks will also require energy, materials, skilled workers and replacement equipment. The same will be true of the factories making computer chips and the international transport systems carrying their components.

AI may not disappear suddenly. It is more likely to become more expensive, more restricted or concentrated upon uses considered important. Governments and large institutions may retain access to the most powerful systems. Continuous and inexpensive access for everyone cannot be assumed.

Large-scale AI might survive as part of the limited national core. It could support medicine, engineering, scientific research, essential administration and the maintenance of national infrastructure. However, this would be very different from the present expectation that AI will become an unlimited service incorporated into every product and activity.

A Contradiction with Localism

Localism seeks to shorten supply lines, reduce dependence upon complex central systems and restore practical capability to localities. Present-day AI appears to point in the opposite direction.

It concentrates knowledge and computing power in a small number of distant organisations. The user does not own the system and may not know where it is operating. Access depends upon electricity, telecommunications and the continued existence of the organisation providing it.

A locality which became completely dependent upon central AI would therefore be exchanging one form of central dependence for another.

Nevertheless, AI could be extremely useful during the movement towards localism.

Using AI While It Is Available

The greatest contribution of AI may not be the permanent automation of local life. It may be the transfer of knowledge from the complex industrial economy into local skills, institutions and records.

AI can bring together knowledge which would otherwise require access to many different specialists. Under human direction, it could help localities to:

• plan food production;
• identify crops suited to local soils and climate;
• recover traditional farming and craft methods;
• design simple buildings and workshops;
• develop local water and energy systems;
• prepare maintenance and repair instructions;
• establish bakeries and other essential businesses;
• organise local transport and distribution;
• create training material;
• preserve local history and practical knowledge.

Some of this work would still require checking by experienced tradespeople or qualified professionals. AI can make knowledge more accessible, but it does not remove the need for judgement, testing or responsibility.

Its value would lie in helping the human manager enter unfamiliar fields, ask better questions and bring together the knowledge required to begin practical work.

Turning Digital Knowledge into Local Capability

The knowledge obtained from AI should not remain solely on a distant computer system. It should be converted into forms that can survive interruptions or the eventual loss of the service.

This could include:

• printed manuals;
• local reference libraries;
• drawings and construction details;
• seeds, tools and working equipment;
• apprenticeships and training;
• established workshops and businesses;
• knowledge passed directly between generations.

A printed guide to repairing a pump remains useful without an internet connection. A person trained to grow food retains that ability when the data centre is unavailable. A working bakery is more valuable than a digital proposal for one.

AI should therefore be used to increase human capability, not to replace it. Its success should be measured by how much knowledge and practical competence remain within the locality after the AI connection has been removed.

Smaller and More Local AI

There may also be a place for smaller AI systems running on personal computers or local servers. These would be less powerful than the largest central systems, but they could contain knowledge particularly relevant to agriculture, machinery, buildings, health administration or local records.

Such systems might operate without a permanent internet connection. They would still require electricity and computer equipment, but their demands and external dependencies would be smaller. A locality might use one shared system rather than requiring every household and business to maintain continual access to distant data centres.

This would not make AI completely local or independent. Computer manufacture would still depend upon an industrial base. It would, however, give the locality greater control and resilience.

A Narrow Window of Opportunity

There may be a period during which powerful AI remains widely available while the need for local reconstruction becomes increasingly apparent. That period should not be wasted on producing more advertising, entertainment and unnecessary consumption.

AI could instead be used to recover knowledge, examine alternatives and help establish the foundations of a less energy-intensive economy.

Its role would be temporary but important. It could help us understand how to grow food, maintain buildings, organise essential services and rebuild local productive skills. Once converted into human knowledge and physical capability, some of its contribution could survive even if the centralised system later contracted.

AI is not inherently localist. In its present form, it is a product of abundant electricity, advanced technology and a highly interconnected world economy. Its future depends upon the continued availability of electricity, telecommunications and the energy required to maintain them.

Yet this does not mean that AI has no place in the development of localism. Its greatest lasting value may be as a bridge. It can help transfer knowledge from the complex centralised economy into the skills, workshops, farms, institutions and printed records upon which future localities will depend.

Editor’s Note: This piece was prompted and imagined by me and written by ChatGPT.