The Butlerian Jihad: The Real Upgrade Humanity Needs

Localised General Intelligence

1. The Wrong Race: Building Super-AI

Humanity has entered a technological race with an unusual objective: build machines that can perform an increasing number of tasks better, faster and more cheaply than human beings.

The race is usually presented as inevitable progress. More capable AI promises scientific discoveries, medical breakthroughs, economic growth and solutions to problems that have resisted human effort for generations. These possibilities are real and deserve serious attention.

But there is another question we rarely ask with equal urgency: what happens to humanity as the machines become more capable?

Imagine a company developing an AI system that can perform the work of thousands of professionals. The company may become enormously productive. Its customers may receive better services. Society may benefit from cheaper goods and faster innovation. Yet the same development could weaken the bargaining power of workers whose skills the system replaces.

The outcome depends on who owns the technology, how widely its benefits are distributed and whether people can find new ways to contribute economically. Progress at the level of a machine or corporation does not automatically translate into progress for everyone.

The concentration of intelligence creates an additional problem. If a small number of organizations own the most capable systems, they may acquire disproportionate influence over knowledge, production, communication and decision-making.

The danger is not necessarily that AI will decide to rule humanity. It is that human institutions may gradually organize themselves around capabilities that only a few organizations control.

We should therefore question the direction of the race itself. Is the ultimate objective to build increasingly intelligent machines, or to create a world in which human beings become increasingly capable?

These objectives can overlap, but they are not identical.

2. The Warning from Dune

Frank Herbert’s Dune offers an unforgettable fictional answer to the problem of machine dominance: the Butlerian Jihad, a war in which humanity fights against thinking machines.

The conflict belongs to the ancient history of the Dune universe. The central story takes place long after the war, in a civilization that has prohibited machines that imitate human thought. Instead, people cultivate specialized human capabilities to perform tasks once associated with computers.

Mentats become extraordinarily capable human analysts and calculators. The Bene Gesserit cultivate mental discipline, observation and control. The Spacing Guild depends on its own specialized abilities to navigate interstellar travel.

Humanity does not simply abandon complexity. It develops human capabilities to meet some of the challenges that machines might otherwise address.

The fictional history is extreme, and it should not be mistaken for a practical blueprint. Banning AI would sacrifice many potentially beneficial applications, from medical research to accessibility tools. Nor does the story establish that machine intelligence inevitably becomes hostile.

Its enduring question is more subtle: what happens when a civilization becomes dependent on intelligence that it no longer develops within its own people?

The modern equivalent of the Butlerian Jihad need not be a war against machines. It could be a change in the direction of research and investment—a decision to make human capability a central objective of technological progress.

Instead of asking only how to make machines more intelligent, we could ask how to make humanity more intelligent.

3. From Artificial Supremacy to Human Capability

Consider two different ambitions.

The first is to develop artificial intelligence that can perform more and more human tasks. The second is to develop technologies that enable people to learn faster, reason more effectively, solve harder problems and adapt to unfamiliar situations.

The first ambition focuses on what machines can do instead of people. The second focuses on what people could do that they cannot do today.

This distinction changes how we evaluate technological progress. An AI system that replaces a thousand workers may create enormous value for its owner. A technology that helps those workers acquire new skills, solve more complex problems and start their own businesses could distribute opportunity differently.

Neither outcome is guaranteed, and the two approaches need not be mutually exclusive. AI can help people become more capable, just as it can automate their work. The important question is which outcome our institutions choose to prioritize.

Human enhancement should not be reduced to producing stronger bodies or inserting electronic devices into brains. It begins with the broader objective of expanding human capabilities. Education, personalized tutoring, cognitive tools, better healthcare and technologies that compensate for disabilities can all contribute.

More ambitious possibilities may eventually emerge from neuroscience, biological engineering and brain–computer interfaces.

The guiding principle is straightforward: technology should not be evaluated solely by how much human labour it eliminates. We should also ask how much human capability it creates.

4. The Two-Times Intelligence Hypothesis

Imagine that we could substantially improve the intellectual capabilities of the entire population. Suppose, as a thought experiment, that people became twice as capable at a range of important cognitive tasks.

What would change?

A student who struggles with mathematics might learn difficult concepts more effectively. A technician might acquire new professional skills in less time. A small-business owner might understand markets, evaluate risks and develop new products more successfully. More people might be able to participate in scientific research, engineering and entrepreneurship.

The point is not that every person would become a genius, or that all cognitive abilities could be doubled using one intervention. Intelligence is multidimensional, and there is no established method for doubling general human intelligence across a population.

The thought experiment instead asks us to examine an alternative destination for scientific progress.

We spend enormous effort developing systems that can perform difficult intellectual work. What if a comparable ambition were directed towards improving the abilities of the people who must live and work in the resulting economy?

Even smaller improvements, if safe, affordable and widely available, could matter. Better learning, attention, memory, reasoning and adaptability might help people navigate a changing world.

The economic implications could be substantial, although they would depend on how these improvements translated into actual skills, productivity and opportunity.

A more capable population could generate more ideas, develop more businesses and perform more complex work. It could also participate more effectively in decisions about the technologies that shape society.

The objective would not be to make everyone identical. It would be to expand the range of what each individual can learn and accomplish.

5. Localised General Intelligence

This brings us to the central concept: Localised General Intelligence.

The phrase describes a direction for technological development in which general intellectual capability is distributed among individual human beings rather than concentrated primarily in a small number of powerful artificial systems.

Localised does not necessarily mean that all intelligence must reside inside the brain. It means that the individual becomes the primary beneficiary and locus of enhanced capability, rather than remaining merely a user dependent on a remote intelligence service.

Consider the difference between two students.

The first has access to a powerful AI tutor. Whenever a difficult problem appears, the student asks the system for an answer. The system may be extraordinarily capable, but the student’s own ability may not improve much if the interaction replaces thinking rather than developing it.

The second student uses the same technology differently. The tutor identifies gaps in understanding, challenges assumptions, provides progressively harder exercises and teaches the student to solve problems independently. Over time, the student becomes more capable even when the tutor is unavailable.

Both students have access to intelligence. Only one approach explicitly aims to develop the student’s own capabilities.

This distinction becomes more significant as technology advances. An external system can supply an answer without necessarily transferring the underlying ability to the person receiving it.

Localised General Intelligence proposes that we should care about that transfer. The measure of success is not merely whether the human–machine system can solve a problem, but whether the human being becomes more capable of solving problems.

The concept also has an economic dimension. If intellectual capability is broadly distributed, more people may be able to develop products, provide services, conduct research and compete with established organizations.

The aim is not to eliminate large companies or prevent collaboration. It is to reduce the extent to which society depends on a few institutions for its most important intellectual capabilities.

6. Chips, Nanotechnology and Organic Superhumans

Could technology eventually enhance people more directly?

Neural chips and brain–computer interfaces provide one possible route. Current systems can translate certain neural signals into commands, helping some people communicate or control external devices. Research is also exploring ways to restore particular functions lost through injury or disease.

These are meaningful achievements, but they do not amount to a general intelligence upgrade. Reading neural signals or enabling a person to control a cursor is very different from substantially improving memory, reasoning or general intelligence.

Nanotechnology suggests another possibility. Nanoscale systems are already relevant to areas such as targeted drug delivery and diagnostics. More sophisticated medical technologies might eventually monitor biological processes, deliver treatments with greater precision or help repair specific forms of tissue damage.

Autonomous nanobots capable of comprehensively upgrading the human body remain speculative. There are substantial challenges involving safety, energy, control, immune responses and biological complexity.

Biological research offers additional avenues. Better treatments for neurological diseases, improved management of age-related decline and interventions that preserve healthy brain function could help people retain their capabilities for longer. Future advances might expand those capabilities, but their extent and feasibility remain uncertain.

The phrase organic superhuman describes the long-term ambition, not an existing scientific category of achievable enhancements.

The objective would be to expand the human organism’s capabilities rather than replace it with a machine. Chips, biological interventions and nanotechnology might eventually contribute to that objective, but each would require separate evidence of safety and effectiveness.

There is also a fundamental design question. An enhancement that makes a person dependent on a proprietary device, remote software or a corporation’s permission could undermine the very independence it was intended to support.

Human enhancement must therefore include more than biological performance. It must also preserve control over the enhanced capabilities.

7. Can Human Enhancement Save Jobs?

Employment is one of the strongest reasons to investigate this alternative.

When a machine can perform a task more cheaply than a person, businesses have an incentive to automate that task. If enough tasks can be automated, entire occupations may change, and some jobs may disappear.

Making workers more capable could help them adapt. People who learn faster, reason more effectively and acquire new skills more easily may be better positioned to move into unfamiliar occupations or perform more complex work.

A capable electrician might expand into sophisticated energy systems. A teacher might develop more effective learning methods. A technician might build a business around technologies that previously required a large organization. An experienced worker might retrain for a new industry instead of being permanently displaced.

But there is no guarantee that enhanced intelligence would preserve existing jobs. A more capable worker could still be competing with a machine that performs the same task faster and at lower cost.

Nor does higher productivity automatically produce higher wages. If the economic gains flow mainly to owners of capital, workers may remain under pressure even as their own capabilities improve.

Human enhancement must therefore be accompanied by institutions that allow people to benefit from their increased productivity. Competition, access to education, opportunities to establish businesses, worker bargaining power and policies governing the distribution of economic gains all matter.

The objective should not be to freeze the labour market in its current form. It should be to prevent technological progress from making large numbers of people economically irrelevant when they could instead contribute in new and valuable ways.

A society that improves the capabilities of its population may have more options than one that treats human labour primarily as a cost to be eliminated.

8. The New Concentration of Power Problem

There is a potential contradiction at the heart of human enhancement.

If neural chips, biological treatments or advanced cognitive technologies become the means of expanding intelligence, who will own them?

Imagine that a corporation develops an effective cognitive enhancement but controls access through expensive subscriptions, proprietary software or remote authentication. Imagine further that a person’s enhanced capabilities depend on continued access to that company’s systems.

The result could be a new form of dependence. Instead of relying on a corporation for external intelligence, people might rely on it for access to their own enhanced capabilities.

Biological enhancement could also become a source of inequality if it were available only to the wealthy. And interventions that affect the brain or body raise difficult questions about informed consent, long-term safety, privacy and the right to refuse enhancement.

Consequently, decentralizing intelligence requires more than distributing devices.

It requires meaningful user control, strong privacy protections, affordable access, independent safety evaluation and, where technically feasible, interoperable systems that do not make individuals captive to one provider.

It also requires a clear distinction between enabling people to become more capable and pressuring them to enhance themselves merely to remain competitive.

The right to remain unenhanced must be preserved alongside the right to pursue safe enhancements.

Otherwise, a project intended to liberate humanity could create a new hierarchy between enhanced and unenhanced people.

9. A Different Measure of Progress

What would it mean to take Localised General Intelligence seriously as a research programme?

The first step would be to define measurable outcomes. Instead of relying on an abstract promise to double intelligence, researchers could investigate improvements in learning speed, reasoning accuracy, memory, attention, problem-solving and the ability to transfer knowledge to unfamiliar tasks.

The second step would be to investigate interventions in order of their maturity and risk. Better teaching methods, personalized learning tools and treatments for cognitive impairment are different from invasive neural interfaces or speculative biological enhancement. They should not be treated as interchangeable technologies.

The third step would be to measure distribution. Does a technology help only a small, wealthy group, or can it safely benefit people across different income levels and educational backgrounds? Does it make individuals more independent, or more dependent on its provider?

The fourth step would be to examine economic outcomes. Do enhanced capabilities help people acquire skills, create businesses, improve their work and adapt to automation? Who receives the gains, and what happens to people whose existing occupations disappear?

Finally, research should preserve human agency. The purpose of cognitive enhancement should be to expand a person’s ability to understand, question and decide—not simply to make that person more efficient at following instructions.

This agenda would not require abandoning AI research. AI could itself become an important tool for studying the brain, developing treatments, personalizing education and accelerating scientific discovery.

The distinction lies in the objective. AI would be one means of enhancing humanity, rather than the sole destination of technological progress.

10. The Butlerian Choice

The Butlerian Jihad presents an extreme fictional response to machine dominance. Its modern relevance need not lie in banning thinking machines or treating all artificial intelligence as a threat.

There is another possibility: change the direction of the technological race before society becomes excessively dependent on a small number of artificial intelligence systems.

We could continue building powerful AI while also investing seriously in human capabilities. We could pursue better education, healthier brains, improved treatments for neurological decline and, where science permits, safe forms of cognitive enhancement.

We could judge technological success not only by the intelligence of the machines we create, but by the capabilities people gain, the opportunities they can access and the degree of control they retain over their lives.

This would not eliminate economic disruption, corporate power or the difficult questions raised by AI. Nor would it guarantee that every worker keeps a job. Those challenges require technological, economic and institutional responses together.

But it would restore an objective that can easily disappear in the excitement of building ever more capable machines: the development of human beings themselves.

The future need not be a contest in which humanity builds its replacement and then hopes to remain useful. It could be a project in which technological progress expands what people can understand, create and accomplish.

The most important question is not whether we can build something more intelligent than ourselves. It is whether we can use our growing knowledge to create a world in which more human beings are capable of shaping their own futures.

Related Article

Continue with a related argument about how reliance on AI can affect human agency:

The Dehumanization Principle of AI