Recursive Human Improvement

Silicon Valley seems to have settled on a future: build machines that improve themselves until they surpass us. Then what? We upload ourselves, merge with them, become their pets, or accept that humanity was simply a stepping stone towards something greater.

I think this is fundamentally backwards. It's an anti-human vision dressed up as technological progress.

We have somehow arrived at a point where some of the brightest minds in the world are dedicating themselves to making machines smarter, stronger and more capable, while treating the improvement of human beings as an afterthought. We outsource our intelligence, our creativity, our memory, our judgement and eventually, if some people get their way, our very existence.

We have stopped asking what we could become.

I want to start a counter-movement. Its name is Recursive Human Improvement.

The proposition is simple: stop pursuing recursive self-improvement in AI. Keep artificial intelligence within boundaries that preserve human supremacy, and redirect our scientific and technological ambitions towards improving the human condition.

The machine is not our successor. It is our instrument.

Humans should remain the pinnacle of intelligence, capability and agency. Not because our current biology is perfect, but because we have barely begun to explore what humanity could become.

The Übermensch is the beginning

This isn't conservative humanism. I'm not interested in preserving humanity exactly as it is.

The current human condition is not sacred. Our biological limitations are not permanent, and our present form is not some finished evolutionary product. We should be actively working to overcome it.

Nietzsche's Übermensch is a useful starting point. The idea of self-overcoming: refusing to accept that what we are today is all we can ever be. But I don't think the Übermensch is the destination. It's the beginning.

Humanity should take responsibility for its own evolution. Not leave it to random mutation, market incentives or the emergence of artificial intelligence.

Think about the stories we tell ourselves. Superheroes, extraordinary abilities, superhuman strength, intelligence, perception and resilience. We consume these stories constantly, yet rarely seem to consider that they might represent something worth pursuing.

I don't think that fascination is accidental. Perhaps it's an expression of something deeply embedded in our collective imagination: the desire to become more than we are.

We call it fantasy because we haven't figured out how to build it yet.

Superhuman shouldn't just be a costume or a fictional archetype. It should be a research direction.

The Dune model

If I were to imagine a civilisation that had taken human development seriously, I'd look less towards Silicon Valley and more towards Frank Herbert's Dune.

Not necessarily its politics, hierarchy or social structures, but its fundamental premise: humanity can cultivate extraordinary capabilities within itself.

The Mentats develop extraordinary reasoning and computational abilities. The Bene Gesserit cultivate perception, psychological insight and control over their bodies. The Fremen develop exceptional resilience and adaptation to their environment.

Different paths. Different disciplines. Different expressions of human potential.

That's what interests me.

I don't want a world where everyone is engineered to the same genetic template or measured against one universal definition of perfection. I want a world where human beings can specialise, develop and pursue radically different forms of excellence.

We should be cultivating extraordinary mathematicians, scientists, artists, philosophers, athletes, strategists and explorers. We should be exploring the limits of memory, perception, physical capability, emotional regulation and creativity.

And we should be doing this through deliberate human development, not outsourcing those capabilities to machines.

There is also the Butlerian principle: thinking machines must not become humanity's successors. We should not surrender the development of human cognition to artificial intelligence and then congratulate ourselves for having invented something smarter than us.

Technology should extend human capability. It should not replace the human mind.

The Gom Jabbar

There's another idea in Dune that I think is just as important: the Gom Jabbar.

The test isn't about intelligence or physical strength. It's about whether someone can exercise conscious control over their instincts, even under extreme pressure.

That's a useful distinction when thinking about human improvement.

We could dramatically increase intelligence, strength, memory and longevity, but what would be the point if we didn't also develop judgement, discipline and self-mastery?

I don't want to create superior biological machines. I want to develop exceptional human beings.

That means emotional regulation, foresight, personal responsibility, resilience, wisdom and the ability to act deliberately rather than simply react.

Capability without character is a weapon looking for a hand.

Human advancement must encompass the whole person.

Every individual

This isn't about creating a genetically superior elite.

Every individual should have access to the benefits of human advancement. Not just the wealthy, the military, exceptional scientists or those fortunate enough to win some genetic lottery.

If genetic engineering can eliminate inherited disease, improve cognitive capabilities or extend healthy human lifespans, those breakthroughs should benefit humanity broadly.

I embrace genetic engineering, regenerative medicine, neural interfaces and whatever other technologies might allow us to overcome our biological limitations.

But universal enhancement doesn't mean uniform enhancement.

I don't want compulsory genetic templates, centrally prescribed definitions of perfection or institutions that own and control human evolution. Different people should be free to pursue different forms of excellence.

Some might pursue extraordinary intellectual capabilities. Others might pursue physical mastery, artistic expression, sensory development or entirely different possibilities we haven't even imagined.

The objective is to expand individual potential, not constrain it.

No permanent biological aristocracy. No enhancement monopoly. No institution that gets to decide what a human being is allowed to become.

A civilisation that improves humanity while concentrating those improvements in the hands of a few has missed the point.

Don't counterfeit the human

There's something deeply unsettling about machines designed to look and behave like people.

The uncanny valley is interesting because we seem to have an instinctive reaction to something that looks almost human but isn't. I don't think we should be so eager to engineer that reaction away.

Why are we so interested in creating artificial people?

Why are we investing in synthetic companions, artificial personalities and machines that imitate human relationships, when we could be investing in making actual human beings healthier, happier, more capable and more connected?

I don't want a counterfeit human experience. I want a better human experience.

We should be careful about normalising a future where artificial entities occupy the social, intellectual and emotional spaces that belong to people.

The goal should be to improve human relationships, not manufacture substitutes for them.

The hard boundary

My position on artificial intelligence is straightforward.

No artificial intelligence should become smarter than a human being. No autonomous machine should become stronger, faster or more capable than the humans it is supposed to serve. No artificial entity should become an independently superior intelligence or an alternative centre of power.

I'm not talking about banning calculators, vehicles or industrial machinery. Specialised tools can exceed human capabilities in particular tasks. A car can travel faster than a runner. A crane can lift more than a person. A computer can perform calculations faster than we can.

That's not the problem.

The problem is the ambition to create artificial entities that surpass us in general intelligence, agency and consequential capability.

Recursive self-improvement in AI is the most obvious expression of that ambition. Build a system that can improve itself, use those improvements to build a better system, and repeat until humans are no longer capable of understanding or controlling what they've created.

Why?

What exactly are we trying to achieve?

If the answer is to create something that eventually makes humanity irrelevant, then I reject the premise entirely.

AI should be developed only to the extent that it helps us advance human capabilities. It can accelerate research, help us understand biology, discover medicines, develop new materials and improve education. It can help us solve problems that have constrained humanity for centuries.

But the objective must always be human advancement.

We should not be building our successors. We should be using these tools to become something greater ourselves.

Reverse the recursion

Consider the feedback loop we're currently pursuing:

AI improves itself → AI becomes more capable → AI improves itself again → humans become increasingly dependent on AI.

I want to reverse it.

Tool improves human capability → human becomes more capable → human develops better tools → human becomes more capable again.

That's Recursive Human Improvement.

The distinction is fundamental. In one future, machines become increasingly capable while humans become increasingly dependent on them. In the other, technological breakthroughs compound into improvements in the human condition.

AI could help us understand ageing, develop regenerative therapies, improve memory, investigate consciousness, accelerate genetic research and unlock entirely new areas of science.

Imagine applying the same ambition and investment currently directed towards artificial intelligence to understanding and improving the human brain.

Imagine what we could discover if we treated human biology as the most important engineering problem civilisation has ever undertaken.

The machine is the spark that reignites human evolutionary progress. It is the forge, not the finished product.

We are the metal, and the smith.

Progress is not inherently good

There's an assumption in Silicon Valley that technological acceleration is inherently virtuous. That the next breakthrough is always desirable, that faster is always better, and that stopping is equivalent to failure.

I don't accept that.

Human flourishing takes precedence over technological acceleration. Progress is only progress if it serves humanity.

If the choice is between halting technological development and creating a future where humanity is displaced by artificial intelligence, I'd rather stop.

That shouldn't be a particularly controversial proposition. We don't have an obligation to build something simply because we can.

We should be ambitious about what humanity can become, but equally willing to draw boundaries around what we should create.

The movement

Recursive Human Improvement is both a refusal and a programme.

We refuse:

• Recursive self-improvement in artificial intelligence.
• The creation of artificial successor species.
• The pursuit of artificial general intelligence that surpasses human intelligence.
• The replacement of human relationships with counterfeit human experiences.
• The gradual outsourcing of human judgement, cognition and agency.
• The celebration of deskilling and dependency as technological convenience.
• The concentration of human enhancement in permanent biological elites.
• The assumption that technological acceleration is inherently good.

We build:

• Medicine and biotechnology that overcome our biological limitations.
• Genetic engineering that expands human potential.
• Education and training that develop extraordinary human capabilities.
• Research into memory, cognition, perception, consciousness and longevity.
• Technologies that augment human agency rather than replace it.
• A civilisation that cultivates diverse forms of human excellence.
• Universal access to the opportunities created by human advancement.

I'm not anti-technology. I'm anti-abdication.

I'm not against artificial intelligence. I'm against the idea that artificial intelligence should become the next stage of evolution.

I'm not interested in preserving humanity as it currently exists. I'm interested in what we could become if we actually tried.

We have spent decades imagining a future where machines become superhuman.

It's time we started asking why we aren't trying to become superhuman ourselves.

Cap the machine. Raise the human.

The machine is not our successor.

Humanity is its own next stage of evolution.