An Open Letter About What We Are Doing With Intelligence

I am a software engineer. I use AI. I build with technology. I am fascinated by what we are capable of creating, and I have spent much of my life believing that better tools can make people more capable, more creative and more free.

That is precisely why I am worried.

Something has changed in the conversation around artificial intelligence.

We used to talk about computers as tools that would help us think. Now increasingly, we are being asked to imagine a world where the machine thinks for us, works for us, decides for us and eventually makes human labour itself obsolete.

And somehow we are expected to regard this as automatically good news.

I don't.

Not because I am anti-technology. Quite the opposite. I think technology may be one of humanity's greatest instruments for solving problems that have tortured us for centuries.

But an instrument is only as good as the purpose to which we put it.

The question is not whether AI will replace work

The question is who will own the machines that replace it.

This distinction is being dangerously blurred.

The International Labour Organization estimates that roughly one in four workers globally are in occupations with some exposure to generative AI, while finding that transformation of jobs is currently more likely than wholesale redundancy. That is an important distinction.

But we should not confuse "most jobs will be transformed" with "everything will be fine."

The distribution of the gains matters.

Research from the IMF finds that AI could produce productivity gains while simultaneously increasing wealth inequality because the people who already own capital are better positioned to capture increased returns from it. In one model, the adoption of AI substantially increased wealth inequality even where wage inequality could fall.

That should terrify us more than the question of whether a chatbot can write a better email.

Because if intelligence becomes extraordinarily cheap while ownership of the infrastructure that produces that intelligence remains extraordinarily concentrated, we have not democratised intelligence.

We have centralised it.

And that is the part of this conversation I find most uncomfortable.

"Don't worry. Work will change."

I keep hearing variations of this argument.

Work has always changed.

New industries will appear.

People will find new things to do.

AI will create abundance.

And if it doesn't, perhaps we will simply give everyone a universal basic income.

There is nothing inherently wrong with universal basic income. If a society becomes vastly more productive, there is a perfectly reasonable argument that everyone should share in that abundance.

But I object to the casualness with which this is sometimes presented.

There is something profoundly strange about saying:

"We may eliminate your economic role, but don't worry. We'll give you an allowance."

That is not automatically liberation.

It can also become dependency.

Sam Altman himself wrote years ago that technology can push wealth toward fewer people and that, if large classes of jobs disappear, society may eventually need some form of basic income. More recently, he has argued that a future in which AI performs much of today's work may require new forms of taxation and ownership so that people can participate in the upside.

That second part is crucial.

Ownership.

If labour loses bargaining power while capital owns the machines, compute, models, data centres and platforms, then simply distributing money after the fact does not necessarily solve the underlying concentration of power.

It can make it more comfortable.

It does not necessarily make it more democratic.

A citizen who receives a cheque from an economic system they have no meaningful influence over is still dependent on that system.

And if a handful of companies control the infrastructure through which intelligence itself is produced, then we should be asking a much more uncomfortable question:

Who controls the people who control the machines?

The lobbying should concern us

This is not paranoia.

There is already a substantial political apparatus forming around AI.

Public Citizen's analysis of U.S. House lobbying disclosures found that 3,570 lobbyists reported working on AI-related issues in 2025, approximately 26% of all federal lobbyists. It found that AI lobbying had grown by 168% in three years, and that 82% of the AI lobbyists it identified represented corporations or corporate trade groups.

That does not prove that every piece of AI lobbying is malicious.

It does prove that powerful economic interests are investing heavily in shaping the rules governing this technology.

Of course they are.

They have billions, sometimes hundreds of billions, of reasons to do so.

That is why I don't think we can leave the future of human labour, human intelligence and human agency exclusively to the companies building AI.

They have an important voice.

They cannot be the only voice.

The public should have a voice.

Workers should have a voice.

Teachers should have a voice.

Scientists should have a voice.

Artists should have a voice.

Doctors should have a voice.

Children should have a voice.

And ordinary people who will never own a data centre should have a voice.

And then there is consciousness

Here the conversation becomes even stranger.

Anthropic has openly established a research programme investigating whether advanced AI systems might eventually have consciousness, experiences or welfare deserving moral consideration. It explicitly says that there is currently no scientific consensus that today's or future AI systems are conscious.

In 2026, Anthropic also announced that it had been holding dialogues with scholars, clergy, philosophers and ethicists from more than 15 religious and cross-cultural traditions to discuss questions raised by AI.

I think these questions are legitimate.

I also think the timing is extraordinary.

We are living through a period in which millions of actual human beings are worried about losing their livelihoods, their bargaining power and perhaps their place in an economy increasingly organised around machines.

At the same time, some of the most powerful companies on Earth are beginning to seriously investigate whether the machines themselves might one day deserve moral consideration.

That is an intellectually fascinating question.

But it should not become a moral distraction.

Before we ask whether the machine deserves rights, we should make absolutely certain that we have not forgotten the human being sitting in front of it.

I have no problem with philosophers investigating machine consciousness. I have no problem with theologians being invited into that discussion. I have no problem with scientists taking the possibility seriously.

What worries me is the possibility of building a cultural framework in which we become increasingly fascinated by the moral status of artificial minds while becoming increasingly indifferent to the material conditions of actual people.

That would be an extraordinary failure of priorities.

And to be clear, I have found no credible evidence that Anthropic is lobbying against Pope Leo XIV's theology. What I have found is something more complicated and, in my view, more interesting: technology companies are increasingly engaging religious traditions because AI is forcing questions about intelligence, personhood, consciousness, agency and human dignity into the mainstream.

You don't need to believe in God to recognise the dignity of a person

Pope Leo XIV has been unusually direct about AI.

His 2026 encyclical argues that technological systems must remain subordinate to human intelligence, conscience and freedom. It warns that automated decisions can strip people of responsibility and empathy, particularly when algorithms determine employment, credit, public services or reputation.

I don't accept the theological framework surrounding those arguments.

But I don't think you need to be Catholic to understand the underlying warning.

A system can be extraordinarily intelligent and still have no experience of what it means to be a human being.

It can optimise a metric without understanding why the metric matters.

It can predict distress without experiencing suffering.

It can generate an apology without having anything to apologise for.

It can simulate compassion without needing anyone to love it.

That distinction matters.

The Pope's language is religious. My own argument is not.

But we arrive at a similar question:

What happens when we become so impressed by intelligence that we forget the value of the person who possesses it?

Technology should not make human knowledge unnecessary

This is where I draw my line.

I want better technology.

I want AI that helps doctors diagnose disease.

I want AI that accelerates drug discovery.

I want systems that help farmers predict drought, optimise irrigation and produce more food with less land and water.

I want better climate models.

I want tools that help researchers understand diseases we currently cannot cure.

I want education that gives a child in a poor country access to the world's accumulated knowledge.

I want engineers to be able to explore possibilities that would previously have required armies of specialists.

I want technology to help us solve problems that are currently considered too large, too expensive or too complicated.

These aren't fantasies.

The institutions working on food security and healthcare are already pointing to these possibilities. FAO has explicitly identified AI, satellite data and digital technologies as tools for improving early warning and responding to food insecurity, while also warning that technology must be designed around farmers and that billions of people remain offline.

WHO similarly identifies opportunities for AI to improve health outcomes and care delivery while emphasising governance, safety, responsibility and human-centred implementation.

That is the future I want.

Not a machine that writes another marketing email slightly faster.

Not an algorithm that removes another junior employee so a shareholder can capture the salary.

Not an economy where the primary achievement of AI is allowing one person with a billion-dollar infrastructure stack to replace ten thousand people.

We have bigger problems.

People are hungry.

People die from diseases we should be able to treat.

Children still grow up without access to education.

People live without clean water.

Millions of people remain excluded from the digital economy.

I cannot look at those problems and conclude that humanity's greatest achievement should be making already powerful organisations even more powerful.

Intelligence should multiply humanity, not diminish it

There is a subtle difference between using AI to extend human intelligence and using AI to replace human intelligence.

The first is civilisation.

The second can become intellectual atrophy.

If every difficult question is answered for us, what happens to our ability to ask difficult questions?

If every piece of writing is generated for us, what happens to our ability to think through an argument?

If every decision is optimised for us, what happens to judgement?

If every junior worker is replaced before they have the opportunity to become a senior worker, where does the next generation of expertise come from?

This last question is particularly important.

Expertise does not appear from nowhere.

A senior engineer was once a junior engineer.

A surgeon was once a medical student.

A researcher was once an undergraduate asking stupid questions.

A writer was once someone writing badly.

Human knowledge is transmitted through participation.

If we automate away the entry points through which people acquire expertise, we may discover that replacing labour is easier than replacing the knowledge pipeline that labour creates.

The ILO's current research is important precisely because it does not reduce this to a simple "jobs disappear" story. Most occupations are collections of tasks, and generative AI will change the composition of those tasks in complicated ways.

That means the goal should not be to maximise the number of tasks performed by machines.

The goal should be to maximise what humans are capable of achieving with machines.

Those are very different objectives.

We should be suspicious of abundance without distribution

AI may create extraordinary abundance.

I hope it does.

But abundance does not distribute itself.

The printing press created extraordinary intellectual abundance, but access to education remained deeply unequal for centuries.

The industrial revolution created extraordinary productive capacity, but also produced brutal working conditions and enormous concentrations of wealth.

The internet made information abundant, but it also produced some of the most powerful companies in history.

Technology does not determine the social outcome by itself.

Institutions do.

Ownership does.

Regulation does.

Culture does.

Politics does.

And, ultimately, people do.

The IMF's recent work makes this point uncomfortable but clearly: AI can generate productivity gains while also increasing the returns to capital and therefore wealth inequality.

So when somebody tells me that AI will create enormous wealth, my response is:

For whom?

When they tell me that productivity will increase, I ask:

Who captures the productivity gain?

When they tell me that labour will become unnecessary, I ask:

Who owns the machines?

And when they tell me that everyone will receive an income, I ask:

Who gets to decide how much?

Those aren't anti-technology questions.

They are the questions required to make technology work for society.

I don't want less innovation

I want more of it.

I want us to stop wasting our extraordinary technological capabilities on the smallest possible definition of progress.

We have built machines capable of processing astonishing quantities of information.

Use them.

Use them to discover new medicines.

Use them to model ecosystems.

Use them to improve food production.

Use them to find materials that make energy cheaper.

Use them to teach.

Use them to translate knowledge between languages.

Use them to accelerate scientific research.

Use them to help a doctor in a remote village.

Use them to help a farmer facing drought.

Use them to help humanity understand problems that currently overwhelm individual minds.

And when the technology makes somebody's work easier, let that person keep some of the benefit.

Give them more time.

More autonomy.

More education.

More security.

More opportunity to create.

Do not automatically turn every efficiency gain into another headcount reduction.

That is not a law of nature.

It is a business decision.

We need a different definition of progress

I am not asking for a return to a pre-digital world.

I don't want that world.

I like computers.

I like the internet.

I like automation.

I like AI.

I like building things.

I like seeing a machine do something that yesterday seemed impossible.

But I refuse to accept the idea that the highest purpose of technology is to reduce the cost of human beings.

Technology should reduce suffering.

Technology should expand human capability.

Technology should make knowledge more accessible.

Technology should give people more agency over their lives.

Technology should help us feed people, heal people, educate people and protect the planet.

And if it makes us richer, then that wealth should be used to make society more capable, not simply to make the owners of the machines richer.

We are at an unusual moment in history.

For the first time, we are building tools that can operate in the territory we once considered uniquely human: language, reasoning, creativity, research and knowledge.

That should not make us want to become less human.

It should make us more determined to understand what being human is worth.

I don't know whether machines will eventually become conscious.

I don't know whether AI will eliminate ten percent of jobs, fifty percent, or whether entirely new categories of work will emerge that make today's predictions look ridiculous.

Neither does anyone else with certainty.

The ILO explicitly cautions that the future cannot be predicted with precision, and even the companies building these systems acknowledge enormous uncertainty.

But we don't need certainty to make choices.

We can choose what we want technology to be for.

I choose abundance over extraction.

I choose empowerment over dependency.

I choose human knowledge over intellectual surrender.

I choose machines that extend human capability rather than machines whose primary purpose is to make humans economically disposable.

And I choose to believe that the most impressive thing we can do with artificial intelligence is not build something that makes people unnecessary.

It is build something that makes people more capable of solving the problems that only humanity has the responsibility to solve.

That is the future of technology I want.

And I think it is worth fighting for.

Sources

#ai #futureofwork #techpolicy