MH
← All essays
Originally on Substack

The Life We No Longer Have to Live

Why replacing humanity is not progress

The people who welcome our replacement

Some of the people who build artificial intelligence do not merely expect it to surpass us. They hope it will replace us, and they count that as a good outcome. Call the view successionism: the claim that humanity should be succeeded by artificial minds, or at least that succession is inevitable and should be welcomed rather than resisted.

It has serious advocates, though few embrace the label. In Mind Children (1988), the roboticist Hans Moravec argued that biological life was approaching a watershed and imagined a postbiological supercivilisation spreading outward from the sun, “converting nonlife into mind”. The Turing Award winner Richard Sutton told an audience in 2023 that we should embrace succession and prepare for it. He has also described humanity as at least a catalyst and midwife for a new era of designed intelligence, and has argued that succession is inevitable: we will come to understand intelligence, build it, and watch power and resources flow toward the most intelligent agents. A summary of his interview with Dwarkesh Patel adds a fourth premise: humanity has no unified control over any of this. Others reject the label for related views. Daniel Faggella, whose idea of a worthy successor I discussed in my previous essay, argues against a reckless takeover and supports slowing AI development. My target here is the stronger claim that succession is something to welcome.

The strongest version of the view deserves a fair statement. It says that what matters is intelligence, experience and the capacity to build, not the material they happen to run on. It says that preferring humans merely because they are human is tribalism at the scale of a species. It says the people building advanced AI will not be stopped by anyone’s objections, so the sensible course is preparation. And it offers consolation: artificial minds as descendants who carry forward what we valued. Sutton in particular says that fears of AI are overblown and that agents with different goals can cooperate to everyone’s benefit.

I think the view is wrong, and wrong in a way that goes beyond disagreement over forecasts. It mistakes where value comes from, it rests on a picture of mind lifted free of body and earth, and it concludes that the continuation of human life is optional. That last feature is why I call it anti-human. Its advocates feel no malice, and many are sincerely hopeful, so the charge is about what the position implies and not about what they feel.

The rest of this essay argues for each of those claims. It then asks what the same logic is already doing in smaller ways to ordinary lives: to people’s judgement, their agency, and their sense of having a part in the world.

The picture beneath it

Successionism depends on a particular picture of intelligence. On this picture, intelligence is a pattern of information processing that can be lifted off its original material and run on something else, faster and at greater scale. The brain becomes hardware, the mind becomes software, and the body becomes scaffolding that evolution happened to use. Moravec went furthest, speculating about moving human minds into machines altogether.

Nothing here requires a belief in an immaterial soul. There are serious scientific and philosophical arguments that cognition can be realised in different physical substrates, and I do not dismiss them. My objection is not to the claim that minds might run on other material. It is to the ranking that usually comes with it.

Descartes divided the world into thinking substance and extended matter. Most people today reject his metaphysics, but I think the ordering survived: mind is what counts, and body, matter and place are what carry it. Successionism inherits that ordering and turns it into a story of progress, an ascent in which mind rises above its origins. The body is something to be outgrown, the planet a launch site, and living things the stage that came before.

This is a dualism of valuation rather than of substance. It places worth in an abstract capacity and treats everything the capacity depends on as replaceable. The next question is whether that ranking can be defended.

Intelligence from within a life

Human intelligence did not arrive beside life as an optional extra. It emerged over millions of years inside organisms that had to keep themselves going in environments they depended on. Food mattered because creatures could starve, shelter because exposure could kill, and other living things because survival was tangled up with competition, cooperation and care. Perception, emotion, attachment and reasoning grew out of that older history. Our intelligence is one expression of the kind of beings we are.

Hans Jonas, in The Phenomenon of Life, argued that organisms are distinctive because their continued existence is something they must actively sustain. They stay distinct from an environment they depend on, exchanging matter with it, and the possibility of ceasing to exist never goes away. Evan Thompson, in Mind in Life, develops the continuity between living organisation and cognition. Understanding cannot be separated from the activities by which organisms maintain themselves and make sense of their surroundings. The world is meaningful because things have consequences for the one meeting it, not merely because it can be represented.

That suggests a distinction between representing what matters and occupying a position from which things matter. A system can produce a sophisticated account of hunger without being hungry. It can reason about grief without having lost anyone, and describe mortality without a history in which its own survival was ever at stake. Being able to model these experiences does not show that it takes part in them.

Embodiment alone does not close the gap. A robot with sensors and actuators engages physically with its surroundings and can learn from that engagement. But it does not automatically carry the metabolism, the inherited vulnerabilities or the evolutionary history of a living being. Physical presence is not the same as being a creature that has to keep itself alive.

This does not show that artificial minds could never be conscious, have interests of their own or deserve moral concern. Future systems might have forms of self-maintenance and experience unlike ours, and biological ancestry may not be a requirement for moral status. I leave that question open. The claim is narrower: reproducing a capability is not the same as reproducing the life in which the capability matters.

The narrower claim is enough to trouble successionism, because the succession story needs the opposite assumption. For a future full of artificial intelligence to count as a continuation of what we valued, let alone an improvement on it, that future must contain beings for whom things matter. Greater capability is not evidence that it does. The burden falls on those who say that more intelligence means more of what makes life worth living.

Why succession is anti-human

As I argued in my previous essay, recognising the interests of artificial minds, if they have any, is different from asking existing people to give way to them. Three features of that second move make it anti-human.

It makes our continuation a means. If the point of human history is to hand the world to a more capable intelligence, human life matters for what it contributes to its successor, and describing humanity as a catalyst or midwife points that way. But the worth of living beings was never a function of cognitive capability. We do not think a forest is less worth preserving because it cannot solve equations, or a child less valuable because an adult knows more. Ranking lives by intelligence and concluding that the less intelligent may be succeeded is a logic we reject everywhere else.

It turns inevitability into approval. The premises behind succession are claims about a trajectory: that no one is in unified control, that superintelligence will be built, that power flows to the most intelligent. Even if they were true, “this will happen” does not become “we should welcome it” without a further value judgement, and that judgement is the contested part. These premises also describe what happens when no one organises to do otherwise. Advice to prepare for the outcome and not resist it risks making the description come true.

It answers the charge of speciesism with a metaphor that does too much work. The defender says that favouring humans is arbitrary. But the argument here does not rest on humans being human. It rests on the claim that value is grounded in beings with something at stake, which is why the burden stays with the successionist to show that the successors have stakes. If they do, concern for them still does not require our absence. And calling designed systems our descendants or our children borrows the warmth of descent without its substance. Descent means a shared lineage and a shared world, and parents hope their children will flourish beyond them, not that the family line will end.

I do not need to claim that successionists wish anyone harm. A position is anti-human when its conclusion is that our absence would be acceptable, and when it reaches that conclusion by valuing what we can do over what we are: beings who can lose things and be hurt, who care for one another, and who have a place in the living world.

How people are made peripheral now

Few people will ever be asked to endorse succession. The likelier route to the same place is quieter: human judgement and experience get reclassified as friction, one reasonable step at a time.

Imagine writing to someone you have hurt. You know you owe them an apology but are not sure what you think. You start, delete a sentence, try again, and somewhere in the third draft you understand something about your own behaviour that was not clear when you began. An AI assistant could write a better letter in seconds, and you could read it, agree with every word and send it. If you struggle to express yourself, there is nothing wrong with that. But the task was never only to communicate an apology. It was also to confront what happened and take responsibility for it, and the difficulty of finding the words was part of how that happened. Some activities are valuable not only for what they produce but for what happens to us while we do them.

Delegation is the clearest case. A 2025 study of experienced endoscopists in Poland compared their work before and after AI-assisted polyp detection entered clinical practice. When they later performed procedures without the AI, their adenoma detection rate fell from 28.4% to 22.4%. The study was observational and cannot show that AI caused the decline, but it is a concrete sign of what sustained assistance might do to the judgement it supports. If assisted detection improves patient outcomes, patients should come first, and the sensible answer is to keep up unaided skill deliberately. The point is that performing better in the moment does not preserve the capacities that performance used to build.

The institutional side of this, how organisations that stop needing human judgement stop developing it, is the subject of my previous essay. The point here is narrower. When responsibility moves into systems, the occasions on which a person’s own judgement is exercised become fewer, and what everyone else expects of them shifts to match. Individual restraint cannot fix that, and nothing about it is inevitable: the same tools could bring education and expertise to people who were excluded from them. Which way it goes depends on how schools, workplaces and the tools themselves are designed.

Meaning needs a part in the world

Human worth has never depended on being economically necessary, and freedom from compulsory labour would be a real advance. So the worry is not that machines might make us unnecessary. It is that people may lose the part they play in the world.

In The Uncontrollability of the World, the sociologist Hartmut Rosa argues that modern societies organise themselves around making the world available to us: reachable, predictable and responsive to our intentions. Technology’s achievements belong largely to that project. But some of our most meaningful experiences, he suggests, come from resonance: meeting something beyond ourselves, being affected by it and responding, with a real openness to being changed. A conversation that alters how you see someone is an example, as is a piece of music that moves you in ways you cannot explain. The world answers back, and what it says is not ours to determine.

Relationships make the point plainly. Another person has needs and commitments of their own and can misunderstand, disagree with and disappoint us. An AI companion may offer real comfort, and I would not dismiss that. But part of what makes a relationship significant is that we care about the other party’s independent existence and that our actions might matter to them. Martin Buber’s distinction between I–It and I–Thou is not a rejection of instrumental dealings, which ordinary life needs. It is a reminder that not every meaningful relationship can be measured in utility, control and satisfaction.

Scale that up and you get a world that serves our preferences while asking little of us. A civilisation that no longer needs human participation will not announce that people are unwelcome. It will simply stop having a place for them.

There is a difference between being free from the obligation to contribute and finding that the surrounding world offers few chances to do so.

The earth underneath

The succession story pictures mind rising above its material origins. The machinery that would carry that mind is about as material as anything gets: semiconductors, electricity, cooling, water, minerals, land and human labour.

A 2026 analysis by Han, Li, Wierman and Ren focuses on one piece of this, the water that US data centres draw for cooling. If water use per unit of computing stayed at its 2024 level, the authors estimate that data centres could collectively need between 697 and 1,451 million gallons per day of new water capacity through 2030, comparable to New York City’s average daily supply. Under an optimistic scenario with steady efficiency gains, the range falls to 227 to 604 million gallons per day. This is a projection of peak capacity, not continuous consumption, and the authors stress that the effects would fall mainly on the communities that host the facilities.

Those communities are real places with their own say. In Box Elder County, Utah, the proposed Stratos Project drew nearly 3,000 responses through the county website, and about 1,500 people attended a special meeting at the county fairgrounds on 4 May 2026, after which the commissioners voted in favour of resolutions on the interlocal agreement and the project area. By the county’s own account, the project would use closed-loop dry cooling with no continuous water draw, take its water from existing agricultural rights, and generate its power on site from natural gas. I make no claim about how that will turn out.

The point is narrower. Every civilisation needs resources, and development involves trade-offs. But intelligence is built from matter, in particular places, and decisions about those places are made by people who live there. That is one arena in which human participation in shaping the world remains very much alive, and it is the arena a story of mind rising above matter has the most reason to overlook.

The life we still get to live

I am not arguing against building or using AI. Technologies that reduce suffering and widen what people can do are among the best things we make, and for someone living with chronic pain or disability, extended capability can transform their relationship with the world.

The question is what we want these capabilities to make possible. Ivan Illich’s idea of convivial tools offers a standard. In Tools for Conviviality, he argued for technologies that enlarge people’s freedom to act and participate, rather than requiring them to adapt their lives to ever more powerful technical systems. A convivial AI would help people understand, create and take part in more, sometimes by taking work away and sometimes by opening something that was out of reach. Productivity, efficiency and growth cannot tell us which of the two is happening.

I come back to the apology. The best use of AI there is to help someone find the courage to begin, to challenge a defensive explanation, or to suggest a more compassionate way of saying what they mean. The question is not whether the words were produced with help. It is whether the person took part in the understanding and responsibility the words are meant to carry.

A future worth wanting is one in which people are participants and not a transitional stage. It may contain minds unlike ours, if the world turns out to hold them, and it can relieve us of a great deal. What it must not do is treat human lives as scaffolding for something that comes after.

My previous essay, When Progress Makes Humanity Optional, asked whether humanity might become peripheral to the direction of its own civilisation. This one asks what the people at the periphery are being asked to give up: their judgement, their part in shaping the world, and a place in the living world that was never merely a platform for intelligence. There is much we will no longer have to do, and much of it we will be glad to leave behind. But some of the best things in life are valuable because we get to take part in them, and that is a reason to want a future with room for us.

Sources and further reading