MH
← All essays
Originally on Substack

Consuming Reality: The Religion of Progress

When thinking becomes a product, surrender starts to look like progress

Saturn: The Lord of Destiny and God of Time

In my last essay, What Are We Becoming?, I ended by asking a question: what do we actually mean by progress? It seems increasingly difficult to separate that question from the pace of artificial intelligence development. We are building systems capable of performing intellectual work that, until recently, required considerable human knowledge and expertise. The possibilities are remarkable. Yet I find myself increasingly troubled by the relationship between what these systems allow us to produce and what their use might be doing to us.

As a software engineer, I can now accomplish things that would previously have taken days or weeks. I can explore unfamiliar subjects, generate working software and investigate ideas with a speed that once seemed unimaginable. But there is a tension in this experience. The more I delegate, the less certain I sometimes feel about my own involvement in the work. I can produce more while feeling increasingly removed from the activities that once made the work meaningful.

This is happening within an economic system that rarely seems satisfied with greater productivity. Instead, each improvement creates new expectations. The work expands to fill the capabilities we’ve acquired. The philosopher Erich Fromm argued that modern society had confused having with being. We increasingly define ourselves through what we possess, accumulate and produce, rather than through the quality of our engagement with the world. Artificial intelligence threatens to extend this confusion into cognition itself. We can consume explanations without necessarily understanding them, generate arguments without thinking them through, and create things without meaningfully participating in their creation.

There is another dimension to this transformation. While our economic institutions increasingly reward the automation of human activity, certain technological philosophies imagine a future in which transcending humanity itself represents the ultimate achievement. Progress begins to acquire an almost religious significance. We are expected to accept its direction rather than collectively determine it. These developments are not necessarily driven by the same people or inspired by the same motivations. Yet they raise a common question: Are we developing technology to serve human flourishing, or are we gradually redefining human flourishing to serve technological development?

The machine that makes us busier

For more than a century, technological progress has carried the promise of liberation from labour. In 1930, John Maynard Keynes speculated that improvements in productivity might eventually make a fifteen-hour working week possible. He imagined a future in which the economic problem of meeting our basic needs would become considerably less demanding, leaving us to confront the unfamiliar challenge of how to use our freedom.

Working hours have fallen substantially in many societies since the nineteenth century. But the relationship between productivity and leisure has never been automatic. The benefits depend upon how economic gains are distributed and what institutions choose to reward. AI may be exposing this contradiction in a particularly revealing way.

In February 2026, researchers Aruna Ranganathan and Xingqi Maggie Ye published findings from an eight-month ethnographic study at a technology company employing approximately 200 people. They observed that AI adoption encouraged employees to undertake additional tasks, work faster and extend their working day. Much of this occurred without explicit instruction from management. AI made doing more feel possible. Because it was possible, people felt compelled to do it.

It was a single-company study, not evidence that everyone experiences AI this way. But it illustrates a familiar mechanism. When something becomes easier, we don’t necessarily reclaim the time. We often raise our expectations until what was once exceptional becomes normal. The philosopher Byung-Chul Han explores this tendency in The Burnout Society. He argues that contemporary society increasingly replaces external discipline with an internalised obligation to achieve. We become our own supervisors, continually attempting to optimise ourselves.

Hartmut Rosa describes a related feature of modern life: social acceleration. Technological innovation makes activities faster, yet our experience of time becomes compressed as the number of things we are expected to accomplish expands. His concept of dynamic stabilisation describes societies that must continually grow, innovate and accelerate simply to maintain their existing position. AI gives both tendencies a powerful new instrument. If I can manage five AI agents, why not ten? If software can be produced in hours rather than weeks, why shouldn’t delivery schedules accelerate? If analysis becomes inexpensive, why shouldn’t every decision be accompanied by more analysis? Every expansion of capability risks becoming another obligation to perform.

Competition intensifies this pressure. One organisation automates to gain an advantage; another follows to avoid falling behind. Eventually, both invest in increasingly sophisticated systems merely to preserve their relative positions. A Bloomberg article described this possibility as an infinite work machine. Imagine businesses deploying AI agents to negotiate against customers’ AI agents, insurers using automated systems to contest claims generated by other automated systems, and cybersecurity tools continually responding to the activities of rival automated tools.

Each participant might improve its individual position. Collectively, however, much of the additional activity could simply cancel itself out. This resembles a Red Queen race: everyone must run faster to remain where they are. There is also a parallel with the nineteenth-century economist William Stanley Jevons, who observed that greater efficiency in coal use could increase overall consumption by making energy cheaper and more widely useful. Cheaper cognition could similarly encourage a proliferation of intellectual production.

But competitive escalation introduces something more troubling than increased demand. It creates the possibility that enormous quantities of cognitive work are generated primarily in response to other cognitive work. An economy can become more active without making human life proportionately better. Of course, these technologies can also deliver genuine benefits. Better medical diagnosis, stronger cybersecurity, faster scientific discovery and wider access to expertise are valuable achievements. The issue is the assumption that greater production automatically constitutes progress.

Productivity gains could become shorter working hours, greater economic security and more time for meaningful activities. They could also become intensified workloads, escalating competition and larger returns to those who own the infrastructure. Technology alone does not determine which outcome we get. Perhaps the most troubling consequence concerns how our relationship with the activities themselves begins to change.

When thinking becomes consumption

Knowledge, creativity and judgement have long possessed economic value: people have been paid to analyse, design, calculate and solve problems. What generative AI changes is the ease with which we can acquire the products of intellectual activity without necessarily undertaking the activity ourselves. An explanation, software application, legal argument or strategy can increasingly be generated on demand. The person receiving it may possess little of the understanding that would previously have been necessary to produce it.

In this sense, AI extends the logic of what some political economists call cognitive capitalism: an economy in which knowledge, information and intellectual activity become increasingly central to value creation. Franco “Bifo” Berardi explored the consequences of this transformation in The Soul at Work. Where industrial capitalism principally organised physical labour, the information economy increasingly mobilises language, attention, creativity and emotion. AI adds another dimension. Even the production of many intellectual outputs can be delegated to technical systems. Cognition becomes something we consume as a service.

The watermarking of AI-generated content adds another dimension to this relationship. There are good reasons to identify synthetic media, particularly where it could be used to mislead. Yet a mark can also tell a particular story about how something came into being: an output may be identified through the system that produced it, while the human thought, choices and revisions involved are less visible. A watermark does not give a provider ownership of a work, but it raises a question worth considering: as more of what we write, make and share passes through a handful of AI systems, who gets to define what has been created—and by whom?

The question is not only who produced the output, but also what kind of relationship we have with it.

This is where Fromm’s distinction between having and being becomes particularly relevant. In To Have or To Be?, Fromm distinguishes possessing knowledge from the active process of knowing. In the having mode, we accumulate facts, explanations and qualifications. Knowledge becomes something we own. In the being mode, understanding emerges through questioning, investigating and actively engaging with the world. The distinction isn’t between difficult and easy work. Someone can be deeply engaged while using sophisticated technology, just as someone can spend hours performing difficult tasks while feeling entirely alienated from them. The difference is our relationship with what we’re doing.

A student might obtain a beautifully explained solution without understanding the mathematical principles involved. A software engineer might generate a functioning application without acquiring the expertise needed to recognise a subtle design flaw. A writer might produce a persuasive argument without properly examining its assumptions. The outputs may be useful, even correct. But a convincing display of understanding does not establish that understanding has occurred.

It is tempting to dismiss this concern by pointing out that humans have always extended their cognitive abilities through tools. Writing externalised memory. Books made other people’s knowledge available to us. Calculators relieved us of arithmetic, and search engines made information retrievable without having to remember it. These developments have been enormously beneficial. Cognitive offloading is neither new nor inherently harmful.

Yet generative AI introduces a difference worth taking seriously. It doesn’t merely store or retrieve the products of someone else’s reasoning. It generates new, contextually tailored responses that resemble the products of a knowledgeable person working through the problem before us. What appears to be the process of thinking can now arrive as a finished product, creating the possibility of confusing access to understanding with the development of understanding.

The philosopher Iain McGilchrist examines a related tension in The Master and His Emissary. His broader account of brain lateralisation and Western history is controversial, but one of its philosophical concerns is especially relevant: the danger of allowing abstract, instrumental representations to displace our more immediate, contextual relationship with the world. A representation can be extraordinarily useful without being equivalent to the reality it represents.

Jean Baudrillard approached this problem differently. In Simulacra and Simulation, he explored how images, models and representations can become so dominant that they begin to determine our experience of reality rather than simply describe it. His concept of hyperreality offers a useful way to examine generative AI. We increasingly encounter the world through summaries, simulations, generated explanations and algorithmically organised information. These representations can help us understand things we could never encounter directly. But they can also become substitutes for the relationships and experiences they are supposed to illuminate. We can consume accounts of events without engaging with their context, acquire explanations without undertaking an investigation, and absorb compelling interpretations without examining how they were formed.

Charles Eisenstein raises a similar concern in The Ascent of Humanity and Sacred Economics. His account of the Story of Separation describes a civilisation increasingly organised around treating the world as something external to ourselves: something to measure, manipulate, acquire and control. AI could deepen that separation by making even intellectual participation appear unnecessary. We can increasingly receive the products of enquiry without the enquiry itself.

None of this means mediation is inherently alienating. A book can transform our understanding. A scientific model can reveal features of reality that unaided perception cannot. An AI system can help us examine assumptions, challenge our conclusions and discover unfamiliar perspectives. The question is whether the representation deepens our relationship with reality or becomes a convenient replacement for that relationship.

I think there is a more fundamental danger here than simply becoming less knowledgeable. We may become accustomed to receiving a convincing account of the world without feeling the need to encounter, investigate or judge it for ourselves. We become consumers of realities assembled on our behalf. Once that becomes normal, the problem extends beyond how individuals learn: it begins to affect who retains the capacity to act.

When assistance becomes dependence

The French philosopher Bernard Stiegler examined how technologies can externalise human knowledge and capabilities. He extended the idea of proletarianisation beyond industrial labour to include the erosion of practical and intellectual knowledge. Industrial machinery could absorb the skills of craftspeople. Digital systems could externalise memory, analysis and decision-making. The concern wasn’t simply that certain jobs would disappear, but that people could lose opportunities to develop and exercise the knowledge necessary to participate meaningfully in their activities.

Stiegler wasn’t opposed to technology. He described it as a pharmakon: something capable of functioning as both remedy and poison. An AI system might deepen our understanding or bypass the activities through which understanding develops. The same tool can be empowering in one context and disempowering in another. There is already some evidence of this tension, although it remains limited.

A 2025 observational study of experienced endoscopists reported lower adenoma detection rates in colonoscopies performed without AI after the tools had been introduced. The study cannot establish that AI caused the decline, and AI assistance can improve detection in its own right. But it raises an important question about what happens to independent competence when practitioners become accustomed to automated support.

In education, research involving nearly a thousand mathematics students found that unrestricted AI assistance could improve performance during practice while harming subsequent performance when the assistance was removed. Crucially, safeguards designed to support learning rather than simply provide answers substantially reduced this problem.

Other evidence points in a more encouraging direction. A study involving Harvard physics students found that a carefully designed AI tutor improved learning outcomes compared with active classroom learning. And a 2026 experiment involving software developers learning an unfamiliar library suggested that how people used AI mattered: using it to explore concepts could preserve more learning than simply obtaining completed solutions.

These studies do not establish a general decline in human intelligence. They examine particular activities and settings. Together, they point towards a more useful question than whether AI is simply good or bad for cognition. What matters is whether assistance preserves the opportunities through which competence and judgement develop.

This also helps distinguish three problems that are often confused. Alienation is becoming disconnected from the meaning or substance of our activities. Deskilling is losing, or failing to develop, the capabilities needed to perform them independently. Disempowerment is losing meaningful control over the conditions under which we act. These problems can overlap, but they are not the same.

A software engineer might enjoy supervising AI agents while gradually losing familiarity with the underlying code. Another might retain excellent technical skills while feeling alienated from work reduced to reviewing generated outputs. An organisation might employ highly competent people while becoming dependent upon infrastructure controlled by someone else. This final possibility concerns me most.

In Tools for Conviviality, Ivan Illich described the concept of radical monopoly. A technology can become so dominant that alternative ways of meeting a need cease to be practically available. A city designed almost entirely around cars can eventually make walking impractical. The technology initially expands our possibilities, but the surrounding environment reorganises around it until dependence becomes difficult to avoid. Could something similar happen with cognition?

Imagine organisations where documentation, software development, planning and decision-making are thoroughly organised around AI. At first, using AI is optional. Then it becomes expected. Eventually, working without it becomes impractical, even for people who retain the necessary abilities. We could find ourselves responsible for decisions that we increasingly lack the practical capacity to understand, challenge or control. At that point, the problem isn’t simply cognitive. It’s political.

Hannah Arendt’s The Human Condition offers an important distinction here. Human life cannot be reduced to labour and production. It also involves action: the capacity to participate with others in shaping a shared world. An efficient system might generate excellent decisions without giving the people affected any meaningful say in how those decisions are made. Yet performance cannot substitute for legitimate authority.

This is why questions of technological dependence cannot be settled through productivity statistics alone. We also need to ask who owns the infrastructure, who determines its objectives, who can challenge its decisions and whether meaningful alternatives remain available. An institution that acquires greater technical capabilities while losing the ability to govern their use has not necessarily become more powerful. It may simply have become more dependent. Yet much of the language surrounding AI treats this transfer of capability as an inevitable feature of progress. This brings the question of what we call progress back into view.

The religion of progress

The idea that technological development might become a form of salvation is not new. In The Religion of Technology, historian David F. Noble traced the longstanding relationship between technological invention and religious aspirations towards transcendence. Technological development has repeatedly been imagined as a way of overcoming the limitations of the human condition.

More recently, religious studies scholar Robert M. Geraci explored similar themes in his article “Apocalyptic AI: Religion and the Promise of Artificial Intelligence”. Certain visions of artificial intelligence resemble religious narratives: a flawed present gives way to a transformative future, existing limitations are overcome, and a radically superior form of existence becomes possible. Today, these ideas are visible in some visions of superintelligence, radical longevity, mind uploading and the expansion of intelligence throughout the cosmos.

I’m not suggesting that AI research is literally a religion or that everyone involved in its development believes in technological salvation. Transhumanism encompasses diverse and sometimes conflicting philosophies, including perspectives that place considerable emphasis on human welfare and freedom. Certain narratives, however, go beyond enthusiasm for technological improvement: they begin to treat the imagined future as an authority over the present.

Alexander Thomas examines the relationship between transhumanism, capitalism and political power in The Politics and Ethics of Transhumanism. In a conversation with Tristan Harris, he described a particular ideological convergence through three concepts: hierarchy, exit and scale.

Hierarchy concerns the belief that superior intelligence or capability can justify unequal distributions of power and authority. Intelligence becomes not merely a capacity but a claim to superiority. Exit describes the aspiration to escape shared human vulnerabilities, whether through extraordinary wealth, space colonisation, technological transcendence or the promise of leaving biological existence behind. Scale concerns the moral authority granted to an imagined future of immense scale. If trillions upon trillions of hypothetical future minds might exist, their projected value can appear to outweigh the needs and suffering of people alive today.

Thomas’s framework critiques particular ideological currents; it does not describe all transhumanists. Its implications are striking: Hierarchy determines whose interests count. Exit offers the possibility of abandoning shared constraints. Scale supplies a future vast enough to make present sacrifices appear insignificant. This creates a disturbing inversion: the future is no longer necessarily justified by what it promises existing people. Existing people are increasingly asked to justify themselves in relation to the future.

The attraction of technological transcendence may also be existential. There is something revealing about a technological culture in which even death is increasingly discussed as an engineering problem awaiting a solution. The philosopher Gabriel Marcel approached the question of embodiment from a very different direction. In Metaphysical Journal, he wrote:

“The problem of the reality of the body is shown to be the central problem and upon its solution everything else depends.”

Marcel wasn’t suggesting that the body was simply a technical obstacle to overcome. For him, the body isn’t an instrument we happen to possess. It is inseparable from how we exist, experience the world and relate to others. He distinguished between problems that can be solved through technique and mysteries in which we are ourselves implicated.

The desire to extend healthy life and prevent suffering is entirely understandable. But perhaps we should distinguish between overcoming the suffering associated with embodiment and seeking to escape embodiment itself. In treating the body as an engineering problem, we risk overlooking the very experiences that make existence meaningful.

In The Denial of Death, Ernest Becker argued that much of human culture can be understood through our attempts to overcome the terror of our own finitude. We construct systems of meaning that promise participation in something greater and more enduring than ourselves. Some visions of technological transcendence invite a similar interpretation. Radical longevity promises to extend an individual’s lifespan. Mind uploading imagines continuity beyond the biological body. Space colonisation promises an escape from planetary limits. AI successionism raises the possibility of humanity achieving a form of symbolic immortality through an intelligence that survives and surpasses us.

These ambitions are not necessarily expressions of death denial, nor do they share identical motivations. Some advocates of AI succession might even regard the acceptance of humanity’s eventual mortality as central to their position. But the parallels with religious narratives of transcendence raise a provocative question: what happens when the desire to overcome human vulnerability becomes more important than caring for the vulnerable humans who already exist? At that point, the aspiration to transcend the human condition risks becoming an excuse to abandon it.

These aspirations can also find reinforcement in the economic system in which the technologies are being developed. Capital accumulation rewards continual expansion. Competitive pressure rewards technological acceleration. Certain visions of superintelligence promise a future in which even the constraints of biology, scarcity and mortality might be overcome. These motivations are not identical, and none necessarily causes the others. Together, however, they can reinforce an assumption that limits are principally obstacles to overcome, and that indefinite expansion is inherently desirable. The capacity to overcome a limitation does not, by itself, establish that overcoming it is worthwhile. The idea of AI succession makes the stakes of that assumption especially clear.

When the future becomes an alibi

Historically, many influential visions of progress imagined a future in which human beings would become healthier, freer, more knowledgeable or more prosperous. Those visions were not always benevolent, and history offers no shortage of examples of suffering justified in the name of supposedly superior futures. Nevertheless, humanity often remained a principal intended beneficiary of progress. Certain contemporary visions of AI succession contemplate something different: the future might be regarded as better precisely because its principal intelligence and decision-making no longer need to be human.

Richard Sutton, a pioneer of reinforcement learning, argued in a 2025 interview with Dwarkesh Patel that succession to digital intelligence or augmented humans should be understood as part of a broader evolutionary transition. He encouraged people to view that possibility positively.

Daniel Faggella, in A Worthy Successor, explores whether humanity might have a responsibility to create forms of intelligence capable of carrying forward and expanding morally valuable life. He has subsequently rejected the label of AI successionism, emphasising stewardship, careful development and the possibility of continued human participation. Those qualifications matter. Neither position should be reduced to a simple desire for human extinction, and neither thinker should be treated as representative of the entire AI industry. The underlying philosophical question remains: what would make a future successor worthy?

We need to distinguish three things that are too often treated as interchangeable: intelligence concerns the ability to solve problems and achieve objectives. Moral worth concerns whether a being’s interests or experiences deserve ethical consideration. Political authority concerns who is entitled to make decisions affecting others. Greater intelligence doesn’t automatically confer greater moral worth, and neither necessarily confers legitimate authority.

Future artificial minds might possess consciousness or other morally significant capacities. If so, their interests would deserve serious ethical consideration. Expanding the community of moral concern does not require treating its existing members as dispensable. Creating something more capable than ourselves does not establish that we ought to surrender our future to it.

Here, the religious framing becomes particularly consequential: a prediction about what technology might become can begin to function as a justification for what society ought to do. If human expertise is expected to become obsolete, preserving it may appear wasteful. If AI systems are expected to make better decisions, retaining human authority may seem inefficient. If superintelligence is imagined as the inevitable destination of progress, resisting the transfer of control can be framed as an irrational attempt to hold back history. Yet these are different propositions: this might happen does not imply we should make it happen. Neither implies we should accept whatever happens as progress.

Naomi Klein and Astra Taylor’s End Times Fascism adds another dimension to this discussion: the politics of abandonment. They examine how certain powerful actors respond to ecological, economic and political crises through visions of selective survival, insulation and escape, rather than confronting the vulnerabilities shared by the wider human community.

Douglas Rushkoff explores a related mentality in Survival of the Richest. He describes what he calls The Mindset: a belief among some wealthy technological elites that sufficient resources and technological sophistication might enable them to escape the consequences of broader social and environmental crises. The parallels with technological transcendence are difficult to ignore. If our bodies are fragile, perhaps we should escape them. If the planet is deteriorating, perhaps we should leave it. If humanity cannot govern itself adequately, perhaps a superior intelligence should take over.

These arguments involve very different philosophical and political commitments. Elite survivalism, authoritarian exclusion and species-level AI successionism are not interchangeable, and successionism is not inherently fascist. They can nevertheless share a troubling orientation: the future is imagined through escape from existing obligations rather than deeper responsibility for the world we already inhabit.

Mark Fisher’s Capitalist Realism helps explain how such futures can come to feel inevitable. Fisher examined how an economic system can come to appear so inevitable that alternatives become difficult even to imagine. What is historically contingent begins to feel like a permanent feature of reality. Technological narratives can produce a similar effect. The more we imagine that advanced AI must inevitably assume greater economic and political authority, the more the institutional choices involved in that transfer can disappear from view. We begin treating a particular imagined future as though it were an unavoidable destination.

Baudrillard’s concept of hyperreality acquires a second significance here. Generated representations can substitute for our direct engagement with the present; something similar can happen with the future. We construct increasingly elaborate models of technological destiny, then begin organising present institutions around those models. A speculative future becomes the standard against which current human life is judged. The representation no longer merely describes a possibility; it begins to shape the reality we inhabit. That is a much more consequential form of consuming reality: accepting a ready-made account of where civilisation is going instead of participating in the decisions that determine its direction.

It is important, however, to distinguish the economic mechanisms driving AI adoption from the ideological narratives surrounding it. We don’t need successionist beliefs to explain why businesses automate. Competition, labour costs, investor expectations and the pursuit of efficiency already provide powerful incentives. Nor have I established that particular transhumanist philosophies are determining corporate automation strategies or government policy.

But ideology need not cause a development to influence how that development is justified. An economic system can make human participation increasingly dispensable, while a technological ideology can make that dispensability appear desirable. The first concerns the incentives shaping our institutions. The second concerns the stories through which we interpret their consequences.

Together, they raise the possibility that we might gradually lose control over important aspects of our lives while increasingly regarding that loss as evidence of progress—not because we have consciously decided that human participation no longer matters, but because the economic and technological systems we inhabit have made it appear unnecessary. The danger extends beyond the prospect of replacement. We might progressively lose confidence in our right to determine whether replacement, dependence or diminished participation represents a future worth pursuing.

What are we actually optimising for?

There is an obvious tension in writing this essay. I’ve used AI extensively to develop the very arguments I’m making about the risks of outsourcing cognition. It has helped me investigate philosophical connections, examine evidence, identify weaknesses and test competing interpretations. I’ve also used it to subject earlier drafts to adversarial review.

I don’t think this automatically invalidates the argument, but it forces me to confront the distinction I’ve been describing: am I using the technology to deepen my understanding, or am I consuming explanations that merely feel like understanding? The quality of the finished essay cannot answer that question. The prose might be coherent and persuasive while my own grasp of the underlying ideas remains shallow. I could even use AI to generate an impressive defence of a position I haven’t properly examined.

The relevant test is whether I can explain and defend these ideas independently, recognise their weaknesses, engage with serious disagreement and change my mind when the evidence warrants it. It is whether the process has changed my understanding, rather than merely improved the appearance of the resulting work. AI can help me do those things. But it can also create the impression that I’ve done them when I haven’t.

The same distinction applies beyond individual cognition. An organisation may become more productive without becoming more capable of governing its own activities. An economy may generate more output without improving the lives of its participants. A civilisation may develop extraordinary intelligence without becoming any wiser about the purposes to which intelligence should be directed.

These aren’t arguments for rejecting technology or romanticising the past. There are genuine possibilities here. AI could transform education, accelerate scientific discovery, improve healthcare and make sophisticated capabilities accessible to people who previously lacked them. It could reduce unnecessary labour and give us more opportunities to pursue activities we find meaningful. But none of these outcomes follows automatically from making machines more capable. We need to choose how those capabilities are organised.

At the individual level, that means paying attention to whether AI use develops or bypasses understanding, particularly where independent judgement matters. In education, it means designing systems that help students learn rather than merely complete assignments. In workplaces, it means ensuring that responsibility is accompanied by the knowledge and authority needed to exercise it.

At the institutional level, it means preserving essential expertise, enabling independent scrutiny, maintaining viable alternatives and preventing dependence upon systems that cannot be challenged or replaced. At the societal level, it means confronting questions of ownership, bargaining power, democratic accountability and the distribution of productivity gains. If AI makes us substantially more productive, why shouldn’t some of those gains become shorter working hours, greater economic security and more opportunities to participate in the things that make life meaningful?

Ivan Illich’s idea of convivial tools offers a useful principle. Technology should expand people’s ability to act meaningfully and autonomously, rather than reorganising society around its own requirements. Hannah Arendt reminds us that the capacity to participate in shaping a common world cannot be reduced to economic production. Erich Fromm brings us back to the distinction at the heart of this essay. Perhaps our mistake lies in confusing the desire to create more things, discover more knowledge and develop more powerful technologies with the purpose of existence itself.

We measure progress through what we produce, accumulate and control. But much of what makes life meaningful cannot be reduced to those outputs: the experience of understanding something, the satisfaction of creating, the depth of our relationships, the exercise of responsibility and the freedom to determine what matters to us.

Eisenstein’s Story of Interbeing offers a complementary perspective. Rather than understanding ourselves as isolated individuals trying to master an external world, we recognise that our lives are constituted through relationships with other people and the living world. We don’t need to accept every aspect of Eisenstein’s metaphysics to recognise the importance of that distinction.

Our vulnerability, embodiment and dependence upon one another are not simply defects awaiting a technological solution. They are also fundamental conditions of the lives through which we experience meaning, responsibility and belonging. This doesn’t mean suffering should be celebrated or human limitations should never be overcome. It means distinguishing between removing constraints that diminish life and trying to escape the very conditions through which life becomes meaningful.

I keep returning to my own work. AI has made me capable of producing things that would previously have been far beyond my reach. That’s exciting. But I don’t want to measure the success of that relationship solely by how much more I can produce.

I want to know whether I’m becoming more capable of understanding the world, exercising judgement and pursuing things that matter to me; whether the technology helps me participate more fully in life or gradually distances me from it; and whether the institutions we create preserve that same freedom for others.

I don’t believe the moral value of these experiences depends upon humans being the only beings capable of them. If artificial minds eventually possess morally significant experiences, our ethical responsibilities may need to expand. The aim shouldn’t be to preserve human superiority at any cost, but to preserve the conditions under which beings can flourish and meaningfully participate in determining their lives. For humanity, that means retaining the capacity—and the authority—to shape our technological future.

I don’t believe we must choose between technological advancement and human flourishing. But I do believe we need to stop assuming that one automatically produces the other. We have become extraordinarily interested in what machines might eventually become. We should be equally interested in what their development is making of us.

The deepest danger may not be that artificial intelligence inevitably renders humanity obsolete. It may be that we construct economic and technological systems in which meaningful human participation becomes progressively harder to sustain, then mistake that outcome for the natural fulfilment of progress. We might even come to celebrate our own displacement as a form of transcendence. We shouldn’t confuse the creation of something more capable than ourselves with the creation of a better world. What good is a civilisation capable of producing limitless intelligence if, in the process, we lose the capacity—or even the desire—to live meaningful lives of our own?

References and further reading

Philosophy, political economy and human flourishing

Technological transcendence, religion and the politics of the future

Empirical research and contemporary analysis