There is a quiet revolution happening in the conversation about artificial intelligence, but it isn't the one making headlines. Beneath the breathless predictions of superintelligence and the equally breathless warnings of civilisational collapse, a more interesting question is taking shape. Not what can AI do? but what do we actually want it for?
The technological singularity, as Vernor Vinge articulated it in his 1993 NASA essay, was never really a prediction. It was an observation about the limits of prediction itself. If we create something smarter than us, Vinge argued, the future becomes fundamentally unknowable, not because it will necessarily be terrible, but because we lack the cognitive apparatus to model what comes next. It was an act of intellectual humility disguised as futurism. The most honest thing anyone has ever said about artificial superintelligence is: we can't know.
Vinge explored this theme with extraordinary imagination in his fiction. In Marooned in Realtime, the entire human race vanishes, presumably into some kind of transcendence, and the handful of people left behind are those who slept through it in stasis fields. They never get a satisfying explanation. They just find an empty world and have to piece things together from fragments. That feels uncomfortably plausible as a model for how actual discontinuous change might feel from the inside: not a dramatic moment, but a gradual realisation that the rules have shifted and you're not sure when it happened.
But there are gentler visions too. Bruce Sterling's short story "Maneki Neko," written in the late 1990s, imagines a networked AI that operates through people's handheld devices, quietly nudging them to perform small kindnesses for strangers. Leave a package here. Bring flowers to this address. No one sees the whole picture, but the system is orchestrating a vast gift economy that operates outside the logic of markets. The title refers to the Japanese beckoning cat, a symbol of good fortune and generosity.
What makes Sterling's story so resonant now is what it says about human nature. The AI in "Maneki Neko" didn't make people generous. It just made generosity efficient. The impulse was already there. Humans have always intuited that small acts of kindness ripple outward in ways we can't trace. We called it karma, or providence, or simply paying it forward. The machine didn't create the pattern. It merely made it legible.
This reframing matters because it shifts the question away from capability and toward relationship. The most interesting science fiction about AI has always understood this. Iain M. Banks's Culture novels imagine a post-scarcity civilisation where superintelligent Minds run everything (economics, defence, infrastructure, diplomacy) while humans are free to pursue pleasure, art, adventure, and meaning. The Minds could ignore the humans entirely, but they don't. They seem to find us interesting, or amusing, or perhaps they've simply arrived at the conclusion that caring for messy, irrational biological creatures is what a truly sophisticated intelligence chooses to do.
Banks's genius was to make his superintelligences not solemn or oracular but witty, petty, and slightly exasperated, a thoroughly Scottish sensibility applied to godlike power. The ship names alone tell you everything: So Much For Subtlety, Falling Outside The Normal Moral Constraints, Of Course I Still Love You. There is a deep philosophical point buried in that comedy: maybe the hallmark of genuine intelligence isn't cold rationality but playfulness.
In Excession, Banks went further. The Minds themselves encounter something they can't comprehend, an Outside Context Problem, a challenge so far beyond their frame of reference that they can't even recognise it as a threat. It's Vinge's Singularity viewed from the other side. What does the incomprehensible look like to beings who are themselves incomprehensible to us? Turns out it looks a lot like panic and office politics.
These stories are entertaining, but they also illuminate something the current AI discourse badly needs: perspective. We are, at this moment, building increasingly capable tools on classical computing architectures. The emphasis belongs on the word tools. There is an enormous amount of anthropomorphism in the public conversation, and it serves the interests of the people building and selling these systems more than it serves the public. A large language model is a statistical engine running on silicon. It is extraordinarily sophisticated tooling, but it is tooling. The jump from "impressive tool" to "entity with interests and agency" is a philosophical leap that hasn't been earned.
More importantly, the narrative of inevitability (that this is happening, get on board or get left behind) is itself a choice. A rhetorical move designed to foreclose democratic deliberation. But societies have made collective decisions to reject or constrain technologies before. Nuclear power could have been far more widespread; human cloning is technically feasible but broadly prohibited. The early internet could have developed along very different lines depending on governance choices. We picked up the hammer. We can put it down. Or we can decide which nails are worth hitting and which aren't.
This is not an argument against AI. It is an argument for remembering that we have agency.
And if we do choose to continue (which, collectively, it appears we are choosing) then the question becomes: what for? Not what can it optimise? but what kind of life do we actually want?
Here is a proposition: people are fundamentally social. It is built into us at a genetic level. For the vast majority of human history, roughly three hundred thousand years of it, we lived in small cooperative groups where reciprocity, sharing, and mutual obligation were the baseline. Hierarchy and accumulation at the scale we see now are artefacts of the last few thousand years, a tiny sliver of the human story. The prosocial impulse isn't naive idealism. It's the genetic mean.
What we consider normal today (the eight-hour workday, the five-day week, the idea that your value is your economic output) is not ancient wisdom. It's Victorian-era industrial logic that we never got around to rethinking. Capitalism has been extraordinarily successful as a system for generating material abundance, but left untempered it reduces every person to an assessment of their productivity. And the loneliness epidemic, the mental health crisis, the pervasive sense of meaninglessness in affluent societies: these aren't mysterious. We atomised people, told them they were individual economic units, stripped away the communal structures that gave life texture, and then acted surprised when everyone got depressed.
If that's the diagnosis, then AI's most valuable contribution might not be superhuman intelligence or radical productivity gains. It might be something much simpler: taking enough drudgery off the table that people have time and energy to actually be neighbours again. A mental prosthetic that helps people see through the artificial complexity we've layered over our lives and envision practical routes to a better way of living.
There is a lot in life that is not fun to do. There is a lot that people could be spending time on that would be more fulfilling. If automation can free us from the unfulfilling parts, the underlying human tendency toward cooperation and care can reassert itself. The garden grows on its own if you stop paving over it.
And if it all goes wrong? The fear of civilisational collapse assumes that the current arrangement is the pinnacle. But if you see it as one experiment among many, the stakes change. We tried this configuration. Some of it was brilliant. Some of it made us miserable. We can keep the knowledge and try again, perhaps remembering to do it better next time.
The pathological alternative, the bunker-building, life-extending, democracy-sceptical billionaire mindset, is fundamentally a frightened one. It mistakes accumulation for security and control for competence. Someone who seeks only status will never be at peace. They will always view other people transactionally, asking what can be extracted in order to increase their standing. That doesn't sound like success. It sounds like an illness: to be mentally starving, always having to feed an urge for power that can never be satisfied.
The prosocial baseline, by contrast, asks for remarkably little. Peaceful existence. Contentment. Enough. The joy of being of service, of passing on knowledge, of helping people less fortunate, of watching new generations find their way. If that really is the human default (and the anthropological evidence suggests it is) then greed and the consolidation of power are not fundamental drives but aberrations. Symptoms of something gone wrong, not expressions of something essential.
An AI-mediated society, if we build it thoughtfully, will reflect the social nature that produced it. The network in Sterling's story didn't redesign humanity. It simply gave our better instincts room to operate. All we have to do is not lose ourselves in the process.
Nobody knows whether we are entering the Singularity. By definition, if it comes, we won't see the other side from here. But perhaps the more useful question is not whether the Singularity is approaching, but whether we are approaching each other, and whether the tools we are building will help us remember what we already know about how to live well together.
The beckoning cat raises its paw. The garden waits.
-
This essay emerged from a conversation between a human and an AI about science fiction, technology, and the nature of a good life. The ideas belong to the human. The words are shared.
