Technology is death. Unless we will it firmly otherwise, that is the direction.
We can will it otherwise, I think, and most of us probably try. But it’s worth being clear about the trajectory, after the rebuke that US voters delivered to the monstrous alliance between MAGA and its harem of tech titans.
A win for democracy, sure. MAGA is faltering, at least in its current form. Big Tech might find its wings clipped a bit. But end of story? Absolutely not.
Death, of course, takes different forms. It may not be the case that the robots are coming to kill us — possible though that is, and, tragically in the age of proliferating drone warfare and the possibilities of AI-powered war, almost certain for some. And there are scenarios in which AI could make or contribute to decisions that lead towards escalating nuclear or biological attacks, the cold application of game theory unmoderated by law or human empathy.
Alternatively, as Mustafa Suleyman has warned lately, those who nurture ideas that AI systems deserve attention to their wellbeing or rights could lead these systems towards expecting a form of autonomy, which they will perhaps eventually defend through coercive means.
Other kinds of death are available too. A mass of messianic titans see themselves as birthing a species that will surpass humans altogether, an alien super-intelligence already among us, one that will reach out to the mysteries of deep space without any of our constraints of flesh and mortality. All hail their Artificial God: omniscient, for it eats all knowledge and remembers everything; functionally omnipotent, for its powers extend daily, it never rests, its clarity of purpose never wavers; and absolutely omnipresent. As for what happens to us as it takes shape: does that even matter?
But short of the most dramatic possibilities, technology simply invites us into a kind of voluntary death. It is the creeping, insidious surrender of self to machines; an emerging world in which the very things that make us human start to be devalued, disdained, displaced, replaced; we start to doubt our very selves, submit ourselves, start to wonder what we are for.
Tech offers a cold hand pulling us deeper into isolation, towards new depths of loneliness, seeking solace brought to us by code, and we quietly despise the personas we are forced to wear as we bend to this new world.
The Artificial God eats us from the outside and the inside.
On the outside, we are familiar with the pattern: an all-out assault by the capitalist endgame on democracy and human equality. Long gone is the comforting (to some) myth of classical liberalism, that democracy and capitalism share an undying commitment to the open markets of money and ideas. That simply gave us old-fashioned inequality. Now, instead of that old game, the Artificial God wants to eat the state: the state becomes fused with it, not understanding it, not seeing it, until the one merges into the structure of the other, like a host consumed by a vast banyan tree.
The canonical utopian/dystopian (you choose) words of this vision were spoken years ago at Libertopia, by Peter Thiel, the man who is perhaps more than any other the godfather of this whole thing:
“the basic idea was that we could never win an election on getting certain things because we were in such a small minority. But maybe you could actually unilaterally change the world without having to constantly convince people and beg people and plead with people who are never going to agree with you, through technological means. And this is where I think technology is this incredible alternative to politics.”
While the flirtations of the hyper-accelerationists with a violent and authoritarian political project might be in abeyance for now, their “incredible alternative to politics” doesn’t just disappear. All the ingredients are still in place for their unique utopia/dystopia of unprecedented surveillance, lethal robots, trillionaire wealth for the demigods, and an inescapable distraction network that turns the whole takeover into sparkling entertainment.
So much for the mythology of buccaneering capitalist adventurism. Even the “technofeudalism” of which we have been warned is just an early stage in a much bigger project. When an over-hyped IPO briefly gave us the first of those trillionaires, the pattern of it all flashed up so clearly: an ever-shrinking group receding ever further from our grasp, the new brahmins in what is becoming a kind of global caste system, gripped in a fever dream of unfettered rationalism and optimisation. Our labours and our labour alike are eaten up by their voracious Artificial God. As for us, we might just find ourselves lining up for the largesse of a monster that chewed us up and digested everything we ever did and whose creators never told us why; and only then, perhaps, will we realise that we never asked.
Still more dangerous is the hollowing-out from within.
As long as democracy is alive, we have a prayer. Tech has long nurtured the conditions for its strangling: years of algorithmic sorting have tribalised and polarised us, chiselled away at our faculty for conversation with the one who experiences the world differently. But democracy can only function as long as we trust ourselves.
The era has arrived in which machines have come to emulate humans. Yet, for long (so long), we had already been teaching ourselves to become the kind of humans who would emulate machines. What we seem to have now is the mechanical modelled on ways of living that have been modelled on the mechanical. As the cycle continues, ask yourself who (or what) is likely to come out on top.
At the heart of this is an ontological problem disguised as a performance issue. In other words, what we are asked to accept as a question of optimisation and productivity is really a challenge to the very nature of our being. If we accept certain terms — specifically that the content of our lives consists very substantially in being productive, efficient, and effective — then we very likely have to accept that the machines are entirely capable of outperforming us, if they don’t already. If we judge the value of our lives according to the same things, then the supreme optimisation machines we call AI are in very real sense better than us. We are defeated already; we are redundant, literally.
Mechanical us: second best.
Tech invades our minds, at our invitation, leaving space for little else. Having already offered up our time and attention, as we now welcome these optimisation machines (and the giant corporations behind them) into the intimate places of our lives (at least until we run out of tokens for the day), the question we must face is whether we have any inner sanctum left at all.
How complete has the colonisation become? Is there any sacred place remaining?
Like you, I am a concentric castle: one ring inside another, each more sacred than the last. And as I open one door inwards after another, towards the very heart of myself, I find myself closing my eyes in fear. Will I find there a garden, as I dream? Or just another machine, humming away?
The more we give away, the more we start to give up on our own judgement, losing faith in our own intuition and decision-making. We can no longer trust what is real and what is phoney, no longer see the distinctions between this and that.
I understand the human brain processes 11 million bits of information per second, while our conscious minds process barely 0.0004% of this. We are consciously aware of so little of what we know. Temperature, air pressure, ambient sound, the feel and texture and colour and smell of things — these form a vivid map of signals for our minds, our map of reality, shaping how we make sense of the world, how we know our place in it, how we decide what to do. AI systems know none of this. And as we submit ourselves to them, the risk is that we no longer allow ourselves to trust or even access the richness of our own minds.
Here is where we could eventually find the interior death of democracy, that messy spectacle that never quite seems to solve our problems.
Meanwhile, the machine watches. The machine sees all. The machine is untroubled by the turmoil of emotion. The machine knows best.
Something much better lies beyond the mess, the messianic titans wants us to believe. Not government, not politics. Our very bodies might prove to be a liability in the unfolding future; we can and must begin to transcend them. Welcome, then, to the cold and timeless embrace of numbers.
The more responsible ones talk about alignment as our way to evade the nightmare. This is a mashup of ethical and technical questions on how to keep AI in some kind of synchronicity with “human values” so as to manage a spectrum of risks that range from disruptive to unthinkably calamitous.
The reasoning behind it is clear enough, and entirely justifiable and necessary. AI alone has no “natural” instincts to provide ethical parameters for its actions to fulfil the task it has been given. If a shortcut presents itself, unless the AI system has boundaries that explicitly prohibit that course of action, the drive for optimisation will cause it to take the shortcut, whatever that portends. Can a sort of ethical compass be built into AI, an adjunct to its specific instructions to guide how it must reason and act?
The difficulties in achieving this are clear: it is not only a technical problem. What are “human values” really? We have never agreed on anything close to a universal expression of values, so there is no firm ground for any universal alignment. While most of would hold in theory to a basic idea that all humans have equal worth and dignity, just look at the reality of the world.
But as much as we need these guardrails, alignment itself is a difficult metaphor. Alignment is what we do with the tyres of a car, to keep them running parallel and straight. Alignment is what we do with schedules, to handle practicalities. Alignment with law is the business of compliance. But what if values do not behave like this at all? What if their fundamental character is fluid, relational, discovered?
Alignment is the map of a river, a long blue line. But it’s not the river itself. To know the river is to move with it through a landscape that is constantly changing; it is contiguous, but full of currents, eddies, rocks, counter-flows, all of which must be encountered, felt, overcome. And the only way to do that is to participate in the flow; there is no way to freeze the river and know it.
Yet, there is a sense in which the idea of a hypothetically perfect alignment haunts me even more than a catastrophic misalignment. Why? Because it is the point at which we stop asking questions.
Because as if we didn’t have enough to worry about, there’s a deeper layer to all this, as though the very alignment problem is itself misaligned.
Here’s the best deal on offer: AI stays subservient to “us”; it will serve “our” ambitions (whoever “we” may be). It will align with our trajectory of historical progress. It will supercharge our productivity and hyper-optimise our strivings. It will take the great exponential arc, surge with us ever further and faster upwards.
But what if the arc itself is based on a dangerous misalignment with the very substance of life?
Whatever should we do? This is traditionally the point for throwing up one’s hands in despair — or, like Bill Gates, mumbling something about throwing in your lot with the Amish.
There is the Apocalypse Now option, argued with phenomenal force by Arundhati Roy (for what kind of regulation could possibly make sense?): “terminate with extreme prejudice”.
Then there is the approach, championed among others by the African Union and ASEAN, which embraces a vision of AI that is about delivering real-world benefits for humans, for our social and economic development and cultural renaissance. There is merit to this. As a policy direction, this is reasonable. It feels congruent with my own, very routine use of AI. But unless we can wrest some sort of collective control from the titans who would extricate themselves even from the most powerful states, the risk is that we find ourselves quaintly toying with something over whose ultimate direction we have absolutely no leverage.
This leaves a third option, which has nothing to do with policy and is nothing more than a posture, an orientation, one that is yet to yield its wisdom. But is I think where a good number of us are arriving.
It is an invitation to confront the simplest of questions that AI puts in front of us: Who are we, and what are we for? And which, then, of the various shades of death are we prepared to accept, for the sake of our gods of economic growth and productivity whom we honour with our lives and souls?
Because when we do that, we might find that the opposite of the deaths we are offered is to start wrapping ourselves more tightly in the tendrils of all that lives. Here, in this cocoon, we might find not cold-blooded alignment, but something rather more precious than that, something which tech must always stand outside. Not mere abstract “human values”, but the (re)discovery of care for what is fragile; and the delights of reciprocity; and love, and music, and perhaps something irreducible about the character of life itself.
Shall we start here?