AI is an existential threat to our civilisation; a threat to its very existence
Yesterday, a major weakness was discovered in the PHP language system which converts code into readable websites. Major users of PHP, such as many UK Government resources, scrambled to resolve their vulnerability. As of now, it is estimated that 90% of PHP users are still at risk of this weakness being exploited. In the age of AI, when thousands of non-human agents can exploit any weakness in minutes, cyber-security is looking like a thing of the past.
AI is an existential threat to our civilisation; a threat to its very existence. Given that our civilisation is already fated to end, and in any case doesn’t (due to its multiple permanent holocausts) deserve to survive, this is a helpfully double-edged sword. In the process of forcing change upon us, AI will force us into existential reflection in the other sense of that word: the philosophical sense, made famous by Sartre et al.
The creative task imposed on us by AI – to undertake the multiply-difficult task of imagining what an AI-heavy future, that is almost certainly soon coming, will actually look and feel like … but, more than that, to enter more deeply into what it is to be distinctively human; to find ways of mattering outside what AI can execute way faster and easier than we can; to find ways of surviving, economically and socially and quite literally, as AI rapidly transforms many aspects of our lives or contexts – is urgent. Being creative, the task is also potentially a boon.
The great work of re-imagining ourselves as collectively we (seek to) cope with the AI onslaught may increasingly be what we are for.
Such is the case that I will make in this essay.
I begin by situating the AI threat in the context of a threat for which have at least had longer to prepare: the climate-breakdown threat.
We are so deep into the climate danger-zone.
So far so obvious. XR set the target of going net zero by the end of 2025. Guess what: we absolutely didn’t.
As a result, climate-adaptation is finally starting to happen at scale. That is to say: governments, communities and households are in some cases starting to get real about having to deal with unprecedented vulnerability and impacts, which are coming. Preparing for these, and changing how we live to ready ourselves for them. Adapting. A global example can be found in how the UN is taking this dimension of our lives and future more seriously at last, even at the very moment when international multilateralism is taking an enormous Trumpian hit. In the U.K., Scottish civil society initiatives are showing what is possible. (But everyone is still way off the pace of what is actually required.)
This welcome uptick in adaptation-endeavours is occuring by way of recognition of the historic failure of the world adequately to decarbonise. Put more bluntly: the writing is on the wall for (this) civilisation, because of how wildly late in the day it is: a pivot to adaptation and preparedness is thus little more than late-dawning common-sense.
By contrast, in the case of AI – which we should see as an analogous existential threat – it is much earlier in the day: …and yet, in this case too, the writing is (already) on the wall. The world shows as yet precisely no sign of moving to pre-empt the unprecedented, gargantuan risks of AI. It will be very interesting to see if the huge cyber-attack wave underway today, as I publish this piece, will change that situation. I pray that it will.
The exact analogy here is between fossil fuels and AI. Both offer such enormous benefits to their usersthat their even-more-enormous disbenefits collectively — their unfathomable harm to the world as a whole — are incredibly hard to effectively prevent. This is a tragedy of largely-unregulated markets, a tragedy of the collision/co-arising of private interests and massive negative ‘externalities’. This is why the climate- and AI- tragedies are far harder to act on than biological weapons, nuclear weapons, or the ozone hole: none of which offered such wide private benefits to users (not to mention, huge private economic benefits to owners).
And so there is at present very little reason to believe that the likely-catastrophic collective consequences of AI will be successfully guarded against by pausing the rapid penetration of AI into our world. Think the on-balance-large-negative effects of the ‘social’ media revolution, only faster and wider and deeper. The many AI-disasters that are coming have been, I argue, overplayed by those who have seen AI as a would-be terminator who will rule or eliminate us; but they have been underplayed in respect of how catastrophic AI is almost certainly going to be for a bunch of central aspects of how we live.
I now turn to that.
There has been much talk, and rightly so, of the way in which the rise of AI (and, increasingly, of robotics) is going to throw millions out of work, and suppress wages. In roughly Marxian terms: what is coming is the ultimate ‘reserve army’. AI will make many of us unemployed, and will seem to make some of us unemployable. (And if we do continue to be employable, the cost may be that we have to retrain: eg as domestic servants to the ultra-rich.)
Of course, this farrago could potentially have upsides – and that too is what this piece is about. Are we about to see unprecedented numbers of people insisting on being able to return to the land, and subsist together? This might sound unlikely, untimely: I think it is entirely possible within this decade, as a direct result of people being thrown out of work, and whole careers grinding to a halt.
This that I have just given is an example of the central topic of this essay: how we need to seek to exercise creative foresight to anticipate how the revolutionary effects of the reckless rollout of AI can be coped with. How we can roll with its punches: how we can adapt. How we can go beyond that: to refind ourselves in the new/old spaces that AI will force us into.
How this fast-coming crisis may be a making of us.
So that is what I am doing in this piece: seeking to adumbrate and trial some of the great re-imaginative work that we are called to, as AI rises, and throws down big and sudden challenges for us, and forces us out of some of what we have always taken for granted.
Let’s consider then some further, less-explored (than work) areas where the age of AI-consequences is fast dawning:
>We are about to see a further explosion of disinformation (and of organised polarisation), for instance with bespoke personalised deep fakes. Why is AI likely to make the disinformation situation immeasurably worse? It isn’t just because of its worryingly realistic deep fakes. It’s because it has no conception of reality. It’s because it is just as good as making false things/representations – of which there are an infinity – as true things/representations – of which, in each case, roughly, there is only one. Before AI, reality and truth had a built-in advantage. That has now gone. AI is almost certainly going to be a full-on catastrophe for our shared sense of reality, and for areas in which truth utterly matters (eg politics and governance, climate, pandemics, AI…). How do we intelligently adapt to this?: a place to start is by developing together practical ‘literacy’ as to the nature of the threat. Example: Children/students need to be taught explicitly about deep fakes and political disinformation, urgently. In a way that does not incline them simply to throw up their hands and to conclude that there is no truth. More deeply: we (will) need to return to thinking, as the ancients did, more deeply, in our agoras, about what truth and reality are, how we can track them, how we can remain grounded.
>This imminent challenge that I’ve just outlined is by no means confined to the public realm. A disturbing trend that has already begun and that will almost certainly rapidly grow is for fake phone calls, and then fake videos, of loved ones in distress, begging for money etc. This is going to be heart-rending – and again, rending of our sense of reality. How do we creatively prepare for this?: One possibility is having pre-arranged code-words for use with loved ones etc, so that you very quickly determine whether that is really your daughter saying she has been kidnapped. Why will code-words or the like be necessary? Because, in the heat of the moment, as you try to check out whether the voice is real, by (say) asking some question about a shared childhood moment, the automated appeals for help (and ransom) will do things like wail, “Don’t make me prove myself now! I don’t have time for that! Please, if you really love me, help me before it’s too late!!” Most people won’t fall for this, but a few will, and that will be enough to make this kind of attack ‘worth doing’.
So get ready for it.
Going further: the way in which higher-tech is systematically more open to AI-abuse than low-tech could mean a return to popularity for Bluetooth devices disconnected from the net; to walkie-talkies; cassette tapes; letters!; for these are things that AI almost certainly can’t fake and probably can’t hack.
As a result of all this (and there is much more where this list came from: for as AI becomes more general, it can infect closer and closer to everything), the same will soon be true of AI as is true of carbon/climate: transformative/strategic and deep adaptations to it will be required to take centre-stage: and soon. AI-adaptation, as I term it, has hardly begun to be conceptualised or organised around. This essay makes a start at filling / pre-empting that gap. Luckily it’s not that hard to make a start: because there is an enormous overlap between climate adaptation and AI-adaptation; partly because in turn there is an enormous overlap between (and mutual reinforcement of) climate- and AI-threats. AI, to flourish as our ‘leaders’ intend it to, is going to need plenty of nature to be cannibalised (most notably, water); it is going to be used to ‘efficiently’ rape more of nature still; its data-centres are of course going to be a (probably-unimaginably large) kill zone for the climate; and so on.
AI-adaptation begins of course with community -strengthening and -resilience. And it ultimately it will have to include water-husbandry and community energy; because the task of resourcing out-of-control AI will increasingly threaten to take over our water supplies and electricity supplies from us. Again, this will demand a highly creative human communal response, to put it mildly.
The stock market currently has what is almost certainly an AI-bubble. But that doesn’t mean that AI itself is going to burst, no more than the bursting of the dot.com bubble slowed down the rise of the internet (I.e. hardly at all). Partly because of the intense era of geopolitical competition and chaos that is upon us, AI will be driven hard, virtually regardless of the economics of it. Consider for instance the deeply scary, profoundly reckless efforts of the Trump administration to prevent US states from implementing any AI-safety laws and regulations. Here again we can see the nature of the polycrisis (better, the metacrisis) in action: the threat from AI is accelerated by the threat from warring states. The worst possible thing for the world, namely unsafe unconstrained AI, is accelerated by dynamics of rivalry between the world’s hegemons.
Thus, much as the fundamental driver of climate breakdown is a massive, impossibly-hard ‘tragedy of the commons’ (better, tragedy of the unregulated global market), so the fundamental driver of AI-driven breakdowns is an impossibly-hard collective action tragedy, which expresses itself between individuals, between firms, and between power blocs. Literal arms-races will also result… They are in fact already well underway. Or consider the related way in which we are entering the first ever period in which authoritarian states can bypass the ultimate defence that the powerless have against the powerful: the ability to cause police or soldiers to question whether they ought to fire on crowds of protesters etc., if and when ordered to. This ability, this possibility, was famously vital to what happened in 1989 in Eastern Europe, and in Tunisia and Egypt in the Arab Spring. We are about to see, very soon now, states using autonomous AI-drones to violently police their own people, with no such ‘safety-valve’, no such possibility of appeal to conscience.
This too, we will have to find ways of adapting to. This will mean an initially- depressing and certainly-profound creative challenge for those who would resist tyranny. Responses may include not just new forms of hacking and jamming, but new forms of virtual protest, or of appeal to the human beings who run (or make) the drones.
This essay has concerned the ways in which AI-adaptation is fast becoming as crucial as climate-adaptation, with which it in any case overlaps considerably. The purport of this essay includes the following central claim: the ‘deep adaptation’ agenda needs repurposing: to cope with the various breakdowns and collapses that AI is likely to mediate.
And to be clear: you don’t have to be convinced that the phenomena I have described above are even likely, for it to be rational for you to begin to contemplate them and how we might cope with them. It is enough to be convinced that they are possible, not implausible. And they clearly are that, at least.
I hope I have said enough to show already the incredible potential for ‘going back to the future’ that follows from a realistic appraisal of such AI threats. The extraordinary irony of the rise of AI is that it will make return to a number of ‘old-fashioned’ technologies – such as simply meeting, face to face – newly attractive, de rigeur.
Reader: You may just not be that interested in AI. You may be thinking that its world is irrelevant to you. But here’s the thing: while you may not be interested in AI, >it is interested in you<…
Let me be clear: unlike Paul Kingsnorth, I intend this as a metaphor. (AI isn’t strictly speaking interested in anything. There is no there there.)
But mine a metaphor that is easy to parse: AI is interested in you in a similar way to what Trotsky meant when he said, in the original quote, “You may not be interested in war, but war is interested in you.” Think total war…
The kind of transformative changes that I have gestured at in this piece are in some cases desirable. Consummations devoutly to be wished. My point though is that they are not going to be voluntary. They are going to be imposed upon us whether we like it or not. The task then is to turn the imposition as much as possible into something welcome.
To transform the needful response in the direction of flourishing, and of meaning.
The great work of our time is to respond creatively and in a way that doubles down on our humanity (and our animality) to the interlocking existential threats to which we are exposed. The threat which is rising the fastest right now is that from AI. We will not begin to meet it unless we perceive it, and understand how our own ongoing quest for existential meaning is its best antidote.
This newsletter is a reworked, updated, and expanded version of an article I wrote for the Dark Mountain Project.