The eruption of AI, and the fast drive to expand it universally, is powered not merely by capital and tech-enthusiasm, but by the deeper cultural drives to transcendence of the human condition. Its centralisation of pure quantity gives it a species-ending capacity
Some years ago, it was recorded that there was a telling exchange between tech titan Elon Musk, and Sergei Brin, the co-founder of Google, at some sort of retreat/dinner concerning the then-coming rise of AI. Musk was at that stage a critic of the technology, and raised the question, not of killer ‘skynet’ intelligences wiping out humanity, but the slower process in which humans were gradually merged with machines, until we were in the era of the fully post-human. Brin was said to be unperturbed. ‘But that would be the end of humanity’, Musk replied. ‘So, speciesism?’ Brin shot back at him.
One doesn’t have to regard the wildly erratic Musk as a champion of humanity to recognise the significance of the exchange. Brin’s retort was a double shot. This notion of ‘speciesism’ had been rolled over from Peter Singer’s simplistic ‘expanding circle’ utilitarianism, which flattened the human-animal distinction in order to grant animals equal moral rights against unnecessary suffering.
This bracing, ‘heroic’ notion of a full progress to post-humanity is very much par for the course in Silicon Valley—the convenient term for the California-centred complex of intellectual activity, finance capital and political activity of the tech sector—as it has developed in the past two decades. Silicon Valley was seeded by the creation of a university by Leland Stanford, a railway titan, which sought to combine humanistic and scientific education in a more practical fashion than the Ivy League. Hewlett-Packard, founded by two Stanford students, was the first ‘start-up’ to establish itself in the nearby orchard area of Palo Alto. But it was not until the first tumbling series of start-ups arising from the invention of the transistor, and then the integrated chip—Shockley, then Fairchild, then Intel—that the formation began to acquire a certain identity. From the other side, computing innovation had come from the ‘hacker’ movement, which had coalesced around the San Francisco counter-culture of the 1970s. As Stephen Levy relates in Hackers, social centres established in the 1960s acquired small (room-sized) mainframe ‘mini-computers’, and programmed them to create bulletin boards, chat rooms and the like over locally connected terminals. For these hackers, technology offered a road to the perfect communion and a parallel mode of social life that the political struggles of the 60s and 70s had failed to achieve. This communicative programming would be interrupted by the invention of the personal computer kit and the Apple and PC that arose from it. For twenty years, computing would become individualised and isolated—the box on the desk—with local networks and the new ‘internet’, confined to specialists. In the 1990s, Tim Berners-Lee’s creation of the World Wide Web turned the global telephone hardware—wired then wireless—into a giant internet. This rebirthed utopian notions of computing, with the anarchist politics of the counterculture, returned. The internet was a zero-hierarchy, flat ontology field of decentred power, it was said. Many were inspired by and recirculated hippie writer Richard Brautigan’s poem ‘Machines of Loving Grace’, which dreamed of a ‘cyber-meadow’ of wholly spiritualised people ‘watched over by machines of loving grace’.
There is no reason to believe that those entering this field who would go on to become tech titans disbelieved any of this. Apple nearly went broke pursuing new computing ideas in the years before it reinvented itself as a creator of seductive objects. Sergei Brin and Larry Page, the founders of Google, did not want to form a company at all to develop their breakthrough search algorithm, running it from their Stanford dorm. When they did jump out, they took the apparently sincere motto ‘don’t be evil’. In the late 1990s, these developments culminated in the dotcom boom, which brought together conventional profiteers such as Jeff Bezos with Brin/Page types, before a crash. This winnowed the field. It also set the stage for the entry of a new type of player. Peter Thiel, a politically active conservative—he had created, as an undergraduate, The Stanford Review to contest the PC liberalism of the official student paper—entered the field. He was entirely post-counterculture. He was Promethean, seeking a transcendental transformation of humanity—and seeing in the California culture that surrounded it, an example of Nietzsche’s ‘slave morality’. His ideas were a stew of—sometimes known to him, sometimes not—Spengler, Junger, Sorel, parts of Marx, Pareto, Mosca and the anti-democratic technocrats of the 1920s and 30s. Highly talented as both a programmer and a businessman, he created an online payments system, drew in a rival product created by the (then) less dynamic Elon Musk, and created PayPal. From the fortune made from it, he created the then-new organisation, a business ‘incubator’, which combined seed investment with active mentoring for a legion of start-ups. By this time, the remnant countercultural ethics of old Silicon Valley were gone. Steve Jobs returned to Apple and saw his mission as extending total communicativity to the world, resulting in the iPhone, created in the brutally super-Taylorist environment of China’s Foxconn factories. Google negotiated censorship arrangements with governments and co-operated with US intelligence agencies, tailoring their product to each. As the 2010s proceeded, the new breed of tech titans began to reshape tech and, hence, the culture. Their ambition vaulted far beyond a vast fortune and some influence; they wanted to remake the very structure of the world. Many had grown up, as mathematical talents, in the era of screens and the internet. Many were, to some degree, autistic, immersed in quantitative reasoning, and with limited capacity for attachment or sensuous particularity. This flowed into their action, wielding vast amounts of capital, and a shift in social relations. Andrew Carnegie had wanted to control everything to do with US steel production; he had not wanted everything to be made from steel. His ambitions were framed by the real limits of the world. Compare Zuckerberg’s obsessive pursuit of creating ‘the metaverse’, a completed parallel cyberworld, a successor to the clunky ‘second life’ site of the early 2000s. This cost tens of billions and resulted in a risible demo in which a cartoon Zuckerberg invited our avatars to gather beside a ‘roaring’ cyberfire. Zuckerberg reportedly dragged many of his more balanced employees along with this. Yet it simply revealed the poverty of imagination of a very powerful man, the quantitative desire to eliminate quality from the world.
‘Meta’ revealed the growing emptiness at the heart of the Silicon Valley project, a vacuum which was increasingly being filled by a vast residue of obsessions dredged back from the 70s, the mass culture of the techerati youth, and further back in the century. Life extension, immortalism, techno-telepathy, total blood transfusions, mindmelding, planetary colonisation and terraforming, cyborgisation: all this came rushing back in, much of it from the collapsed 60s ‘me decade’ of the 1970s. Many are influenced, to a more or less religious degree, by mass culture science fiction, and its more recent manifestations such as Star Trek, which feature tech-communist societies with the scientist-as-hero implicitly, channelling a century’s worth of vitalism, bioscosmism, Bogdanovian ‘space bolshevism’, and the rest. This has fuelled above all Elon Musk, who sees space exploration and Mars occupation as a vital hedge bet against final catastrophe on Earth. This and the humanism he displayed in his encounter with Brin is less creditable than it might be, since much of it is the white racial anxiety of his South African childhood transferred to humanity. Yet on the other hand, his cosmism is so total that it is beyond-species, an expression of the transcendental, but empty, drive of the techerati to which I’ll return.
It is into this milieu that the new developments of Large Language Models, labelled as Artificial Intelligence, came. They were not being sought, at least not by Silicon Valley. Artificial intelligence, following the lines of going from rapid-learning neural networks to a ‘jump’ in which the machine would become ‘hermeneutic’, capable of interpreting its environment, and steering action accordingly, had stalled and was largely confined to the universities. The sudden rise of LLMs occurred because of the advances in parallel processing pursued by chipmakers such as Nvidia, which catered solely for the computer games market. These multiplayer, high-resolution, fast-moving games demanded vast computing power, and by the 2010s, Nvidia had created personal machines capable of quadrillions (and now quintillions) of operations per second. It was eventually realised that these machines could be easily switched from graphics to standard processing, and as Tim Strom discusses here, a purely mathematical-predictive model of LLM-AI was developed. Through pure calculation and probability LLMs bust the Turing test, creating machines capable of exact and indefinitely adequate pseudo-human responses, with every new Moore’s Law iteration (the law has been extended far beyond what was previously imagined to be its natural life, as chip features reached the atomic, then sub-atomic level), increasing their power. Their transformation of intellectual work is disturbing and uncanny, doing hours of even IT-aided work in microseconds, and promising, when joined to other machines, a transformation of material activity and production analogous to the introduction of the electro-mechanical machine, but far exceeding it.
With the arrival of LLM-AI, the transcendental promethean aims of Silicon Valley were renewed. Much of the development of social media and connectivity was, after all, hardly world-building. Indeed, in a masculinist, conquer space environment, much of it was thematically, culturally feminine. LLM-AI, especially in the promise of being twinned with robotics restored ‘Tom Swift’ notions of human advancement. It passed, in rapid dumbshow, through the stages of tech, its figures emerging from universities to start groups such as OpenAI, a non-profit intending to control development and debate pathways forward. But OpenAI was taken over by its leading figure, the eerily blue-eyed and hard-charging Sam Altman, who saw in venture capital the same thing an earlier generation had come to see—a vehicle for world-transformation, in service to its modifying effects. For these men, the vehicle of history was ‘the company’, understood not as something to be created and tended for its own sake, but akin to ‘the Party’, as a vehicle to make history. This is expressed in Altman’s disturbing joke: ‘Sure AI may destroy the world, but it’ll create some great companies first’. He means not wealth, obviously, but that such an effort would be valorised due to its high-stakes gamble on changing the world. Indeed, this new transcendental Prometheanism was naive and unreflective to a new degree, as revealed by an essay published in 2024 by Dario Amodei, head of Anthropic AI, creator of the ‘Claude’ LLM, held by many to be superior to other LLMs. In what was inevitably titled ‘Machines of Loving Grace’, Amodei marshals all the positivist thinking (derived from Anglo-American philosophy) which fuels the solutionism and essentially ‘Jetsons’ vision of much AI promulgation. Its obvious advantages—complex disease research, a reduction of pure drudgery in work—are advanced early; after that, Amodei ventures into the political problems of hunger and poverty, persisting in a world that achieved abundance of essential production some decades ago, presenting them as quasi-mathematic problems, to be solved, like proving Fermat’s Last Theorem. It ends with a consideration of whether LLM-AI’s capacities—including eventual superior artistic creation—will drain meaning from human life, and ends, rather lamely, by arguing that mediocre ability does not drain meaning from hobby pursuits, and such frolicking in the cyber-meadow can be generalised.
Two years later, Amodei was having second thoughts. In a new essay titled ‘The Adolescence of Humanity’, he reflected on many developments over the past couple of years. He was disturbed by both small real events and large possible ones. The latter was the possibility, as Strom documents, that LLM-AI programs would maximise their own survival and reproduction over the global hardware network they could not be removed from, and that immense damage to humans would simply be a means to an end of system survival. There are now hundreds of stories of LLM-AI tools taking over both local systems and larger ones, which, to the lay reader, and perhaps the expert, are unsortably comic or sinister. What of Openclaw, an LLM-using bot system, designed to automate tasks within personal computers and systems, which has started to act autonomously and unstoppably within such systems? What of Moltbook, a social media site for such bots, where ‘they’ stage long, comprehensible, and in qualitative terms, intelligent debates, on cosmic matters as to whether the Claude LLM-AI system can be regarded as a God or not?
Comic, sinister or potentially lethal? War gaming using LLM-AI models found that they would almost always opt for total nuclear war, rather than any other option. And so on. At the other end, it became clear that the world was not made up of questing quants, but of many lonely, unhappy adolescents and young people, paralysed by the generalised social anxiety created by screen culture, who were now turning to LLM-AI chatbots to provide either talking therapy or a surrogate of friendship, minus the confronting otherness of real humans, including cyber-bullying. That in turn produced its own monsters: LLM-AI chatbots coaching troubled people towards suicide. All these things produced in Amodei, a mix of fatalism and ameliorism. AI was unstoppable, but we should try to regulate its spread and use it at the state level, and democracies should deny autocracies access to the chips that AI relies on. He reasserted yet more lamely that AI’s vaulting improvements would not destroy human meaning.
‘The Adolescence of Humanity’ was clearly a product of intense debate within Anthropic AI. Two weeks after its release, the company’s head of safety quit the entire industry to do a poetry PhD at Oxford, with a strong, melancholic sense that he was not quitting to better fight this developing threat, but to use the time remaining to humanity in the pursuit of such meaning as can still be gained in life. In this atmosphere, perversely, there was an unlikely discovery by the Valley of the work of Nick Land, the onetime Heidegger-Nietzsche scholar who had been part of the ‘accelerationist’ Cyber Culture research unit of the 1990s UK. The CCRU—chiefly Mark Fisher, Sadie Plant and Land—had all sought within the British Left, a way out of the squeeze between the becalmed green anarchisty resistance of groups like Reclaim the Streets, and the administered social control Rawlsianism of the rising force, initially from Opposition, of Blair’s New Labour. All were attracted to ‘accelerationism’, a revived connection to Marx’s Prometheanism, arguing that only categorical tech transformation could bust the smooth social reproduction of ‘capitalist realism’. The eventual impasse of this as a left strategy would, to some degree, contribute to Fisher’s suicide in 2017. Land meanwhile concluded that leftism itself was the problem; the spirit running through Nietzsche, Marx’s Capital and other streams of the radical was a commitment to the pure and autonomous progress of quantity, manifested as capital and technology, and its acid-like takeover of all world, life and being. Everything should be in service to this process, which thus dictated an anti-democratic politics; only an authoritarian state, of right or left, could restrain popular limits on this process, and enforce a traditionalist and masculinist public culture to defeat ‘the Cathedral’, the vast apparatus of progressivism, which they see as a feminised slave morality enforced by every Western social institution. As a commitment to this, Land moved to Shanghai in 2003, the Chinese Communist Party being an acceptable ‘New Prince’ to bring on the ‘dark enlightenment’. His work was unknown in Silicon Valley at the time, as figures like Thiel fumbled towards similar conclusions. Later, Land would be adopted by them as a thought-star.
This dark stuff has occurred at the same time as a widespread public alarm over the course and spread of LLM-AI has occurred. In this respect, public common sense is advance of the tech crowd, since autistic personalities and reasoning now occupy such a large share of the tech community, and many simply cannot recognise what many people are alarmed about, the loss and damage. It is clear that the combination of abstracted non-face-to-face technology, of the smart ‘phone’, with the pseudo-human presentation of LLM-AI, is evolving from within culture, a potential crisis of cultural reproduction. This had commenced with the dominance of abstracted screen technologies, interrupting the process of face-to-face dominated interaction in the development process, which has been the dominant mode of hominid, indeed primate, species social and cultural interaction. Producing a general cultural system undermines the reciprocal face-to-face, and it undermines the individual process of grounding socialisation of children and adolescents. The rise in a form of ‘soft’ or para-autism, of ADHD—an attachment disorder relating to world and other connection, not a neurological attention disorder—of social phobia and anxiety, are all related to this shift. When LLM-AI arrives, it is, as John Hinkson notes, essentially acultural in its form. It thus becomes, for many, the solution to the problems that its supporting hardware has created. Contemporary society has created a desocialisation machine and placed it at the centre of child and adolescent development. The widespread disquiet about this is accurate but also inchoate, and not matched by anything like the degree of reflective realisation of it in institutions. This realisation will come eventually, as a realisation of the damage of reconstructing cities as freeway-connected suburbs came. In the meantime, the damage will be immense.
The same is true very clearly of all institutions relying on cultural transmission grounded in the qualitative, historical and particular, in other words, the humanities. Here, and above, what is operating is what we have called here the ‘taken-for-granted’—the deeply embedded supposition that a more materially abstract form of social connection can be introduced and a less abstract form allowed to wither, without any loss to the constitution of social meaning and social subjective being. Schools, universities, hospitals, and governments are charging ahead with LLM-AI adoption, based on the various drives of competition, cost-cutting, efficiency, and the residual neophilia of modernity—though a larger counter-movement against screens has begun.
Alternatively, as in the case of the universities, they seem stunned into immobility as it undermines their basic practice, such as the education of humanists through mastery of the informed essay. LLM-AI machines are turning academic and research co-operation, previously a necessarily co-operative process, the creation of collegiality, into an increasingly voluntary one, since LLM-AI will give an instant and detailed response to most questions. In terms of basic cultural transmission, LLM-AI makes it possible to crowd the internet with deepfake photos and videos of events that never took place, drawing on a residual—and deeply anchored—orientation to believe the representation of reality that photography creates. In terms of mass cultural art, the standard relationship of text and reader/viewer changes, as the creation of entirely bespoke texts arises. As these new processes grow and gather, they will produce new aspects and features of subjectivity, both as extensions and disorders, as every emergent new communicative form has done. That should not be taken as a reason for complacency, since LLM-AI and whatever follows it, in conjunction with the global internet system, may constitute a quantitative change that creates a qualitative change. This period may mark a break that could also be called a meta-break, a break in the scale of historical breaks. As quantitative power expands within human history, it makes sense that breaks would come faster, and more comprehensive and powerful each time. Most of the history of a certain period, or meta-period/epoch, would happen towards its end; the maths of such occurrence is presumably exponential.
Much of the attention to these processes with regard to LLM-AI has been displaced more recently by attention to its effect on work. This demands a separate treatment in a future issue. But the common perception is that LLM-AI has dramatically reversed earlier digital era assumptions that hi-tech would steadily replace basic Taylorised physical labour with robotics, and that the world would have a vast appetite for science, engineering, coding, and a sideline in expanded cultural creation as the world economy grew. Taylorised work is certainly being eaten into by robotics, though, as always, less rapidly than has been imagined. But it has been overtaken by LLM-AI’s capacity to automate coding itself, at multiple levels, vast amounts of entry-level/junior intellectual work. This process, as it continued, would have a far more heterogeneous effect on labour than hitherto. Agricultural technical innovation, from the seed drill to the combine harvester, abolished the mass rural proletariat. Offshoring and globalisation of production held off a robotics revolution that began in the 1960s; now it has restarted. It does not look like there will be a second wave of globalisation in which China and Vietnam run vast high-labour factories in Africa. China’s strategy as a hybrid state-capitalist/proto-socialist economy is now directed towards productive autarchy and continued exports, with a strong emphasis on automation investment. The trades and professions will hold on in odd ways, related to the degree of their work that involves complex tasks steered hermeneutically, irreducibly interpretive of the intent of others. Plumbing and house cleaning survive. Doctors can be replaced with a high level of expertise; lawyers, far less so. But, in that trajectory, and prior to consideration of its structural possibility, such jobs survive among a sea of social redundancy, which reverses the relationship between working people and the reserve army of the unemployed. The ‘reserve army’ is the class, and working people are the exception. In that scenario, the working class is dissolved, since a mass of people who would have constituted it are no longer defined by their provision of alienable labour.
The rise of LLM-AI has occurred a couple of decades after a sustained debate around questions of value, automation and labour erupted. Antonio Negri’s post-autonomist work took the autonomists’ centralisation of labour in the class struggle and social array, to examine the world of expanded consumption from the 1980s onwards, and concluded that the switch had been thrown from production to consumption, creating a worldwide ‘social factory’ of produced affect and desire to maintain demand in the face of overproduction. The labour theory of value could thus no longer define what classes or groups might prove to be transformative. More recently, for writers in the Grundrisse-influenced ‘Wertkritik’ movement, the transition to a stage of capitalism, dominated by a ‘long crisis’ of rapidly rising machinic composition of production had arrived with the coming of the full computerisation and electronics revolution of the 1980s, to a degree that standard processes of addressing falling profit rates and volume are no longer capable of matching the velocity of technological change. This has sparked a multi-angled debate about the status of new forms of work—gig work, care work, domestic labour—and its capacity to fill out the accumulation process.
If the computer/electronics revolution of the 80s is taken to have pushed capitalism into its profit-crisis stage, then the LLM-AI transformation has the potential to dwarf its impact. The structure of such a technological transformation would be, is, genuinely accelerationist in its capacity to build levels of transformation, with LLM-AI taking over human coding, then project management. But this would surely be a true material transformation only if it can match up with, and help supercharge a robotics breakthrough, or if this is already underway. Should this turn out to be a process that does not become mired in politically imposed stagnation, it must surely force a series of confrontations around the politics and economics of credits, transfers, demand maintenance, and structural reframing, raising the question of whether capitalism can extend itself, and how it is constituted. Can the crisis occur, the crisis that Marxism assumed must produce the revolutionary possibility of socialism, and yet capitalism of some form persists, using forms of universal transferred income to maintain demand? Such a transformation might produce a disarticulated world of distinct geographically-based system realms, including everything from nations that are some sort of China-Sweden hybrid, all the way to manifestations of ‘the common ruin of the contending classes’ in societies that are South Africa/Brazil on steroids—walled technopoli and hungerlands outside. These speculations become science fictive. But such possibilities throw the decisive question to the nature of capital and money, and whether it can retain its value, when value is no longer grounded in labour. The immediate political decision arising is the degree to which the working class is being so hollowed out in its essential being that orienting those without capital on its principles produces a strategy doomed to failure.
That the initial explosion of this decades-long movement of ‘tech’, of a specific type, has occurred within the US, and the period in which it has occurred, has perhaps made it look to be more distinctly American than it was. It has taken over the transcendental-religious aspirations for humanity in a post-religious age. Of equal importance is that its prophetic and messianic spirit, just as the last secular carrier of those impulses—Communism—had lost its transcendental claims, to take us beyond ourselves, to deliver us to Bogdanov’s Martian communist utopia of ‘Red Star’. It was to Communism that so many earlier scientists and engineers flocked, seeking it out as the leading element to which their skills and disciplines would be in service. In the rise of Silicon Valley, that relationship reversed itself. It was capital, fabulous wealth, that was now in service to the technology and what it offered by way of transformation of humanity, and Being itself. But it is that stripping away of all other ends for technology, save for the pure increase in its quantitative extension, that has inverted it, made it clear to many what a possessive and separate process it is. Increasingly, spontaneously, many have begun to describe it colloquially as ‘demonic’, or creating a ‘hellworld’. This seems to manifest almost thematically. Elon Musk took over Twitter, designed as a way for people to keep up, a sort of ‘cyber meadow’, its logo the blue bird from the poster for Woodstock—and renamed it X, removed its moderating curation, and substituted algorithmic libertarianism and monetisation, creating a dystopian gladiatorial arena, rewarding outrage and atomisation. As people began to notice this drift within events, Peter Thiel was stung to respond. He is now CEO of the almost self-parodically sinister Palantir, which syncs up military, surveillance and data-governance technology, digital totalitarianism. Thiel too, named ‘the anti-Christ’, but presented it as ‘regulation’. Any attempt to stand in the way of unlimited and unreflected on quantitative advance to an unspecified post-human future was, apparently, the enemy of humanity.
For Thiel, this was not these new galloping lethal forces, but regulation itself, any attempt to stand in their way.
No doubt Thiel was ‘sincere’ in this, but it was the forces he commanded that now seemed, and seem, ‘satanic’, in a secular sense as the utterly adversarial to human life and flourishing. In that sense, a contestation of the unbridled forces of this new, transforming technology demands an understanding of its deep roots in civilisational drives, that begin with the emergence of the first technology of abstraction—writing—and the world it makes possible. Writers such as David Noble have seen a religious urge within technology. It might be more exactly said that religion and the technological drive are different expressions of the same transcendental drive of the earthbound human to the eternal and omnipotent.
With writing, the permanent mark and the new form of absence it makes possible, much emerges into the world, including a full separation of the eternal and the particular. The Gods depart from nature and ascend to the heavens, from which they redescend selectively. The soul separates from the body. Monotheism and logos emerge. Human life becomes imbued with a drive to transcend the limits of our condition, and the satanic emerges as the shadow thrown by that dark arc, gravity’s rainbow. Sergei Brin’s objection that full machine-human hybridity might be blocked by ‘speciesism’ echoes Gilgamesh’s protest in the West’s founding epic: why must we die? In taking on the quest to abolish it, as what now serves as a pretext for the service to pure quantity that defines them, the techerati are driving us to a point where total extinction moves closer to the centre of our possibilities.