The AI of large language models is a step beyond prior developments, creating an enclosing, arguably acultural, condition
Apart from another World War or climate catastrophes of global proportions—both of which could be just over the horizon—one can say that artificial intelligence has come to be regarded by many as the major threat to our way of life and our sense of security. This has developed quite strikingly, since AI is the sort of development in which, in our era, people have come to invest hopes of development that might deliver longed-for transformations. This dual quality is probably a sign of our times. It deserves thorough investigation precisely because of this combination of hope and despair. AI is especially disturbing to many because it potentially suggests a process of thorough emptying out of familiar social life, a process which leaves nothing to take hold of, only a disappearing world of known institutions—especially those that support the world of paid work essential to life. Yet at the same time, though increasingly occluded by the threat, for many it offers a radical future of hope that development can continue, that progress is still a real possibility.
The negative threat is well illustrated in a recent interview in the Financial Times with Geoffrey Hinton, a Nobel Prize-winning scientist at the centre of the development of AI, who has, as is so common today, benefited personally from his work. In his retirement, he decided that he should warn the world of what it could bring to us soon. Yet he can only speak of AI from the point of view of its effect on the workplace. Here, he is definitely disturbing: ‘It’s going to create massive unemployment and a huge rise in profits. It will make a few people much richer and most people poorer. That’s not AI’s fault, that is the capitalist system.’ Hinton is a socialist from the UK. It is important to him to differentiate capitalism from the impacts of Silicon Valley, even though he does not outline how the non-capitalist outcome could be different. Yet then he goes further. When pushed by his interviewer to consider the Silicon Valley scenario of a future of ‘embodied AI’—where humans progressively become cyborgs assisted by AI—Hinton retorts: ‘What is wrong with that?’ The interviewer counters, ‘We lose a sense of ourselves and what it means to be human.’ Hinton responds: ‘What is so good about that?’ Again, the interviewer responds: ‘It doesn’t necessarily have to be good, but we won’t have it anymore, and that is extinction, isn’t it?’ After a long pause, Hinton replies, ‘Yep.’
Hinton’s view is probably typical of those who know AI primarily from the standpoint of science and technology, and have thought about its future trajectories, but have not thought about its fully social aspects—especially the role it might play in the process of social formation. He does not see the process in terms of the cultural formation of the person as a full human being. The person and humanity are presumed to be stable reference points, able to be taken for granted, no matter what happens. Though it seems almost certain he would not know of it, Hinton’s view of humanity is similar to that of Bruno Latour in his work on Actor Network Theory, in which he seeks an understanding of the human from the standpoint of ‘hard’ science, dissolving distinctions essential to a fully social conception. For Latour, whose hugely influential theoretical work anticipated to some degree the rise of global networked life, this conception disposes of the need for a practical notion of the social. Hinton’s analogous untheoretical rendering allows him to view the effects of AI simply in terms of the destruction of social institutions crucial for the economy.
It is equally interesting that those who are deeply committed to the AI revolution and wish to articulate its positive possibilities also have no way of taking hold of it in genuinely cultural terms. Equally, however, those hitherto constrained within a political-economic paradigm have invested utopian hopes beyond the economic. In the recent edition of Quadrant, Oliver Hartwich, a veteran classical liberal Thatcherite, makes a case for AI, arguing that it allows humans to extend their capacities.
Like Hinton, Hartwich is preoccupied by the significance of the AI revolution. He quotes from the former CEO of Google, Eric Schmidt: ‘AI is underhyped. Within five years, he predicted, nations might literally bomb each other’s data centres to prevent AI dominance. America alone needs ninety gigawatts of new power generation—that is ninety nuclear plants’ worth—just to feed these systems…’ Then, Hartwich notes, ‘came the kicker: Schmidt called this the most important development in 500 years. Maybe a thousand. The biggest thing since the printing press, perhaps since writing itself.’
Hartwich outlines his case by referring to multiple examples of what is actually happening: revolutionising drug discovery; reducing the time by half for drug trials, tripling the density of lithium-ion cells, and, of course, recursive self-improvement and even the emergence of digital agents as opposed to chatbots. He concludes that by and large, along with some disruption, AI ‘does not change people. It reveals them. It amplifies them.’
Beneath their different assessments of the danger or hope from AI, there is no real difference between Hinton and Hartwich in the evaluation of its significance as a system of instrumental practice of immense power. Differences largely reduce to whether this leads to the destruction of human societies, especially of possibilities for employment (the reasonable concern of Hinton, the socialist, uneasy about the technological Armageddon he is deeply associated with), or whether AI can lead to the enhancement of society. What both these intellectuals miss in their analysis is what I would call the cultural level of human formation. Both take our humanity for granted and therefore grossly misinterpret the significance of what is happening today.
It is important to put AI in its proper context. Though it is a fundamental development in the history of technology, it is something more and other than what one might call ‘stand-alone technology’, as could be said about the major developments of the printing press or even writing. This requires that we place AI in a different context that gives distinctive meaning to our times today. AI is the latest example of a string of more recent technological developments that arise from an intellectual revolution in the first instance: one with its roots in the fundamental transformations of the early 20th century or thereabouts. This revolution, based in the new knowledge of ‘deep’ nature uncovered by the sciences—as found with the splitting of the atom that produced atomic energy and atom bombs, or the manipulation of genes in order to shape animals, plants and humans, or the revolutionary world of chemistry—allows intellectual practices, as a social form, to change their historic relationship to society in general. When transferred from science itself and from its initial use by the state in the world wars, it allows intellectuality to enter the process of value production. This occurs fundamentally after the Second World War, when the massive developments of the state’s scientifically revolutionised total war machines were transferred to Cold War and capitalist (and, in the USSR, existing socialist) process, sustaining a systemic process of technological development that has culminated in the increasingly frequent tech revolutions throughout the second half of the 20th century, the fifty year passage from boxy personal computers through globalised networks, and vast computer capacity to this day. It is a systemic process, one that arises out of abstract intellectual ideas that previously had no, or little, role in the world of practical endeavour that transformed nature for human use. Politically, it is what has displaced the working-class movement, successively depriving it of the power to stage a revolutionary overturning of a production system, which is steadily less reliant on it.
More recently, in the process, it has started to seriously diminish the role of the humanities in academic institutions. This has far-reaching consequences: for example, the role played by intellectuals in interpreting and overviewing the directions of social development, and feeding their results back into social debate, begins to unravel. Thus, increasingly, social directions are at the mercy of the short-term thinking of politicians, the powers of scientific intellectuals and, of course, capitalist rationality. It is precisely in this process that many of the AI and tech titans—the Altmans, Musks and Zuckerburgs—invest their most ‘utopian’ hopes: a process of social steering wholly performed by the application of massive Big Data findings to social questions. Humanistic interpretation—involving debated theoretical frameworks—is rendered irrelevant.
Despite all this, the possibility of an AI superintelligence, raised by Hinton and many others, is not the issue in my view. It is not a proper assessment of real possibilities. Yet the threat of AI leading us into catastrophe remains.
Human intelligence and moral action in the world require humans to be cultural beings in a way that the tech titans cannot grasp. Such a being cannot be reduced to a simple accumulation of instrumental practices gained via the computer program, whatever its powers. A cultural being is of a different order.
Cultural being requires a model of far more complexity than is of interest to the tech titans or Latour or Hinton. Historically, this has been pursued via a combination of a distinctive form of social science, which asks the question as to how, via cultural processes, a society and the interiority of its subjects are constituted. This question is unknown to socially interested technoscientists like Hinton, and explicitly rejected by a theorist like Latour. The work of George Herbert Mead is particularly relevant and illustrative here. For Mead, cultural being emerges through the internalisation of the moral and practical views of others who are members of a group with boundaries (what he calls the Me). This internalisation process is not a simple one. It requires internal interventions (by the I) that seek to achieve resolution of internalised contradictions. And as Geoff Sharp emphasised (see, for example, ‘From Here to Eternity’ at arena.org.au and related discussion), a cultured being requires a significant grounding in face-to-face relations where the tangibility of the other helps shape the emotions via the continuity and predictability of others. This orientation towards a tangible world develops complexity over time, especially through the combination of generational relationships, including the valuing of ancestors, that tie past and present together.
If machine learning from large language models can be said to be somewhat closer to human action in the world than was found in earlier forms of artificial intelligence, it should still be regarded as a world of a different kind, a world incommensurable with cultured (that is, enculturated) beings. While language is an indispensable component of cultural formation, culture is of a different order than what can be found in the massive texts of large language models, upon which developers of AI depend. The ‘Turing’ effect of LLMs, their capacity to indefinitely present as human in an exchange, obscures the fact that they are, in the chain of human communicative technical development, the first entirely autonomous communicative element, arguably acultural.
However, this does not mean that the fear about making humanity redundant is without basis. While the cultured machine is not a real threat, AI can still be a threat to the role of culture in social life as such. This is one of the possible implications of the shift in the role of intellectual practices in our era.
A feature of the intellectual revolution that began to unfold in the 20th century is the introduction of technologies that facilitate the substitution of social relations of absence for face-to-face social relations. Technologies that extend the self by mediating between self and other, best illustrated by social media and the mobile phone, achieve this effect. In this circumstance, the ‘other’ of ‘self-other’ need not have a tangible existence; rather, they may merely be a fleeting other, with no more than momentary existence.
Media of this kind are not new. One need only think about the role of the printed word in displacing a proportion of face-to-face relations dependent on oral communication historically, to see the process at work. But the shift in balance between relations based on technological mediation and relations based on face-to-face interaction is striking in the last century. The unfolding of this process results in the attenuation of face-to-face interaction and of culture itself.
Depending on how far it proceeds, the attenuation of culture can achieve the decline of the self and the loss of humanity referred to earlier by the interviewer of Hinton. Ultimately, this can be the ‘victory’ of processes that lead to de-humanisation, where empathy for the other is no longer substantially real. But rather than pointing to AI alone as the cause of this transformation, this cultural shift has been unfolding for at least three generations. Nevertheless, there is no doubt that AI is poised to add to this process in a highly destructive way. This is achieved by a constant thinning out of the role of the face-to-face in the affairs of the world; by massively enhancing processes in the sphere of the automatic that, previously, were embedded in human associations and meanings.
There is no easy way to turn this outcome around. It is a situation that demands much more than implementing this or that policy to ban AI via legislation. It requires a social movement of a new kind that reaches further into formative processes than previous concepts of communal endeavour. It challenges all previous formulations on the Left of how to build a better society.
This does not mean that attempts to pursue conventional left politics, which aim to achieve universal equality and opportunity for self-flourishing, are no longer important but, any such politics, which takes the conventional left route of regarding the given technological ensemble of the period as ‘taken-for-granted’ and outside the political question, will, de facto, serve the extension of that dehumanising and annihilating process. Seeing AI as a way to a post-capitalist order, twinning superfast networked, feedback-looped planning, with universal basic incomes, for example, embeds us into a tech system so totalised that it puts us on the road that leads to the end of selfhood as well as humanity. An emphasis upon change simultaneously requires greater emphasis be given to continuity in relationships, a proper respect for the past, which, for all its faults, also contained crucial elements of what it means to be human. We now know that these elements can be lost forever.
Many years ago, Ludwig Wittgenstein initiated a form of fight-back against the rule of a mathematicised world threatened by the mathematical-analytic philosophy and logical positivism posed by Bertrand Russell, as well as the Vienna Circle. His conclusion, towards the end of his life, that philosophy should be founded upon life forms that serve as the context for language games, can be seen as an important acknowledgement of the grounding of philosophy in life-worlds rather than in mathematical forms. Whatever other limitations can be associated with his work, this can be seen as a crucial focus when faced by developments, such as AI, that pay no attention to, or understanding of, the importance of everyday life in their practices—a strategy oriented towards productivity, but that can only give productivity a bad name.
Wittgenstein never formulated the life-world in a way that helped illuminate its crucial elements and make clear the way that threats could undermine it. He treated the life-world as though it could be taken for granted. But unlike in the past, our humanity is now facing real threats emanating from the systemic development process. The task for intellectuals is not to further high-tech processes already in place; rather, there is a need for a renewal of intellectual support (cultural, not financial) for those embedded in everyday life, and for strengthening bonds of association that deepen our experience of others. This would constitute a revolution: a transformation in the relationship between intellectuals and ordinary people in everyday life. And a crucial aspect of this process must lie in the recognition of the role of the face-to-face in cultural formation and renewal. This means that ways can be found that limit the impacts of processes, such as AI, by surrounding them with cultured relations that can sustain boundaries. Consciousness of the limits of AI, and the like, needs to increasingly become an aspect of shared culture.
Comments
AI does not dream. It does not taste, smell, hear, feel or see. It regurgitates what has been already written and combines that material in a possible new arrangement. It rearranges. This new combination can be useful but remains very constrained by what it can never experience as a human does. It is highly overblown as a functioning capability when considering what it can never experience or know. It may be that what we learn from it (AI) is that our human capabilities are being seriously marginalised and diminished comparatively when it would be more valuable to recognise what it is not capable of and make that very clear. The overrated AI and the underrated human reflects our own self betrayal.