ChatGPT and the Dream of Universal Knowledge

This piece was originally delivered as a speech at Pusey House's conference on Intelligences: The Making an Unmaking of Humans at Exeter College, Oxford, in February 2024. Published in Røyst.

AI and the Idea of the “World Brain”

With the rapid growth of AI has come the promise of something Western civilisation has long dreamed of: a single, self-contained body of all knowledge available to mankind. From as early as the Middle Ages, numerous attempts have been made to compile information into encyclopaedic forms, which became increasingly advanced following the 18th century French Encyclopèdists. [1] In 1914, the science fiction writer H.G. Wells updated their ideas for a technological age, proposing to bring into existence “a new social organ” which he called the “World Brain”. This global encyclopaedia, he hoped, would “bring all the scattered and ineffective mental wealth of our world into something like a common understanding”,[2] and serve as an “undogmatic bible to a world culture”. [3] Already it would seem the internet has turned Wells’ fantasy into a reality, but technologies like ChatGPT take it further by allowing users to retrieve information instantly, and without even having to undertake research themselves. The ultimate “world brain”, ChatGPT not only stores our data but also supposedly “thinks” on our behalf.

In doing so, it might seem to bring mankind closer to the attainment of a perfect knowledge. By allowing us to overcome the finitude of human memory and the burden of manual mental processes, it appears to possess an almost transcendent, perhaps even divine quality. In the first instance, the encyclopaedic urge had its origins in the Christian belief that our present state of ignorance and forgetfulness is a consequence of original sin, which we should strive to overcome as a form of spiritual redemption. This was the rationale of the 17th Century philosopher Francis Bacon, perhaps the founder of modern technology (or, at least, the ideology behind it), whose own encyclopaedic endeavours were motivated by the belief that it was possible, with the aid of machines, to get back to a state of perfect Adamic knowledge. [4] At first glance, AI might seem to fulfil this essentially theological aspiration.

We should be sceptical, however, towards this – not least because modernity is full of “false transcendence”, [5] from the self-affirming pseudo-spiritualities of the New Age to the naive utopias of Silicon Valley. With the decline of traditional faith, the West has become so hungry for any sense of the transcendent that it all too easily misplaces this hope in immanent things. Technology is perhaps a prime example of this. We ought, then, to think carefully about what AI really means for human knowledge. Upon inspection, it becomes clear that large language models will not bring us closer to genuine wisdom, but in fact could actually worsen our present forgetful condition. This is because, in replacing our mental faculties with artificial prostheses, they inhibit the human acquisition of wisdom as it was classically understood. To understand this, we ought to review the history of the Western encyclopaedic urge, and how its original intentions have been subverted.

The Christian Origins of Encyclopaedism  

Human beings have always sought to record and expand our knowledge, but it is in Christianity that this project takes on a moral dimension. In the writings of Church Fathers such as Athanasius of Alexandria, [6] we find the idea that the Fall deprived mankind not only of his innocence, but also his knowledge: whereas Adam could at first comprehend the entirety of creation, postlapsarian man has only a partial, limited understanding of things. The medieval Italian scholar and rhetorician Boncompagno da Signa believed that original sin was responsible specifically for our present condition of forgetfulness, whereby our memories are weak and easily corrupted. [7] Accordingly, to improve our memory is to take a step towards being redeemed from our fallen state. This is what motivated medieval Christians to make the first encyclopaedic volumes and develop educational methods that would maximise students’ abilities to retain facts and information.

The Christian pursuit of learning was also deeply influenced by the philosophy of Plato, who saw the improvement of human memory as a necessary precursor to wisdom. Our memories, he said in the Theaetetus, are like blocks of wax which hold the impressions of things from the sensible world. [8] The stronger our wax is, the more defined these impressions become, and so the more able we are to perceive patterns between them and identify the essential natures of things. Subsequently, we are better able to associate the ideas in our minds with the divine ideas that our souls once knew in the realm of forms. This allows us to move from the lower types of knowledge – the sense impressions and judgements which he called eikasia and pistis – to the higher, rational types of dianoia and ultimately noesis; the apprehension of universals. That is to say, Plato believed that the remembrance of things in this world is conducive to the remembrance of eternal things in the sense of anamnesis; it is in this spiritual recollection that perfect knowledge lies.

It was out of these two ideas – the Christian pursuit of knowledge retrieval and the Platonic association between memory and wisdom – that the Middle Ages gave birth to what has since been called “the art of memory”. This tradition, brought to light by the historian Frances Yates in her book of the same name, is perhaps best associated with the 13th Century Catalan mystic Ramon Llull. He created elaborate diagrams for memorising the virtues in the arts and sciences; the elements of the natural world; different types of flora and fauna; the planets and signs of the zodiac; and the attributes of God. One of these diagrammatic forms was his Arbor Scientiae or Tree of Science, “in which the whole encyclopaedia of knowledge [was] schematised as a forest of trees”. [9]  In the 16th Century, the Italian philosopher Guilio Camilo designed what he called a “memory theatre”; a heptagonal space with seven axes and seven stages, onto which forty-nine symbols representing all the different aspects and levels of reality would be projected. Both of these were designed to encourage the recollection of the whole of creation and so, in turn, inspire the noetic ascent of the soul.

Ramon Lull's Arbor Scientiae (Medieval edition)

These early attempts to organise vast quantities of data appear to anticipate the modern encyclopaedic urge, and may even be seen as prototypes of large language models like ChatGPT. Yet there is something that makes them fundamentally unique: these early encyclopaedic forms, be it Llull’s diagrams or Camilo's theatre, were designed so that human beings could internally remember the knowledge they represented. Although they are technically artificial memory aids, their ultimate purpose was to strengthen our organic memory – to “harden the wax”, as it were – so that our souls may be led to the ultimate recollection of the forms. It is only through this anamnesis, the Neoplatonic Christian tradition taught, that we can attain perfect knowledge.

Indeed, this tradition was highly critical of any innovation that weakened human memory. Famouly, Plato himself had said in the Phaedrus that writing posed a danger because it is an artificial form of knowledge retention that makes memory redundant and so – by “softening” the wax – inhibits our internal progression from the impressions of sensible things to their true forms. [10] It is this emphasis on improving natural memory that characterises the early encyclopedism of the Christian West, which was foremostly concerned with redeeming the forgetful, fallen state of mankind. As such, the technologies they developed were never supposed to be used in place of memory, but in aid of it.

Modernity and the Rise of “Information Technology”

Of course, such a paradigm can only subsist when the idea of transcendent truth itself is maintained. Without the belief in a divine reality to which the soul is capable of ascending, the pursuit of acquiring knowledge inevitably loses its moral, spiritual dimension. This is precisely what happened at the dawn of modernity. Although the philosophers of the Enlightenment did not necessarily repudiate the existence of an ultimate reality per se, they were sceptical towards our ability to perceive it. With Kant’s “Copernican revolution” came the consensus that we cannot comprehend the essential natures of things, but only how they appear to us in the sensible realm; we experience phenomena, but not noumenon, or “things in themselves” (in German, “ding an sich”). [11]  Kant rejected Platonic realism, and with it the potential for our knowledge of this world to allow us to glimpse that which lies beyond it. [12] Subsequently, especially after the industrial revolution, societies began to acquire knowledge in the sole interest of material progress, eventually abandoning the search for real insight.

This gave rise to what the French metaphysician and critic of modernity René Guénon called the “Reign of Quantity”: a culture in which knowledge is accumulated purely for utilitarian ends. As a result, Guénon said, the “vertical” dimension of existence – that which leads us to higher truth – is neglected, causing us to fixate only on the “horizontal” expansion of worldly pursuits. [13] Under this “Reign of Quantity”, the encyclopaedic project assumed a new purpose, beginning with the Encyclopèdist movement of the French Enlightenment. In the 18th century, the atheist intellectuals Denis Diderot and Jean le Rond d'Alembert made it their mission to compile an index of everything known to mankind, and did so with the overtly political aim of “liberating” knowledge from its traditional and supposedly “dogmatic” (that is, truth-oriented) religious framework. [14] From then on, Encyclopaedic endeavours ceased to be about leading the soul to wisdom and instead became about acquiring knowledge for more profane purposes – often explicitly as a vehicle for secular liberalism. 

H. G. Wells’ “World Brain” is perhaps the ultimate culmination of this. For Wells, also working within the post-Kantian paradigm, the goal of the world brain is not to lead people to higher insight; indeed, he emphasised that it should be “undogmatic”, by which he meant that it should not espouse any one idea of truth. Rather, Wells’ ambition – typical of the “Reign of Quantity” – was simply to collate as many fragments of information as possible, with which the individual could be free to do whatever he or she wanted. He could not have spelled this out more clearly than when he said, in his manifesto for the World Brain, that “a World Encyclopaedia will have by its very nature to be what is called liberal”. [15] Like that of the Encyclopèdists, Wells’ project was not designed to inspire intellectual noesis, but rather to assemble data into a functional and supposedly neutral form.

As a consequence of this transformation, the forms of technology used to store and organise knowledge also fundamentally changed in character. As we saw, the Medieval and Renaissance encyclopaedias were specifically designed to engage the human faculty of memory. This is because they deemed the internalisation of knowledge as essential for noetic ascent. With the abandonment of the very idea of “noetic ascent”, however, the perceived value of memory quickly diminished. After all, if Plato’s theory of truth is no longer taken seriously, then why would his warnings against the weakening of this human faculty be taken seriously? Soon, everything Plato had said about the danger of exteriorising knowledge was written off as superstition. There was no longer any need for encyclopaedias to be designed in their original, mnemonic format. Instead, they became self-contained bodies of information that could serve as a substitute for memory. As such, modern “information technologies” came to undermine the very faculty that was long deemed instrumental to the acquisition of wisdom.

This has a self-perpetuating effect: technologies built upon the assumption that we cannot ascend to any higher truth are inherently wired in such a way that prevents the disclosure of that truth. On this matter, the French Marxist philosopher and critic of technology Bernard Stiegler, reflecting on the nature of the internet and the digital “knowledge economy”, said – in a succinct yet rather profound passage – that “technical epochs condition epochs of noetic sensibility”. [16] That is to say, the way technology is formed determines what type of knowledge can be obtained through it, and specifically whether it encourages or discourages noesis. Modern technology, Steigler explained, is anti-noetic by its very design. This is because, firstly, it encourages passion rather than reason: algorithms are programmed to feed us content that perpetually arouses emotional and sexual hedonism, distracting us from the very pursuit of truth-seeking in the first place. Secondly, and especially pertinent to AI, they retain and organise information according to their own logic. As such, they make our memories redundant, thus hindering our potential to progress towards any higher truth.

AI takes this to an extreme because, by converting data into the simulacra of ideas on our behalf – be it a paragraph written by ChatGPT or a phantom image conjured up by MidJourney – it wholly circumvents the natural process of internalisation and anamnesis. As such, it results in exactly what Plato warned about the invention of writing, as voiced in the Phaedrus: “you give your disciples not truth, but only the semblance of truth. They will be hearers of many things and will have learned nothing; they will appear to be omniscient and will generally know nothing; [and] they will be tiresome company, having the show of wisdom without the reality.”

It is precisely for this reason that ChatGPT betrays the original, theological aspiration of recovering perfect knowledge. Instead of providing the means for noetic ascent, it offers artificial memory which leads to precisely the opposite outcome. As Stiegler argued, this technology in fact leads to collective stupefaction and, furthermore, makes us susceptible to manipulation – especially in the context of capitalism. The weakening of our mental faculties, he said, turns us into the ideal consumer subjects: the less interested we are in seeking wisdom and virtue, the more likely we are to give into impulsive wants and desires. Already we can see how this is being exploited by corporations, who are now using AI to track consumer behaviours and tailor advertisements most likely to increase sales. But the danger goes even deeper than this.  In inhibiting the rational soul, this technology also diverts us from the course of truth in a more spiritual sense. In encouraging materialism rather than qualitative insight, it keeps us trapped within the worldly “Reign of Quantity”.

We can conclude, then, that AI does not offer real transcendence, but only its semblance. In this sense, it is just like the Library of Babel in the short story of the same name by the Argentinian writer Jorge Luis Borges. He describes a fictional bibliothèque containing vast collections of books on every subject in every language, connected by vestibules surrounded by mirrors; mirrors which create the illusion that the library goes on forever. The narrator says of this illusion: “we prefer to dream that burnished surfaces are a figuration and promise of the infinite”. [17] Similarly, we may like to think that the “World Brain” – with its appearance of boundless knowledge – brings us closer to the infinite. But this too is a deception. It may, in a sense, present us with a kind of “universal knowledge” in terms of the sheer quantity of data it can retain; but it brings us no closer to the knowledge of universals themselves. Only through the noesis of the soul can we attain such a glimpse of eternity. Modern technology, in inhibiting this noesis, will only perpetuate our forgetfulness and ignorance.

 References:

[1] Blom, Philipp. Enlightening the World: Encyclopedie, the Book That Changed the Course of History. New York: Palgrave  Macmillan, 2005.

[2] Wells, H. G. World Brain. Cambridge, MA: MIT Press, 2021. 11.

[3] Ibid. 14.

[4] Rees, Graham & Austey, Peter. The Oxford Francis Bacon, XIII: The Instauratio Magna: Last Writings. Minerva XLI, no. 1 (2003): 89-92.  See also  McArthur, Tom. Worlds of Reference: Lexicography, Learning and Language From the Clay Tablet to the Computer. Cambridge: Cambridge University Press, 1986. 110.

[5] Partridge, Esmé L. K. Disenchantment, or Dark Enchantment? Røyst #21-22.

[6] Athanasius of Alexandria. Contra Gentiles and De Incarnatione. Thompson, R. W. (Trans). Oxford: Clarendon Press, 1971. 7-23.

[7] Boncompagno. Rhetorica Novissima. Gaudcntio, A. (Eds). Bologna: Bibliotheca Luridica Medii Aevi, II. 1891.

[8] Plato, Theaetetus, 191d-202c. Cornford, F. M. (Trans). In: Plato: Collected Dialogues. Hamilton, Edith & Cairns,  Huntington (Eds). Bollingen Series 71. Princeton: Princeton University Press, 1999.

[9]  Yates, Frances. The Art of Memory. Middlesex, England: Penguin Books Ltd. 1969. 187.

[10] Plato. Phaedrus. 275a-e. Hackforth, R (Trans). In: In: Plato: Collected Dialogues. Hamilton, Edith & Cairns,  Huntington (Eds). Bollingen Series 71. Princeton: Princeton University Press. 1999.

[11] Kant, Immanuel. Immanuel Kant's Critique of Pure Reason (Unabridged Ed.). Norman Kemp (Trans). New York: St Martin's Press. 1965.

[12] Braver, Lee. A Thing of This World: A History of Continental Anti-Realism. Evanston: Northwestern University Press. 2007.

[13] Guénon, René. The Reign of Quantity and the Signs of the Times (Fourth Edition). New York: Sophia Perennis. Lord Northbourne (Trans). 2001.

[14] Blom, Philipp. Enlightening the World: Encyclopedie, the Book That Changed the Course of History. New York: Palgrave Macmillan. 2005.

[15] Wells, H. G. World Brain. Cambridge, MA: MIT Press, 2021. 55.

[16] Stiegler, Bernard. Symbolic Misery, Volume 2: The Katastrophē of the Sensible, Norman, Barnaby (Trans.). Cambridge: Polity Press, 2005. 27.

[17] Borges, Jorge Luis. The Library of Babel. Hurley, Andrew (Trans). Penguin Classics. 1998. 38.