Artificial Intelligence, 1588

In this image taken from one of the Folger Shakespeare Library’s two copies of John Harvey’s A discoursiue probleme concerning prophesies (1588) (STC 12908), the early modern English physician and astrologer remarkably employs the phrase “artificial intelligence”:

This predates what the Oxford English Dictionary identifies as its first use in a proposal that John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon authored in 1955 to investigate the feasibility of machine intelligence, which would then turn into the Dartmouth Summer Research Project on Artificial Intelligence in 1956.

It seems easy enough to discount this very early example of the phrase as an unrelated coincidence due to the technical specificity of the later subject matter and the absence of formal digital computers such as the Electronic Numerical Integrator and Computer (ENIAC, completed in 1945) in early modern England. Yet, the circumstances surrounding artificial intelligence’s earliest possible invention and use in the late sixteenth century share some commonalities with its mid-twentieth century counterpart.

Early modern understandings of artificial intelligence may contribute to present conversations about AI, whether as a philosophy, an area of research, a synecdoche for large language models (like Claude or Chat GPT) or even a subject of Pope Leo’s latest encyclical.

However, to understand Harvey’s specific formulation of artificial intelligence, which he compares to a “natural” kind, it will take a bit of contextualizing.

Prognostication and Prediction

In the late sixteenth century, English readers avidly purchased and consulted books of prognostication and prediction from self-ascribed prophets.

They did so in the hopes of rationalizing the rhythms of the weather, their lives, and even possible dangers before them. The most worrisome thing the prophets addressed in the years leading up to 1588 was the designation of 1588 as a ‘annus mirabilis,’ or year of wonders or portents. While some prognosticators offered rather harmless advice, others focused on “greuous [grievous] terrors, perils, and calamities,” along with “Accidents and Consequents” (B1r).

Harvey initially began writing A discoursiue probleme before the start of this miracle year, to dissuade friends and associates from relying on these unscientific fear-mongering books and pamphlets. Claiming concern about the explosion of public anxiety from “even the learneder or wiser sort,” he sought to disabuse readers of the notion that these “prophets” possessed innate qualities to foresee such a catastrophic future. “I would gladly learne what excellent nature,” he writes pointedly to his audience, “what cunning or profound art, what active exercise, or what other general or special means of perfection should breede in them any such profeticall spirit, or appropriate habit of prediction” (B2r). Instead of focusing on the musings of untrained prognosticators, Harvey asked readers to rely on sensible predictions that were based in science.

Harvey defines the role of the prophet from Biblical times to the present, assuring readers that only those anointed by divine authority—presumably of Christ—can share knowledge in this way (he unsurprisingly excludes all contemporary Jewish and Muslim philosophers from this possibility). This, he argues, prevents current prognosticators, even those who are religious, from achieving the role of a prophet. He focuses on the end of the world calculations and predictions made by 16^th century religious figure Rabbi Elias (Eliyahu de Vidas) and other self-identified prophets.

Encryption, Decryption, and Artificial Intelligence

Finding the rationale of religious figures’ calculations of the end of the world wanting, Harvey traces ancient philosophers’ thoughts on the same subject, declaring that Plato, Aristotle, Pliny and others could not comprehend “the depth of this profound divine mysterie”(E2v). He concludes that “the deepest humane science is not comparable to the divine alphabet,” and that there is no possible way to ascertain and calculate the final days of the planet, writing that these “mightie and wonderful proceedings no Poligrapher can expresse nor Steganographer decipher” (E3v). There is no kind of abstracted language to construct an answer to a question that for Harvey lacks any possible answer.

Harvey suggests that the mysteries associated with humanity cannot be reduced to poligraphy (the art of encryption) or steganography (the art of decryption), or the signs invented or used in the formation of this calculus. In other words, he could not imagine a world in which calculation and computation could make a predictive model of it or foretell its ending. There was an impossibility surrounding creation of this kind of knowledge or intelligence for Harvey, even as he anticipates an empiricist response of “Nihil in intellectu quod non prius fuerit in sensu” (nothing is in the intellect that was not first in sensation). To this philosophical truism, he asks, “how can they devise, or imagine a science of that, whose causes are concealed, and hidden from them?” (E3v).

Emphasizing the impossibility of creating knowledge or intelligence that is separate from the divine while also declaring that having intelligence from the divine does not permit access to the world’s mysteries, Harvey then asks:

Here, he seems to differentiate artificiality in two ways, as 1) something deriving from human invention to mimic something that occurs naturally, or 2) as an entity or object crafted with the goal of deception. Arriving at a conclusion that neither kind of intelligence could predict future (often dystopic) events, he debunks the possibility of any kind of intelligence having access to what he identifies as divine or metaphysical mysteries.

While not directly related to automata or other machinery, Harvey’s thinking is in line with the founders of the modern discipline within computer science. He imagines how non-human thought processes work, especially those that are fully invented in relation to predictions. Truth does not exist within the paradigm of predictive text and cannot be formed in that way. The founders of AI as we know it, did not seem to operate in that framework and instead operated under an assumption that machines would eventually be able to exhibit human-like intelligence.

John Harvey, Kenneth Craik, AI, and The Dartmouth Summer Research Project, 1956

The proverbial story is that AI’s genesis took shape in the proposal for the Dartmouth Summer Research Project on Artificial Intelligence in 1956. In the twenty-page document, mathematicians, computer scientists (although not a formal discipline of university study until the 1960s) and engineers John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon contended that “every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it.” For this group, it was theoretically possible (and a bit closer to reality in Claude Shannon’s case) for machines to employ language to plan an action, construct abstractions and “solve kinds of problems now reserved for humans and [for the machines] to improve themselves.”

In the grant proposal, the research collective outlines how they might develop intelligence in a machine. Admitting it was no easy task, they also suggested that human thought—as well as the complexity of human brains—could be reduced to “manipulating words according to rules of reasoning and rules of conjecture.” Their previous work in mathematics, cognitive science and a survey of “advanced human thought activities” guided one of their conclusions: that creative thinking resulted from the injection of “randomness” into the act of straightforward problem solving. In effect, the randomness intersects with the kinds of predictions that Harvey warns 16^th century readers about.

John McCarthy was particularly captivated by the writings of early 20^th century Cambridge philosopher Kenneth Craik, whose ideas were strangely in conversation with Harvey’s in terms of their theories about prediction. Craik believed that all levels of intelligence could be abstracted beyond the divine and that all thoughts (and examinations of thought) were designed to “fulfil predictions or confirm other experiments.” In a sense, Harvey’s writings wrestle with this idea—that it was possible for ‘prophets’ to convince people that their symbolic predictions and thoughts—however bizarre—were entirely valid and worthy of awe. The tenuous role of prediction in Harvey’s formulation of natural and artificial intelligence plays an oversized part in Craik’s understanding of human cognition and intelligence.  Both of these philosophies constitute research still being conducted by contemporary AI researchers.

Ideas undergirding the theoretical models of mental action proposed by the Dartmouth researchers include their summary of Craik’s 1943 work The Nature of Explanation, in which cognition is reduced to “little engines inside the brain, which can simulate and thus predict abstractions relating to environment.” These founders of AI were less interested in empirical knowledge, the kind that Harvey was asking his readers to keep in mind as they were leafing through books of randomized predictions. Instead, the founders of AI were more interested in knowledge that does not depend on complex human experience, but upon abstractions of that experience. The prophets of the 1580s were interested in similar kinds of output—they wanted to predict the next sequence of events using whatever random or calculated inputs available.

As radical as these ideas must have seemed in 1956, and which still seem radical in 2026, they had already been considered, at least theoretically, over 350 years prior. The myriad problems associated with this model of cognition and intelligence were staged in the late 1580s in Robert Greene’s play The Honorable Historie of Frier Bacon and Frier Bongay as a university professor seeks to program a magic automaton to speak aphorisms, or general truths, as a way to understand the future and protect England.

Friar Bacon, The Brazen Head, and Pope Leo

Throughout the play, Friar Bacon works in the same vein as Harvey’s prognosticators—and perhaps even some contemporary artificial intelligence models—as he voraciously gathers whatever knowledge he can (including that from the devil himself) to imbue it into the Brazen Head (the name of the magic automaton). His goal is to produce important mystical statements and surround England with a protective wall. Despite this constant work, Bacon’s goal is never achieved, mostly thanks to his and his assistant Miles’ inability to stay awake for the great reveal from the artificially intelligent head. The Brazen Head essentially experiences a catastrophic time out before Bacon or Miles receive and interpret its output.

As comedic as this moment in Greene’s play may seem, it also indicates anxiety about the role that predication as well as these artificially intelligent entities could possess in the 16^th and 17^th centuries.

Conversations about artificial intelligence did not necessarily originate with Isaac Asimov and 20^th century science fiction writers—concerns about its function and limits have been a part of cultural and philosophical conversations well before that. The apprehensions expressed in the play are clearly articulated by Pope Leo’s latest encyclical in which he reminds readers that their “reliance and the search for ready-made answers…weaken personal creativity and judgement,” and displace the obvious humanity lying behind the predictive nature of algorithms. In other words, all artificially intelligence relies on some kind of prediction—whether in the spirit of John Harvey or Craik and the Dartmouth researchers.  All 16^th century artificial intelligence and its 21^st century counterparts are ultimately imbued with human decision-making patterns, reliance on randomness, and other complex political structures and meaning.