From the Machine of Lagado to ChatGPT: Descartes, Swift, Harris, and the Dream of Automating Language
 |
| The Machine of Lagado, lithography. AI image |
— George Berkeley, Alciphron, or The Minute Philosopher
Introduction
In the third part of Gulliver's Travels (1726), Jonathan Swift describes one of the strangest inventions in literary history. At the Academy of Lagado, Gulliver encounters a gigantic apparatus composed of wooden frames covered with words. Students turn handles, rearranging verbal combinations at random. The machine's inventor proudly claims that, with sufficient effort, anyone could produce books on philosophy, law, science, and politics without the inconvenience of study.
The episode was intended as satire. Yet nearly three centuries later, readers find themselves in an unusual position. Machines now translate documents, summarize articles, answer questions, and generate essays. What once appeared absurd has become technologically plausible. The dream of automating language has travelled a remarkable path—from philosophical speculation to literary parody, and from parody to engineering.
Descartes and the Test of Language
Long before Swift imagined the Academy of Lagado, René Descartes had already identified language as a decisive boundary between humans and machines. In the Discourse on Method, he argued that no mechanical device could participate in genuinely open-ended conversation. An automaton might produce a limited range of responses, but it could not adapt meaningfully to whatever circumstances arose.
For Descartes, this capacity revealed something fundamental about human intelligence. People routinely produce sentences they have never heard before and understand expressions they have never previously encountered. Such flexibility seemed irreducible to purely mechanical operations.
Language therefore became more than a means of communication. It served as evidence of a distinctive form of thought. If a machine could genuinely converse, one of the strongest arguments separating mind from mechanism would begin to collapse.
A century later, Swift transformed precisely this philosophical problem into satire.
The Machine of Lagado
The Academy of Lagado is filled with eccentric projects, but none is more memorable than its linguistic contraption. Swift describes a large frame containing words arranged on movable blocks. By rotating handles, assistants generate new combinations. Whenever an interesting sequence appears, scribes copy it down. The inventor believes that, through enough repetitions, entire libraries could eventually be produced without the aid of learning or reflection.
The target of Swift's satire is often misunderstood. He is not merely mocking scientific inquiry. Rather, he is exposing a particular intellectual temptation: the belief that knowledge can emerge automatically from formal procedures.
The apparatus promises to eliminate judgment, creativity, and understanding. Instead of thinking, one merely manipulates symbols until meaningful combinations appear. The result is a parody of scholarship in which production replaces comprehension.
What makes the episode striking today is its resemblance to contemporary discussions about artificial intelligence. Modern language models do not operate through rotating wooden blocks, yet they also generate text through the manipulation of symbolic relationships on a massive scale. Swift's fictional invention anticipated, in exaggerated form, the possibility that linguistic output might become detached from human reflection.
His joke has acquired an unexpected second life. What began as satire now reads, at moments, like a technological prophecy.
Harris and the Computational Turn
If Swift’s machine turns language into a combinatorial game, later developments in linguistics and computing gradually transformed that satire into a conceptual possibility. The question, however, is not only whether machines can produce language, but what counts as “understanding” language in the first place. It is precisely this reconfiguration that concerned Roy Harris when he published The Language Machine in 1987.
Personal computers were still relatively new, yet he already sensed that a deeper shift was underway. Computers were not merely transforming communication technologies; they were reshaping the conceptual framework through which language itself was being described.
Understanding increasingly came to be framed in terms of information processing. Communication appeared as encoding and decoding. Meaning seemed reducible to the manipulation of symbols according to formal rules. As these ideas spread, the machine ceased to function merely as a tool and began to operate as a model of the mind.
Harris viewed this development with caution. He questioned whether linguistic activity could be adequately captured through computational metaphors. Human communication, he argued, depends upon contexts, purposes, and situations that resist reduction to formal operations.
His concern was not simply whether artificial intelligence would succeed or fail. More fundamentally, he asked what it means for human intelligence when computational models cease to describe machines alone and begin to define the very vocabulary through which intelligence is understood.
Seen from the perspective of the present, Harris was writing at the threshold of a transformation whose consequences are now impossible to ignore.
The Joke That Refuses to End
The path from Descartes to Swift, from Swift to Harris, and from Harris to contemporary artificial intelligence reveals a curious historical pattern. A philosophical impossibility became a literary joke, and the joke gradually evolved into a research programme. The research programme produced technologies that would have astonished all three thinkers.
Yet the central question remains unresolved.
Machines can generate convincing prose, answer questions, translate texts, and sustain dialogue. In that sense, part of the dream of automating language has undoubtedly been realized. Whether those achievements amount to understanding is another matter.
The Machine of Lagado no longer belongs entirely to fiction. Its descendants are already among us, producing essays, conversations, and explanations at a scale Swift could scarcely have imagined. The old inventor's ambition has been fulfilled in ways both impressive and unsettling.
What remains uncertain is whether the modern heirs of Lagado understand any of the words they produce—or whether, despite all the technological advances, we are still turning the same handle.
Related Post
Beyond Gulliver’s Travels: Lessons from the Grand Academy of Lagado
https://derridaforlinguists.blogspot.com/2024/08/blog-post_28.html
References
Berkeley, G. (1732). Alciphron, or the minute philosopher.
Descartes, R. (1637/2006). Discourse on method (I. Maclean, Trans.). Oxford University Press.
Harris, R. (1987). The language machine. Duckworth.
Saussure, F. de. (2011). Course in general linguistics (W. Baskin, Trans.). Columbia University Press. (Original work published 1916)
Swift, J. (2008). Gulliver's travels. Oxford University Press. (Original work published 1726)
Comments
Post a Comment