Depuis 2005

Philosophy / AI - March 2016

Umberto Eco and Artificial Intelligence


Umberto Eco tells the long story of the search for the perfect language, after Babel, whose completion is actually the research on artificial intelligence.

Umberto Eco, The Search for the Perfect Language, 1997, (NB. The references to the pages are from the French edition, over 448 p., for approximation)

In 1992-1993, I attended the comprehensive Umberto Eco's conferences at the College of France, Paris, where he exposed his historical research on the philosophical idea of a "perfect language". It's always worth listening an oral presentation, laxer than an academic text, and this is already part of the question of a perfect language. Before the crowd, Eco had imposed a little essay to give a priority access. Eventually, we had been a few fifteen good pupils to do the book report requested, from Gérard Genette's Mimologiques, acquiring seats in the front row of the big conference room.

In my PhD project ongoing at the time of the conference, while responding to the philosophers opposed to artificial intelligence, I had noted that AI could be considered as the summarization of the history of philosophy. The Umberto Eco's book specifies precisely this historic search for a perfect language of reason (if not God). This was especially so to escape the theoretical misconceptions symbolically embodied by the opposition between the Adamic language and the myth of Babel. In the Christian context of European history, this quest is constantly restrained by the requirement to justify the biblical texts addressing linguistics issues. The erudition of these times was suffering of getting lost in idle exegeses like determining whether Adam or Eve had spoken first.

From the beginning, artificial intelligence is referred to by Eco, talking about the correspondence between sign and reality (p. 38) when linguistics and philosophy were still unaware of the arbitrariness of signs. Indeed, the problem of a perfect language can be understood as a self-analysis of the symbolic function which is constrained by analogical and iconic associations. Maybe the semiotic specialty of Eco historically stems from this confusion. Hence this book has a testimonial value.

The "Cabalistic pansemiotics" of the Renaissance Jews derives from their reference to the biblical Scripture and their privileged access to ancient Hebrew considered as the original language. In this context, the question was bound to arise of what language Adam was speaking and how languages have differentiated. Supporters of the "Kabbalah of names" had built a cryptologic or combinatorial analysis from letters and their numerical value (Notarikon, Temurah and Gematria). They give rise to a magic automatic generation of the world itself, while longing for the reuniting of languages by the Messiah. These proposals will be taken up by Christian philosophers necessarily subjected to the custody of this biblical semantics.

Dante will resume this program in a Christian switch (God talks to create the world and Adam to name things), while beginning to explore practical issues of the historical evolution of language and of the status of vernaculars speeches. Thus, he takes perfection as minor (in contrast to the former idea of decadence), also finding out again the universals of medieval grammarians.

His contemporary Ramon Lull built a semantic algorithmics, ars magna by permutations, with constraints to eliminate empirically false statements, or to discover the rules of reality as for converting the infidels to Christianity. This will result in vertigo before the combinatorial explosion of possibilities. And the magical excesses of the Christian Kabbalah of Ficino, Paracelsus or Bruno also imagined a secret power of Egyptians hieroglyphs. Mixture of initiatory regressions and plentiful discoveries which will set up the academic topics of the following centuries: proto-language, etymology, grammar and classification of natural languages, universal and linguistic harmony...

The discovery and interest in Egyptian hieroglyphs and Chinese ideograms led to the idea of a "perfect image language" where the model of Chinese writing, as verbalized differently in various parts of China, Korea, Japan and Cochinchina, coined an idea of universal ideography. On this principle, the German Jesuit Kircher (founder of Egyptology) invented a useless stenography and interpretation of hieroglyphs that made Eco wonders about "the ideology that led him to amplify its errors" (p. 189). In continuation of Renaissance hermeneutics, Eco sees in it a "hallucinatory device in which all the possible interpretations can be gathered" (idem). In my personal book about AI, I reminded that Umberto Eco himself, in The Limits of Interpretation (1993), retracted on the excesses of too much reader's freedom, which he advocated in his previous books (Lector in fabula, 1985...). Let's note that the frills of erudition always tend to bring Eco, because of his academic Renaissance specialization, very close to Kircher "baroque among baroque" (p. 192).

But modern times were approaching and Vico denied that the hieroglyphs contain any hermetic mysteries concealed for long by ancient philosophers (p. 193) and his meditations were likely the source of Rousseau's theories on the expressive origin of language. The secret or hermeneutics regression reappeared with the Rosicrucians or John Dee's magical language which rediscovered the cabalistic or Lullian's generativity. In his oral conference, as I recalled too, Eco unveiled the secret (reserved for the rare initiates who witnessed it) of how knowing to whom we are dealing with:

"There are very few serious books about Rosicrucians. How to recognize a serious book on this topic? If it deals with the Rosicrucians, it's not serious." (Umberto Eco, The Search for the Perfect Language in the history of European culture, Seminar of the College of France, session of November 12, 1992).

We must remember the state of knowledge in Europe at the beginning of the modern era. The Latin language was international, academic and sacred, Hebrew was seen as the origin language, and the vernacular languages weren't standardized. The discovery of other civilizations in America or Asia, their antiquity or their languages, has given birth to modern linguistics, archaeology, or relativism which is forgotten today as founding modernity against the previous dogmatism!

The collection and classification of this new knowledge had produced curio cabinets, the 18th-century French Encyclopédie and bilingual dictionaries which already prognosticated automatic translation allowing speaking any languages (p. 234). The reference to the Babel myth is a haunting Umberto Eco's book. The Kircher's coded dictionary set fifty-four primitive categories, reduced to forty-four by Schott (pp. 236-237) and Descartes also considered clear and distinct ideas of reason in this point of view. The Comenius' "panglossie" sought to eliminate the ambiguities of natural language and develop a kind of pictorial conceptual pedagogy. The deaf sign language has emerged from these artificial language projects.

Dalgrano and Wilkins have specifically developed thesauri for these semantic primitives but their artificial graphics languages, kinds of conceptual Linnaean classifications, have failed on morphology and syntax to be "readable by anyone in their own language" (p. 273). This compact ideographic formalism aimed to get a transparent language of Nature, which also expressed through signs. Eco sees in it the prefiguring of hypertext (p. 296) or, rather about Lodwick, an anticipation of Schank's Artificial Intelligence primitives (p.  302). The known consequence was the Leibniz's "universal characteristic" as a scientific language identifying thought and calculus (pp. 316-320). The French Encyclopédie will be a concrete outcome of the building of these semantic networks, but from a perspective of universal dissemination which opposes to the former Renaissance esoteric approach.

Other a priori universal or philosophical language projects will emerge during the Enlightenment: Faiguet, Delormel, Maimieux, Hourwitz, De Ria, Vismes, Sudre, Freudenthal etc. (pp. 333-351). At that time, the ideal model of Cabalistic Hebrew was replaced by the Chinese ideographic model. But eventually, the topic of unity of thought and language has mainly produced the romantic conception of national language as determining culture. The French author Rivarol also pitifully solves the issue by considering the French as the already perfect universal language!

The last chapters of Umberto Eco's book only developed the issue of auxiliary international languages like Esperanto and Volapuk. The beginnings of humanities and the nascent linguistics started to eliminate the philosophical quest inanities. Champollion deciphers the hieroglyphics and the issue of Adamic language became the long exegesis of the Indo-European linguistic genealogy.

Eco admitted that "all artificial intelligence projects inherit somehow the a priori artificial languages problematics" (p. 353) without actually deepen the issue and he considered, too much classically, that AI provides only limited or ad hoc findings. We should rather consider, as he says himself, that "various chapters of philosophy, logic and linguistics [derived from] the secular development of the search for a perfect language" (idem).

Strictly speaking, his book should rather be interpreted as the history of misconceptions in this quest. The study of errors is a good method, which is not best-selling within continental philosophy, contrary to the study of fallacies in the Anglo-Saxon world. Yet mistakes do characterize the very steps of knowledge.

Eco's philological skills allow him to put himself in the shoes of the scholars of past centuries when they haven't yet discovered what their successors enjoy effortlessly. Eco is able to understand and reproduce their intellectual approach with their limits or constraints, such as having to comply with the biblical dogma. It follows an identification with their biased approaches which however still exist in contemporary philosophy. If Eco had elaborated further the last steps of his study, from the standpoint of the present stage of knowledge, he would have considered, like I suggested, that the quest of the perfect language led to artificial intelligence.

A bit too quickly, History recorded that this quest is only summed to the Leibniz's calculus ratiocinator program. Eco shows us the precise developments with their fantasy trial and error moves. He concludes his book by underpinning quite well the current state of the problems in the linguistic context. But suddenly, with an overly literary twist, he freezes himself on the usual untranslatability, genuinely compliant with the myth of Babel. Eco remains a Renaissance man.

In his oral submission in the College of France, Eco had not spoken too much of artificial intelligence. I had obviously noticed this relationship and I wanted to ask him if he acknowledged it too, but he was always surrounded by people after the conferences. A few years later, when I was managing a bookshop, I had the opportunity to discuss with a client who told me that her friend was preparing a PhD with Umberto Eco. I mentioned to her what I said in my book on the different way of structuring knowledge in databases which correspond exactly to the thoughts of philosophers about whom Eco spoke in his book. The value of computer sciences is precisely to provide a way to mechanically determine the validity of these old philosophical theories. I also showed in my book that the resistance against the AI precisely consisted in the regressive persistence of old illusions carried by the ritual repetition of the traditional philosophical mistakes, whose list is mostly established here by Eco.

Those who are interested in the history of philosophy should bother reading this Umberto Eco's Search for the Perfect Language to be aware of this relationship with artificial intelligence. Those who are interested in AI could read it too, if they are not intimidated by the abundant Eco's erudition, to explore the origin of their field and find probably some new ideas there. In 2012, in a brief comment on the blog of Jean Véronis, data scientist who prematurely died in 2013, I mentioned the Umberto Eco's book. Véronis replied to me that it was his favorite book and the first recommendation he made to all his students.

This book should be considered itself as a full research program to deepen the points presented in it and to better elaborate those which are only skated over, making our mouth water. This program could meet philosophers and computer scientists to achieve unity of knowledge that was formerly seen as such. The philosophers of the times which especially interested Umberto Eco were much more aware of it that contemporary ones.

Jacques Bolo

courrier Recommander cette page à un ami
Voir aussi :
Cet article en français Francais
Subscribe to Exergue
Tous les articles

Accueil Ball © 2005- Exergue - Paris Ball Légal Ball Ball Partenaires Ball Contacts