Personhood, Logos and Artificial Intelligence
Y a-t-il un homme si stupide qui n’invente au moins quelque signe pour se faire entendre? Y a-t-il une b�te si rus�e, qui ait jamais trouv�? Et qui ne sait que la moindre des inventions est d’ordre sup�rieur � tout ce qui ne fait que suivre?”
Jacques Benigne Bossuet: De la Connoissance de Dieu et de Soi-m�me (pg. 103)
“En arche en o Logos”. This sentence is the beginning of the Gospel according to St. John. The tradition of Christianity has always translated this sentence as “In principio erat verbum” , “In the beginning was the Word”, where the term “Logos” corresponds to “Word”; this translation is accepted today by all the scholars. But the Greek term “logos” is polysemic, i.e. can be translated in Latin as “verbum” or as “ratio”. As an example, the usual translation of Aristotle’s definition of man (human being) in Metaphysics (Z12, 1037b13-14) ”zoon logikon” is “rational animal”, where, in this case, the root “logos”, in the adjective form logikon, has been translated as “ratio” instead of “verbum”. Had the Stagirite translator used the same meaning as the Gospel according to St. John, we human beings would be defined as “verbal animals” instead of rational animals.
This possible mistranslation of Aristotle can be seen as a remote origin of the concept of intelligence as self awareness, but this equivalence (intelligence = self awareness) can be shown not to be the case, as Ludwig Wittgenstein did; we present below an abstract of Wittgenstein’s point of view.
The philosopher who mainly contributed to this misunderstanding was almost certainly the Frenchman Ren� Descartes. Among the multiple merits of Descartes is the fact that he was the first philosopher that explicitly studied epistemological problems, putting aside or devoting less interest to the ontological problems that had been, up to then, the main subject of study amongst philosophy scholars.
From our own point of view, those epistemological problems started in the Renaissance with the discovery of the certainty provided by the mathematization of the sciences of nature after centuries of acceptance of so called “first principles” as self-evident, without any further investigation. This discovery drove philosophers to try to find a similar degree of certainty in the sciences of spirit. It is possible to summarize briefly three possible solutions to this problem. Let us list them from the most recent to the oldest.
The commonest solution today of the problem of how to find certainty in the sciences of spirit is that of being able to give a sense to a text; this philosophical school is called hermeneutic. It was originated by the “hermeneutic turn” of the phenomenology of Martin Heidegger.
An older solution to the problem we are studying is the one that denies any possibility of certainty, reducing it to the sciences of nature, only to what can be measured, counted, verified experimentally; this philosophical school is called positivism and it is particularly important in the logical positivism school typically characterized by the Vienna Circle.
But the oldest solution to the problem was rationalism, root and origin of the other solutions, and this school was pioneered by Ren� Descartes.
The philosophical problem created by rationalism was not the question of the quest for certainty in the sciences of spirit, but Descartes’ answer. Looking for a certainty criterion of apodictic value, he found that he could be sure of the truth of a statement when it was a clear and distinct idea (“id�e claire et distincte”).
Descartes, as is well known, started from a universal doubt about everything, or so he thought, even if Wittgenstein showed, centuries afterwards, that this was not the case. The flaw discovered by Wittgenstein at the end of his life is precisely related to language. The Viennese professor at Cambridge emphasized the fact that the Frenchman never doubted the meaning of language: he, like everybody else, needed to found doubts on some certainty, i.e. total scepticism, even if methodical, is impossible, leads to nonsense [see 7,8 for further precisions]. Once more, we find that the verbal aspect of the epistemological problem has been neglected.
The escape for Descartes’ skeptical method was, as is well known, the “cogito ergo sum” insight. This sentence was felt as absolutely true without any possibility of doubt. It was seen as an “id�e claire et distincte”, a clear and distinct idea. Generalizing afterwards on this subject, he arrived at the conclusion that every such clear idea should always be certain beyond any doubt.
This is an “internalist” criterion and therefore impossible to validate objectively, but the main problem with it is that it leads to the principle that certainty, capacity of attaining truth, intelligence in a word, is essentially the ability of being perfectly aware of his own mind (self- awareness).
In fact, this philosophical position led to a kind of contamination between certainty and personhood. Using the (implicit) criteria that only persons could attain certainty, it follows that personhood must be identified with the ability of having these “id�es claires et distinctes”, that’s to say, self-awareness.
The impact of this new (in the XVII century) idea was enormous, not only among professional philosophers, but also among the general public. Descartes translated his own works from Latin to French and achieved a very large diffusion of his ideas in Europe. Suffice to read the “M�ditations Cart�siennes” of Edmund Husserl (1929) to see that Cartesian points of view are at the origin of Phenomenology and with it, of all the “continental” philosophy up to the most recent years, up to the hermeneutical turn. Incidentally, the “M�ditations Cart�siennes” were a set of speeches given in Paris in the Descartes Hall of the University.
As we said, this is not only the case with professional philosophers, but also with the general public; open any newspaper and on any page you will find, explicitly or -more often- implicitly, the equivalence between intelligence and self-awareness. Consult, for example, the International Herald Tribune of February 25, 2008, on killing elephants in South Africa to end overpopulation, to read:
The announcement follows months of impassioned debate, with some conservationists arguing for elephant killings to protect the ecosystem, and animal welfare groups outraged at the prospect of slaughtering one of the planet’s most intelligent and self-aware creatures.
The non-continental (Anglo-Saxon) analytical philosophy, and particularly Ludwig Wittgenstein, has pointed out how flawed is the Cartesian way of seeing intelligence. To defeat his very general statement, it suffices to show a counter example, and this can easily be taken from everyday life. How many times were we sure of knowing something, for instance a subject matter before an examination, just to find afterwards that our idea, even if clear and distinct, was wrong? But to quote a famous, more illustrious, example, we transcribe from St. Augustine’s Confessionum libri XIII, liber XI, caput XIV: “Quid est ergo tempus? Si nemo ex me quaerat, scio; si quaerenti explicare velim, nescio” i. e. “What is time? If nobody asks me, I know; if I want to explain it, I don’t”. Is there a more eloquent objection, stated eleven centuries earlier, to the “Id�es claires et distinctes” Cartesian thesis? Intelligence can often go along with self awareness, but the real discriminatory criterion between intelligence and non-intelligence, must be found elsewhere, maybe translating the Greek word “Logos” into “Verbum” and not into “Ratio”.
Even the very motto of this Metanexus conference, again from St. Augustine, “Mihi question factus sum”, implies a self awareness far from being “claire et distincte” and needs words to formulate it: one can be aware of hunger, fear or joy without terms, as a dog often does, but a question cannot be thought, let alone formulated, without words. (Philosophische Untersuchungen Part II, I; Part I, nº 250)
The explicit refutation of the Cartesian point of view, without naming Descartes himself, came from the Viennese philosopher Ludwig Wittgenstein when teaching at Cambridge University (Philosophische Untersuchungen Part I, nº 311 and ff.). His argument is just the one in the preceding paragraph: a dog can show joy knowing that his master is present, this action (even this action of communication with tail movements, jumps etc.) needs no language; but to manifest joy knowing that his master will be present tomorrow at the same time is impossible without the aid of some language similar to the human one.
One of the “side effects” of Wittgenstein’s philosophical position is related to language (and therefore personhood) as something depending on a collection of individuals. In fact it is well known that Wittgenstein emphasized that language, apart from being a creation of a group of people, must belong to a collectivity: private languages, as Wittgenstein called them, are impossible (Philosophische Untersuchungen nº 243, 258). But, if a person is a verbal animal, then our very personhood, which depends on language, depends also, therefore, on the group to which we belong.
The bibliographic reference [9] discusses this point, but we can also, in a more pleasant way, remember Lewis Carol in [13]:
“When I use a word” Humpty Dumpty said… “it means what I choose it to mean – neither more nor less”
“The question is” said Alice “whether you can make words mean so many different things
This opinion of personhood depending upon the human community through language is shared with the Systemic School (also called Palo Alto School) of psychiatry. This school sees personhood as a “picture” depicted in words, that must constantly be contrasted with the images reflected on other individuals, also in words [see12, no. 3 – 33]
Some of Wittgenstein’s disciples remarked that the refutation of Cartesian ideas allowed philosophers to put again on the table the work of some older philosophers that had been much neglected after the XVII century, particularly medieval scholars, and mainly Thomas Aquinas. This movement of Wittgenstein’s disciples is known today as “Analytical Thomism”. To put it in the words of one of them, Anthony Kenny:
…Wittgenstein’s importance in the history of philosophy lies especially in his criticism of the Cartesian framework within which philosophy has been conducted through the modern era, well beyond the critique of Kant. One side effect of Wittgenstein liberation is that it enables us to read the writings of pre-Cartesian philosophers, and in particular of medieval scholastics.
This is important , because it allows us to define again “mind” as a handful of possibilities, just as medieval philosophers defined the soul as a set of capacities, denoted in latin by the term “potentiae”, possibilities, power to do something. In fact, mastering language is the possibility to effectively speak, and even small (or unborn) children have the ability to acquire such an ability, a second degree capacity or “meta-potentia”, i.e. an ability to get an ability, either by growing or by maturation (learning).
I think that the authors that best treated the problem of the inner mental world, the pretended “mind theatre”, as it is often called, are Jacques Bouveresse [15] and particularly Gilbert Ryle (see his chapter titled “Self awareness”, and particularly pg. 181 – 184 in [14]). But for the sake of simplicity, let me explain the ideas of both philosophers with a personal anecdote.
Last autumn, I was preparing a course on this subject one afternoon on the top of a lookout in southern Spain. Going down the hill back home, a man looking like a beggar tried to sell me some tropical fruits. I rejected the offer, suspecting a dubious origin of the fruits, when the man asked me about the book I was holding in my hands. I explained that the book served me to prepare a lecture at the university.
“Do you lecture at the university? That’s fascinating! Which subject are you going to talk about?”
“Listen” I told him “I am going to explain it in simple words. Let’s take an example: where must you look to see that boat?”
“That’s a silly question” answered the man, “I must look at the sea, of course” And he pointed his finger in the right direction.
“I can read in your face that you are afraid that you will not understand what I’m going to tell you. Is that right?”
“Yes, I must confess I have a little apprehension.”
“And where must you look, to see that you have a little apprehension?”
“That’s an even sillier question” he answered quickly. “I know it, that’s all. I need not look anywhere to see that! It is like a tooth that hurts, I know it hurts, I needn’t look anywhere!”
“OK, you are a clever man. You understand everything on my subject.” I emphasized. “You need not attend my lecture.”
That man understood clearly that introspection is not a kind of inner perception [14]; that is the commonest mistake that philosophers as well as non-philosophers make after Descartes, as Jacques Bouveresse and Gilbert Ryle explain in their books.
* * * *
One field in which defining intelligence is crucial, is precisely the one called (perhaps improperly) “Artificial Intelligence”.
In fact, before one can answer the question about the possibility of such a thing as “artificial intelligence”, the term “intelligence” itself should be defined. But, obviously an “internalist” definition like the Cartesian one (self-awareness) would not do, because it would be useless to check if a machine shows self-awareness or not. Nothing simpler than programming a computer to say: “Yes, I am perfectly self-aware”. This not only useless and even inappropriate, but just flat nonsense.
Is ability to add up two numbers an intelligent operation? Certainly! But we cannot say that the adding machine devised by Blaise Pascal in 1645 was an “intelligent machine”, it was just a product of Monsieur Pascal’s intelligence. The same can be said of the really wonderful multiplying machine of Leibniz, the Analytical Engine of Babbage or the programmes that play chess at grandmaster level today. Personhood implies the ability to do all these things, or the ability to learn to do all these things, but not the other way around: by themselves alone those abilities cannot be said to define intelligent behaviour, they are tools created and used by intelligent people. When we use one of these machines it is its creator who solves the problem, via the previous solution of a meta-problem, but the machine does not solve anything; in the same way as a hammer doesn’t pound a nail, but it is the user of the hammer who does it. The same concept is true, speaking about a more complicated tool such as a computer.
The most accepted criterion for ascertaining if a device succeeds in showing artificial (or even natural) intelligence is the Turing test, called that way because it was first proposed by Alan M. Turing in his article “Comput ng Machinery and Intelligence” published in “Mind” in 1950.
To perform the Turing test, some judges are connected via some appropriate devices to a human being and to a machine. Both are allowed to chat with the judges and, if the judges are unable to distinguish the machine from the human being, the machine is said to have passed the Turing test and is considered to be artificially intelligent. This seems to be an application of the Leibniz principle of identity as indistinguishability.
Let me say immediately that no machine, computer or programme has ever passed this test. As a matter of fact, a contest is performed every year (Loebner Prize in collaboration with the Cambridge Centre for Behavioural Studies, in a little more sophisticated manner than the one described) and a significant amount of money will be awarded to the first machine able to cheat its judges. A dozen human beings and a dozen computer programmes are made ready to “talk” via a screen and a keyboard and some judges should decide if they are chatting to a machine or to a person; as we say, no programme has ever succeeded in cheating any judge. To render things more likely (and perhaps more interesting) a “restricted Turing Test” is used. In its restricted way, only a particular subject (for example soccer, quantum dynamics or Shakespeare works) is permitted in the exchanges between judges and programmes, instead of just general unrestricted chat, making things easier for the programmes. In spite of this simplification, no programme has succeeded either. [The interested reader can easily consult the transcripts of these conversations in the appropriate internet site].
But, again, mastering language doesn’t mean just the ability to chat like a person, as Turing seemed to tell us. As we say above, the ability to speak (“potentia”) should allow us to acquire other abilities, as we did in primary school. How did we learn arithmetic? By being told, as we learnt practically every other ability, even nonverbal ones (riding a bicycle, for example). But a machine able to pass the Turing test will not in principle be able to learn by being told, it will not be really intelligent in the same sense as a person is. Some attempts were made in the 60’s to do so: a programme called “Eliza” by Weizenbaum, was able to learn a little “foreign language” (he primarily talks only English) but not in a way to be confused with a human being.
Another important consideration in this subject is that it is at least as important as being “verbal” to be “animal”, in order to be intelligent, to be a person, as Aristotle told us.
This assertion looks a little strange at the first sight, but a minute reflection will show that personhood, even in the general non-technical sense, implies something more than pure verbal intelligence.
A few months ago, I was attending a meeting in which poems were read by a poetess. This woman artist lives, as is usual in these cases, a different life from everybody else. She spends most of her time reading, writing or just thinking, instead of watching TV like everybody else, and this behaviour leads people to consider her as “a brain on two legs”, as an inhuman body.
After the reading of her poems, somebody approached her saying “I am happy to see that you are, after all, human too. That you suffer, expect, feel depressed etc.”. This utterance deeply struck me as paradoxical. After all, considering inhuman a person just because she thinks all the time looks strange, because thinking is, of course, a typically human action. Moreover, considering her human because she shared feelings also found in animals (joy, fear, worry, sadness etc.) was adding one paradox onto another.
But some thinking about this fact made me remember the Aristotelian definition of personhood: rational animal! And I realized that we have some tendency to forget the “animal” part of the definition.
Aquinas defines the human soul as intellect plus will, and people sometimes call “inhuman” certain, perhaps very intelligent, people lacking human feelings, unable to be generous, or insensible to other people’s sufferings etc. This goes in the same direction as Aquinas when he says that man differs from other animals, not only in intellect, but in the most sophisticated internal senses, even -he adds – if sometimes external senses are less perfect in persons than in animals. (Summa Theologiae I pars, questio 78)
Conclusions:
- It seems more appropriate to define personhood as “verbal animal” than as “rational animal” as has traditionally been done.
- In this line of thought, the Turing test is appropriate to define a kind of intelligence in a machine, putting emphasis on verbal abilities and being “externalist” and, therefore, objectively checkable.
- The Turing test seems to be incomplete, as it fails to show the ability of learning by being told.
- The “animal” part in the “verbal animal” definition of personhood is an essential one. Fear, joy, pain and even hunger, thirst, and sexual drive, have a specific way of being in human personhood.
Bibliography:
The gospel according to St. John. Trilingual version (on the net)
Aristotle: Metaphysics
Aurelius Augustinus: Confessionum liber XIII; http://www.sant-agostino.it/latino/confessioni/index2.htm
Thomas Aquinas: Summa Theologiae (I pars, q 75-90); http://www.corpusthomisticum.org/sth1075.html
Ren� Descartes: M�ditations M�taphysiques GF Flamarion
Ludwig Wittgenstein: Philosophische Untersuchungen Editorial Cr�tica, Barcelone
Ludwig Wittgenstein: �ber Gewissheit, Suhrkamp Verlag KG
Anthony Kenny: Wittgenstein, Blackwell Publishers Ltd. 1973
Saul Kripke: Wittgenstein on Rules and Private Language, Blackwell Publishers Ltd.
Edmund Husserl: M�ditations Cart�siennes, Vrin 1992
Alan M. Turing: Computing Machinery and Intelligence, Mind 1950
P. Watzlawick et al. Pragmatics of human communication, W.W.Norton & Co. 1967
Lewis Carol: Through the looking glass, Penguin Classics
Gilbert Ryle: The Concept of Mind, Penguin Books
Jacques Bouveresse: Le mythe de l’Interiorit�,