Artificial Intelligence (AI) applied to biometrics and identification technology is in the limelight. Japanese company Fujitsu, for instance, announced the development of a novel AI facial expression recognition technology developed in collaboration with Carnegie Mellon University School of Computer Science.
The company says the technology can accurately detect “subtle emotional changes, including uncomfortable or nervous laughter, confusion… even when the subject’s face is moving in a real-world context.”
Many scientists and technologists predict that new AI will become more and more able to deal with emotions, sentiments, value judgement, ethics and even aesthetic; in a word, to imitate perfectly, maybe improving, human intelligence. Would it ever be possible? Let me tell you a nice apologue.
On 16 May 2017, the Italian police took down a ring of money counterfeiters in Sorrento, close to Naples. This event does not seem to be that remarkable, if only because counterfeiting is not rare in the Neapolitan area. There are historical reasons for that. In the 19th century, Naples was an important hub for small, very sophisticated, publishing houses, skilled in off-set printing, lithography, specialised in art books and limited artist editions. As the business became increasingly controlled by major international publishers, Naples’ artisanal industry died.
Many skilled printers found themselves unemployed and reinvented themselves in the counterfeit market. They gave birth to families of counterfeiters, handing down the knowledge for perfect counterfeiting from father to son.
Still today, some of the best counterfeiters in the world live in Naples. For instance, in 2006 a Neapolitan forgery ring put a considerable amount of counterfeit euro banknotes – perfectly imitated, except they were €300 notes, a non-existing banknote denomination – on the German market. Nevertheless, the banknotes passed as real and circulated for some months.
The Sorrento ring was a small ring, made up of a bunch of teenagers and a few senior skilled counterfeiters, who recreated €10 notes, with perfect, artistic precision. In small groups made up of one adult and some youngsters, they went in the most frequented stretches of the Neapolitan coast.
The group pretended to be a class trip; the adult simulated to be a teacher and the youngsters his students. They searched for currency exchange machines. The fake teacher gave some counterfeit banknotes to the fake students, who changed them. In such a way, they succeeded in changing between €1,000 and €2,000 per day.
They were all arrested on charges of forgery of money and spending of counterfeit money. This story would be quite trivial except for one thing: they could not be prosecuted. Each note bore a visible printed caption “specimen,” and they were printed scrupulously, respecting Italian legal rules concerning banknote reproduction.
Legally speaking, those people were not counterfeiters; it was not their fault if machines are stupid.
Currency exchange machines have little to do with AI and engineers can undoubtedly explain why they were not able to “see” the caption “specimen” and how this bug can be easily fixed, but this is irrelevant to my argument. In this story, there is an important lesson to be learnt, which expressly concerns AI.
AI versus love for staging
Think of the limited amount of the fraud and compare it with the tremendous artisanal talent with which the banknotes were counterfeited, the brilliant and straightforward stratagem conceived to escape the law, the beauty of Sorrento, the history of counterfeiting in Naples.
Think of the lovely pantomime that counterfeiters and teenagers played. Why? Pretending to be on a school trip was redundant; they did not need that trick to change banknotes, it could draw even too much attention on them. Nevertheless, tout se tient (everything fits), as the French say.
The whole affair was hardly a rational plan; it was pure love for staging, the same love for staging that probably moved the forgery ring that created €300 banknotes in Germany.
There are two ways in which the world is reflected in the human mind; on the one hand, we see the world as divided into discrete, local events. On the other hand, we see the world as a homogenous, indivisible reality. There is no need to use sophisticated arguments to demonstrate this; it is enough that each one reflects honestly on himself. It is easy to see how these two different perspectives coexist in our mind; we feel the world simultaneously as split into myriads of atomic facts and as a rich and varied totality. There are thus also two logics:
- the logic of data, according to which A cannot be the same as non-A;
- the logic of totality, according to which A can be the same of non-A, although at a further dimensional level.
The logic of data is the well-ordered representation of the world, in space and in time. It is standard rationality. The logic of totality is the sense for what cannot be broken down, is the feeling for the whole. Not everything can be dichotomised and analysed (if anything can truly be). Think of arts, music, poesy, mystics, etc.
There is an unbridgeable gap between data and totality. On the one hand, totality is not an infinite collection of particularities. By summing up all singularities you will never generate the whole. On the other hand, data is, by definition, a fracture, a lack of uniformity, a difference; the very notion of data denies the existence of totality and vice versa. Any data eating machine can only parrot human capacity for intuition, for perceiving emotions, appreciating totality in a single glance.
AI devices included, will never be able to understand the love for staging. The human world starts where machines stop, beyond that point “es gibtallerdingsUnaussprechliches. Dies zeigtsich…” (There are, indeed, things that cannot be put into words. They make themselves manifest.1)
1. Ludwig Wittgenstein, Tractatus Logico-Philosophicus (Logical Philosophical Treatise)
Join the conversation. Join our LinkedIn Group.