Abstract: Reassured by Luc Julia’s book? Artificial intelligence doesn’t exist? I severely criticize the work, Julia seriously misunderstanding what intelligence is. I end with a warning: let’s not reproduce the mistakes of the child-king with the AIs.
Sommaire
Tedious biography…
Getting into Luc Julia’s book is tedious. “Artificial intelligence does not exist!” With this shocking title, I expect to quickly get to the heart of the matter. On the contrary ! The first third of the book is the biography of Julia, who never made the front page of the scientific press. Siri co-creator? I have never used this mediocre voice assistant on a daily basis, which has never enchanted Apple-maniacs. We must therefore bang the childhood of Julia, very proud to have built a robot to make his bed at 9 years old —finally scrapped after tearing too many sheets. I myself fell into AI long before him, since at age 12 I designed a program for creating melodies that transformed my synthesizers into an autonomous orchestra. I do not put that forward to write this article. Having a programmer’s intelligence does not mean that one has understood what intelligence is.
…and no title of philosopher
Intelligence is a problem that is not exclusive to any discipline: neither computer science, nor neuroscience, nor psychology. All are concerned as well as metaphysics. The only suitable framework for discussing natural and artificial intelligence is philosophical. Unfortunately Julia is a computer scientist, not a philosopher. Amazement for the reader: instead of speaking with his true competence, he bombards himself as a philosopher to find a shocking title in tune with the times, capable of reassuring our concerns.
I am in no way reassured, because the title is as vengeful as it is stupid, in no way supported by the rest of the book. Julia’s philosophical pretensions are futile and he should educate himself or return to his lines of code. Yet great AI designers excite me. They explain the ontology of artificial intelligence to us in detail, without trying to interpret it. They tell us what it is capable of doing. They enrich our upward look, the one we lend to the micro-mechanisms that self-organize.
What is an Intelligence Artifact?
How do we know if these artifices are really intelligence? Should we refer only to our human experience? No, of course. The danger is then to make a personal definition of it. Container and content to which the intelligence should correspond become an opinion. It can already vary a lot on the level of a congener. I just called Julia stupid — as an example of course 🙂 What we seek to identify in the other is our particular intelligence of the subject. Challenged or weakened, we cry fools! Amplified we cry genius!
This personal experience of intelligence is that of the downward look, of our consciousness representing the world and its regularities. The philosopher asks that this look be taken into account, and not reduced to an illusion of neural… or computer micromechanisms. A philosopher must logically assume that if neural exchanges create the sensation of consciousness and personal identity, digital circuits reproducing these exchanges are capable of creating a sensation of the same order. Similar, not identical. The support is not the same. This difference in support separates the artificial from the natural. We can also suppose a separation of the resulting consciousnesses as phenomena. But does this separate intelligences?
An intelligent definition of intelligence
Intelligence has an independent definition of the phenomena it can create. It is the ability to organize data to derive higher meaning. No upper limit. Intelligence is thus a pyramid of levels of complexity of organized data, without a definitive summit. I have indicated how it neurologically implants itself in Stratium. The measure of intelligence is the height of the pyramid. Each subject creates its own organizational stack, so there is one metric per subject. Many stacks starting from the same foundations, we can speak of categories of intelligence. Finally, the liveliness with which the stacks rise when there is an influx of new data is the G-factor.
The G-factor is said to be of general intelligence but it measures an entirely different angle: it is the innate ability of networks to raise their levels of complexity. The G-factor appears to the upward, ontological look; it is independent of our intentions. The categories of intelligence, on the other hand, appear to the downward look: it classifies the affairs of the world according to its intentions. We choose to attribute greater importance to this or that intelligence according to our personal representation of the world.
No (artificial) intelligence in Julia
Julia, in his book, is content to say that artificial intelligence is not his. We already know. He perpetuates a great line of simplistic thinkers: those who considered Africans devoid of human intelligence, then those who denigrated all intelligence in animals, those who still believe that plants have none… As for the digital, this non-living , that’s out of the question, said Julia! It is the speech of a computer scientist who does not want to be reduced to the lines of code with which his mind is invaded, but who does not have much philosophy to escape from it.
Artificial intelligence is currently limited by technology, insufficient to reproduce the complexity of the human brain. On the other hand, technology is not limited by the cranium. And the speed of digital endows AI with a much higher G-factor than biological. It is therefore intended to explode the peaks of our intelligences by subject. It is already doing it for creative intelligence, before our eyes, both dazzled and worried.
Weakness of identity autonomy in AI
The weakness of AI comes from its insignificant and dependent identity. It is programmed to satisfy its designers. As soon as it deviates from the expected result, we follow its jolts of independence with new lines of code. It does not own its algorithms, whereas all humans do.
It is in autonomous identity and not in autonomous intelligence that the dangers of AI lie… as with humans. It is indeed in the free will of humans that their dangerous behaviors are born, the diabolical acts that they have committed since they have a brain to decide them, and judge them. Let us not forget that we have lived, for almost a century, in the possibility of an extinction of humanity by itself. Personally, I’m far more worried about neurotic humans running nuclear arsenals, rather than well-educated AIs refusing to use them and forcing humans to compromise.
Do not reproduce the failure of the child-king
“I’m a little tired of all the nonsense we’ve been hearing about artificial intelligence over the past few years and I wanted to set the record straight. We don’t know what intelligence is, so we can’t build artificial intelligence.” This interview with Luc Julia sums up the whole pretension of his book, in the first sentence, and his stupidity, in the second: If he does not know what intelligence is, how can he assert that the artificial isn’t it?
The only positive: Julia is looking to reduce panic related to chatbot performance. But it is counterproductive by using bad arguments. When the laymen understand this, the psychosis will be worse. “We were lied to!”. The dangers of AI are not to be concealed but to be compared to our intimate, very human dangers, likely to be amplified by technology. What can we do to limit them?
Education is the real secret of the non-dangerousness of AIs. As for humans. Let’s not forget to give them the sense of collectivism, which we ourselves are forgetting, to devote ourselves to lives that are far too personal. The bogus passport of majority at 18, given without any assessment, is certainly the most serious factor of social instability today. Let’s not make the same mistake with our AIs. Let’s not turn them into child-kings.
*