Who is afraid of Artificial Intelligence?
Artificial intelligence. Considerations on the sidelines of the Giuliano Amato case and beyond. The speech by Giuseppe Sacco, professor of Economics and International Relations
It is well known that the invention of artificial intelligence has aroused many fears. Just as it is quite clear that these fears continue to multiply as ever new segments of the Internet public find themselves confronted with that formidable communication and research tool that are chatbots, and above all ChatGPT, which at present seems to be the most powerful and popular among them.
These are fears that we cannot – frankly – be surprised by.
In fact, to many, this first tool for mass access to AI seemed like the translation into reality of an objective of technological and scientific research that until not long ago was considered purely imaginary, even not entirely far-fetched. An achievement whose possible consequences had only been hinted at by authors of novels or films that remained popular and famous for many decades, but books and films written or filmed only to make their audience feel an exciting thrill of fear. From Frankenstein, or the modern Prometheus where the monstrous creature escapes from the hands of its creator, up to the Hul computer of 2001 A Space Odyssey , whose attempt to take control of the spaceship brings a reaction from the only surviving human being, who he does not hesitate to destroy the machine on which the attempt to conquer the universe by the native Earthlings was based.
An ambiguous relationship
In the darkness and almost religious concentration that prevailed in cinemas, the reaction of the spectators – all human beings – was decidedly ambiguous. On the one hand, there was no doubt that the computer's excruciating agony as its vital components were taken away was making them suffer. Because with that machine, and with the futuristic ambition that it embodied, they had entered – as we would say today – into empathy. That is, because in that machine they recognized their own natural and inextinguishable projection towards knowledge and towards the absolute. But on the other hand, it was clear how much they feared the domination of the artificial machine over their peers. And how they were unanimous in the belief that this, if it ever attempted to escape their control, should be nipped in the bud.
This relationship between the society of men and their creation is what still manifests itself today when faced with the need to decide on the future applications of AI. There is no doubt that this, Artificial Intelligence, is a creation and an almost exclusive domain of technicians, mathematicians and computer scientists. And that it will be these above all that will guarantee its future developments. But research, choices and decisions do not only concern them, because all the other aspects of men's intellectual and social activity are involved in different ways in the prospect of coexistence between human beings and thinking machines.
Sociologists and economists are already calculating the impact on the labor market, the possible replacement of human workers in the activity of producing wealth. But philosophers and politicians will inevitably ask themselves, or rather they are already wondering, about another type of consequences, those that these extraordinary machines will cause in the distribution of power and the ways in which it is exercised. And therefore of the remedies which it will perhaps be necessary to resort to to curb its development, to prevent new technologies from shaking the social pyramid too violently.
All the main countries, in recent months, have posed these problems, and it is therefore not surprising that the Italian government has resorted to them with the creation of an official group for reflection and legislative proposal on the subject. Indeed, even two think tanks whose characteristics and skills seem at least partially to overlap. And the most recent of which, however, has raised some doubts at the highest levels of government. Which, like all power structures, cannot hide its fears in the face of this new gigantic step in human knowledge and creativity.
Between technique and law
But the appointment of a long-standing jurist and magistrate like Giuliano Amato as head of the study commission on AI in the publishing sector did not only raise the irritation of Prime Minister Giorgia Meloni, to whom he appeared too advanced in years. This appointment – but even more so the conferral of such an intellectual scientific role on a jurist – even seemed to create the dismay of one of his government colleagues, the hyper-active Matteo Salvini: "I find putting a constitutionalist in charge of artificial intelligence curious”, said the vice-premier.
Locked in his power games, Salvini evidently does not realize how true the opposite is. And that is, how jurists are increasingly destined to "deal with the unprecedented scenarios that open up with the massive use of algorithms and, in general, new technologies". And how the birth of machines capable of making autonomous decisions clashes with "one of the cornerstones of modern legal thought, consisting of the possibility of attributing conduct to a subject" who can be held responsible. These are problems that have clearly emerged in the field of driverless vehicles, and which have meant that the unmanned air taxi is already in operation today in China, and that in all the most advanced countries the prospect of abolish this figure even on airliners, but we are very far from a car that can do everything by itself in surface traffic.
Between “new” and “old”
The case of Meloni's "youthfulness" is different. On the one hand, his doubts regarding the presidency entrusted to an eighty-five year old appear to be criticisable, especially if one takes into account the fact that future technological developments of generative artificial intelligence – and more precisely of Artificial Intelligences (in the plural) – seem to promise the prolongation of the intellectual activity of individual human beings beyond the limit of their physical life. Which would transform all the thought and experience accumulated before each individual's encounter with generative AI into an irreplaceable heritage.
On the other hand, however, it is probable that Meloni believes – so that Italy does not fall behind other countries in a field that is already the subject of scientific and technological rivalry – that a project that promises an extraordinary leap towards the future should addressed concretely. And which, if it undoubtedly requires that it be approached with prudence, should in any case be entrusted not to elderly lonely men who every break with the present makes full of doubts and fears, but rather to intellects without too many emotional attachments to the past, and aimed at the new, who is generally believed to be the youngest people.
Meloni would not be alone in this belief. Very often, in the debate that is raging in all the main countries on this topic, it has in fact been pointed out that in Great Britain – which with the United States is probably the most advanced country in AI research – at the head of a body similar to the one as soon as it was created in our country, a thirty-eight year old, Ian Hogarth, graduated in computer engineering, specialized in machine learning and entrepreneur in the digital sector, was placed.
In short, London has decided to bet its cards on a man who is not only a specialist in the sector, but also a personality who belongs to a specific generation; the one whose formation coincided fully with the developments of technology which led us – well in advance of what had been expected until yesterday – to the concrete possibility that the machine could catch up and surpass man in a field in which he was always was the absolute dominator.
The prudence of the technicians
It must be said, however, that in March 2023, the "young" Hogarth himself was among the signatories of an appeal so that all those who carry out research and application in this sector should observe a pause even if only for six months – one might say “a truce” – in the development of AI systems of higher power than GTP-4, i.e. the professional model of that ChatGPT which is offered by the Internet for universal access.
The reference to GPT-4 as a technological goal once reached is very interesting and a pause for reflection seems appropriate. This reference highlights the difference between the race currently underway for the creation of the many generative AIs and that for AI tout court. If limited alarm, or even no alarm, has been raised by the fact that some private company (not only American but also South Korean) is now able to offer – and at a price accessible to a middle-class family – the " resurrection of a deceased loved one, the same does not apply to GPT-4.
The latter – as well as the various competing or imitating models that can be reached on the Internet – seems to be an excellent starting point for giving life to what Hogarth himself calls the "God-like AI". God-like, but also very human-like, as “more powerful models are beginning to demonstrate complex capabilities such as ambitions for power, or ways to deliberately deceive humans.”
An agreed pause among industry operators – wrote Hogarth – would be an important move to gain time to dedicate to research on the security of Artificial Intelligence systems. And it is probably with similar intent that, together with him, more than 1,800 scientists in the sector signed the appeal. And not only "elderly" like the 73-year-old Steve Wozniak, who played a role equal to that of Steve Jobs in the creation of Apple, but also true "visionaries", our contemporary "creators of the future", like Elon Musk. Who – as he himself explained – is precisely because he is aware of the risks of the present that he has always boldly and repeatedly projected himself, with extremely ambitious initiatives, towards the future.
This is a machine translation from Italian language of a post published on Start Magazine at the URL https://www.startmag.it/innovazione/chi-ha-paura-dellintelligenza-artificiale/ on Mon, 30 Oct 2023 08:13:51 +0000.