Artificial intelligence, who criticizes and why the EU AI Act?
Despite not being among the protagonists in the development of artificial intelligence (AI), the EU wanted to be the first to arrive with a law regulating it. Now he has it, but according to various experts there is little to boast of the record. Round of (critical) comments on the AI Act, approved yesterday by the European Parliament
Yesterday the European Parliament approved the AI Act, the world's first law on artificial intelligence (AI). However, according to some experts, there is not much to celebrate…
A RULE FOR CARVING OUT A ROLE
For Stefano Mele, partner and head of the cybersecurity & space economy law department of the Gianni & Origoni law firm, interviewed by Formiche , the AI Act is the European Union's strategy "to try to influence the 'race' of states through rules United States and China", "not being able to compete in terms of investments and technological development in the artificial intelligence sector".
In fact, the real race for AI is being played, once again, by the two largest economic powers which, as underlined by Mele, "have been investing billions of dollars in artificial intelligence for years, or rather in what, in my opinion, is one of the main technological macro-trends capable of guaranteeing global leadership in the near future".
This is echoed in a more critical way, in the Corriere della sera , by Alan Perotti, researcher and data scientist at Centai, the laboratory for advanced research in the field of artificial intelligence founded in Turin: “Europe is carving out a role as referee in a technological challenge led by America and China: the referee is useful in competitions, but never wins. It is important that this regulatory part is paired with investments in research, otherwise we end up legislating other people's products."
THE REAL (ITALIAN) URGENCY IS INVESTMENTS
Stefano da Empoli, professor and president of I-Com, again in the Corriere , also hits hard on the insufficiency of investments, especially Italian ones: “The European AI Act is a necessary but not sufficient step for Italy. The clear rules outline a path and raise awareness, but adequate investments are still lacking. While reaching a regulatory goal, effective implementation is needed which is by no means a given."
“Today – declares the teacher – only 5% of Italian companies use AI, with an enormous potential for diffusion among small businesses which make up over 90% of the total. The challenge is to increase public and private investments in research and innovation, currently insufficient, with greater European coordination to reach the critical mass necessary to compete with the USA and China".
A (USELESS?) RACE TO ARRIVE FIRST
In Domani , Giovanni Maria Riccio, jurist and full professor of comparative copyright law at the University of Salerno, reiterates Europe's role as a spectator with respect to the United States-China duopoly and also raises another question, namely that the The EU does not seem to take into account the rapid evolution of these technologies which, "in the near future, could radically change".
Europe, despite not being among the protagonists of the AI race, rushed to be the first to pass a law but the rules of the AI Act, writes Riccio, "at the moment apply mainly to non-European companies , making the vaunted citadel, which has erected its walls around fundamental rights, look like a small town that suffers from a clear technological delay compared to the United States and China (and perhaps also to others) and that desperately tries to defend its competitive delays ”.
Furthermore, the jurist also once again asks himself questions about the few investments in the sector and the EU's not openly declared ambitions: "how much are European countries – apart from France – investing in the development of these technologies? […] is the Ai Act a European regulation or is it something that, from Europe, would like to control what is taking place beyond its borders?”.
WHAT IS WRONG ACCORDING TO THE CO-FOUNDER OF TRANSLATED AND THE M5S
Even more severe is the judgment of Marco Trombetti , co-founder of Translated, which among other things carries out Airbnb translations into dozens of languages with the help of AI: "I see more than one problematic aspect: how they should be the 'datasets' to train the algorithms, to the sanctions that I find disproportionate and which I hope do not discourage young people and startups from investing".
“At the same time – he adds -, some real risks are not regulated: primarily autonomous weapons. But also the transparency behind 'alignment', which is the ability of algorithms to return users answers aligned with their preferences and potentially capable of influencing their opinions and behavior in an increasingly subtle and pervasive way".
Sabrina Pignedoli , MEP from the M5S group (who abstained), agrees: “The text that has emerged risks being more harmful than useful. The definitions are too vague, the 'measurability' of artificial intelligence is a chimera and the AI Act overlaps, in some parts, with other European regulations, even contradicting them."
“In this way – reiterates Pignedoli – it becomes very difficult, especially for small and medium-sized businesses, to be able to have a clear and simple regulatory framework on which to base their investments and the possibility of accessing artificial intelligence technologies. The risk is that instead of promoting the spread of safe artificial intelligence, entry barriers are created, increasing gaps and discouraging European innovation."
A FAILURE FOR THE HUMAN RIGHTS OF THE MOST DEFENSELESS
For Access Now, a non-profit organization founded in 2009 and focused on digital civil rights, we're not quite there. From its website it defines the AI Act as “a failure for human rights, a victory for industry and law enforcement” which allegedly perpetrated “scandalous lobbying activity ”.
The result, according to the organization, is a text “full of loopholes, exceptions and tweaks that will not protect people, nor their human rights, from some of the most dangerous uses of AI.”
In particular, Access Now notes that the AI Act “fails to adequately ban some of the most dangerous uses of AI, including systems that enable mass biometric surveillance and predictive policing,” thus creating “a regime separate for people who migrate, seek refuge and/or live without documents, leaving them with far fewer rights than EU citizens and almost no access to remedies when these rights are violated.”
This is a machine translation from Italian language of a post published on Start Magazine at the URL https://www.startmag.it/innovazione/intelligenza-artificiale-chi-critica-e-perche-lai-act-ue/ on Thu, 14 Mar 2024 15:35:05 +0000.