Europe wants to regulate Artificial Intelligence but perhaps it is already late
robort - 2023-06-03 13:52:00
The explosion of Artificial Intelligence has literally created a fundamental problem between those who designed it and those who instead suffer the consequences of its use. At this point, one wonders how far it is possible to go with this type of technology in order not to create problems for others.
If Europe is asking for it, it wants to somehow run for cover and protect its citizens and would like to implement it with a regulation, the so-called ''Ai Act'', i.e. an organic regulatory act on Artificial Intelligence that has never been implemented debate by the European Parliament.
The acceleration by the European bodies goes hand in hand with that the explosion of ChatGPT and all that has followed. In fact, however, what appears to the eyes of the most expert is a formal delay on the part of politics in actually following the development times of these technologies and in general of all technologies that take hold much faster than a hypothetical response legislation.
And the Ai Act, presented in 2021, has yet to be discussed and then voted on by the European Parliament in April and subsequently negotiated in the trilogue: Parliament, Commission, and Council of the European Union, with times that are certainly not short.
Ai Act: what is the European regulation about?
The AI Act has "the objective of defending rights with the need to compete in a global market at an advanced stage" These are the words of MEP Brando Benifei, co-rapporteur of the regulation.
The idea is to essentially face the challenges posed by ChatGPT with a regulation that will have to have a '' text with a mechanism that makes it adaptable to future challenges, as far as possible".
Yes, because the development of these new technologies evolves almost daily and the idea of fixing them in a static regulation could have more negative than positive implications for the defense of citizens or of those who in any case use this type of technology.
Moreover, between the European Council and the Parliament, there are many opposing ideas that still need to be judged and which could cause the idea of a rapid approval of the regulation to be postponed.
And let's talk about opposing positions on real-time facial recognition, emotion identification systems, the risk of disinformation with ChatGPT, or the risk of supporting regimes that use Artificial Intelligence for repression.
In short, on many other aspects, there is a controversy between Parliament, the Council, and the Commission, and the idea is to find the right balance by spring to follow up on the regulation.
For some, however, releasing a regulation that keeps Artificial Intelligence ''at bay'' or that, in any case, allows the defense of those who use it or those who perhaps suffer from it, may in any case not be useful or completely useful above all on the basis of the definition that will be given of Artificial Intelligence in regulation.
In fact, if we pretend to describe AI as a ''list of methods created by machine learning'', we will only allow the regulation to be easily circumvented. Yes, because there will be companies that in some way will create "non-AI methods" capable, however, of functioning as Artificial Intelligence.
The idea would be to create a regulation that would consider all the parts, or rather the components, of an Artificial Intelligence and not in a more general way because the technology that underlies an AI system is made up of ''components'' and therefore it would be necessary to consider a regulation to keep each type of component at bay.
In short, the question is really difficult and it will certainly not be easy to quickly implement a regulation on Artificial Intelligence that is truly effective and above all is '' adaptable over time''.
What we certainly know is that these new technologies are advancing at a frenetic pace and if we want to somehow ''defend'' ourselves, we can do nothing but run at their pace so as not to fall behind.