Public institutions are taking up the issue of artificial intelligence to frame it and prevent abuse. In April 2021, the European Commission presented a draft regulation. This project includes the supervision of so-called “high risk” systems related in particular to facial recognition or systems used in transportation or education. In case of non-compliance, the National Commission for Informatics and Liberties (Cnil) could be in charge of controlling and sanctioning the companies.
To carry out this mission, the Cnil needs to be trained and understand how artificial intelligence algorithms work. The specialized media l’Usine Digitale reports that the Cnil has already tested two auditing tools: IBEX (Interactive Black-box EXplanation) and Algocate.
The Commission worked with Clément Henin, current referendum adviser in extraordinary service of the Court of Auditors and researcher within the Privatics project team of the National Institute for Research in Digital Science and Technology (Inria). asked by himdigital factorydefines AI as “an algorithm that does not have its behavior entirely programmed by a human being but is capable in some way of learning from data (machine learning) or experiences (reinforcement learning).”
Specifically, for IBEX, for example, Cnil agents worked on a fictitious case: a company that offered consumer credit. They were working from an online platform and had to answer questions about fictitious algorithmic decisions.
Promoted by the Council of State
In a study published on August 30, 2022 on a future regulation on AI, the Council of State encourages strengthening the powers of the CNIL.
He seems optimistic about the future and the use of artificial intelligence. For the administrative jurisdiction, it would help the public service in its daily mission, guiding decisions, and reducing user review times.
The public institution therefore encourages the Cnil to become “responsible for the regulation of AI systems provided for in the future European regulation.” The study thus recommends “a profound transformation of the Cnil into a national control authority in charge of regulating AI systems, particularly public ones, to embody and internalize the double challenge of protecting fundamental rights and freedoms, for one hand, and innovation and public action, on the other”.
To achieve this, the Council of State would like to strengthen the power of its institutions in the matter, through a network of existing market surveillance authorities, such as ARCEP and ARCOM.
Source: BFM TV
