Gide | Artificial Intelligence, European legislation and legal issues

EACCNY Member Gide Loyrette Nouel is pleased to present an analysis of “Artificial Intelligence, European legislation and legal issues”, which aims to introduce the European legislation currently being adopted (AI ACT in the version approved on 2 February 2024), but also to examine the legal issues of AI across 10 sectors of activity.

Author: Thierry Bonneau – Senior Counsel, Gide Loyrette Nouel 

The genesis of the notion of artificial intelligence (“AI”) is a subject of debate. Some trace it back to the founding fathers of information technology1 . Others believe the expression was coined by Johan McCarty and Marvin Minsky, during a lecture given at the Dartmouth College in the summer of 19562

The use of the words “artificial” and “intelligence” may come as a surprise, as the notion of AI refers to machines and robots. “Man, whose intelligence was believed to be the distin -guishing mark, the mark of nobility, has conceded this quality to a digital interface, a machine”3 . AI refers to a robot’s acquisition of human cognitive skills, and therefore of human knowledge and reasoning abilities.

What is AI?

There are many forms of AI. Different approaches have been proposed. If we simplify the oppositions, it is possible to distinguish two forms of AI4:

◆ a so-called weak AI, “where it is possible to conceive that reasoning per se is not necessary, and that the machine me-rely translates innate animal-like reactions into a form of symbolic representation”5;

◆ a so-called strong AI, “where it is possible to conceive that the full range of human intellectual capacities can be reproduced, as the machine can learn from statistical data and thus go beyond the initial symbolic representation”6.

From this perspective, AI is “a process by which an algorithm evaluates and improves its performance without human intervention”7: it designates a “self-learning mechanism”8 .

AI comprises two essential elements: an algorithm, on the one hand, and data, on the other hand9. The algorithm is “the description of a finite and unambiguous sequence of steps (or instructions) for producing results (output) from initial data (input)”10. Data is “information in digital form that can be transmitted or processed” by a computer (Merriam-Webster dictionary definition). In light of these elements, AI can be defined as “a computer system based on an algorithm endowed with cognitive capacities developed with the help of data, enabling it to be autonomous in the choices it makes and the decision-making it carries out”11.

In 2017, the Financial Stability Board (FSB) defined AI “as the theory and development of computer systems able to perform tasks that traditionally have required human intelligence. AI is a broad field, of which ‘machine learning’ is a sub-category. Machine learning may be defined as a method of designing a sequence of actions to solve a problem, known as algorithms, which optimize automatically through experience and with limited or no human intervention”12 .

This approach was taken into account by the European Union (“EU”) in its White Paper published in 2020: “AI can perform many functions that previously could only be done by humans” (Ibid p 13) and in the proposal for a regulation of the European Parliament and of the Council laying down harmonized rules on artificial intelligence (AI Act) and amending certain Union legislative acts adopted by the Commission on April 21, 2021 (hereafter “Proposal for an AI Act”13 ).

To ensure that the definition of an AI system provides criteria that is precise enough to distinguish AI from simpler software systems, the compromise agreement reached by the EU institutions on December 8, 2023 aligns the definition with the approach proposed by the Organization for Economic Cooperation and Development (“OECD”) (see Section 1.2 p.

Read the full analysis here: Gide_Booklet_Artificial_Intelligence_March_2024

Compliments of Gide Loyrette Nouel – A member of the EACCNY