What are the issues?
Artificial intelligence aims to understand how human cognition works and to reproduce it. It refers to technologies that rely on the use of algorithms. These technologies, which have multiple variations, are often characterized by their predictive ability. It is a question of endowing machines with their own intelligence and autonomy.In addition to the impacts on everyday life, they can go as far as changing the boundaries between man and machine. In its report “For a controlled, useful and demystified artificial intelligence” of 14 March 2017, the Parliamentary Office for the Evaluation of Scientific and Technological Choices (OPECST) considers that the advent of a super-intelligence is not part of risks in the short and medium term. It identifies however three major issues are related to artificial intelligence: its economic and social consequences, ethical and legal issues and technological and scientific issues posed by progress in AI.
The main economic and social issues of AI
The OPECST and the working group co-led by the National Council for Digital and France Stratégie, whose report “France IA” is also published in March 2017, draw the conclusion of the emergence of a globalized economy dominated by “platforms”. This “platform” of the economy represents a challenge of international competitiveness, where access to data becomes a comparative advantage. For the first time since the Industrial Revolution, France and Europe are not at the origin of this technological revolution. The United States with GAFAM (Google, Amazon, Facebook, Apple, Microsoft) and China are now duopoly worldwide.
In addition, automation is likely to destroy more jobs than it creates. It also risks creating a two-tier labor market: a minority of highly skilled jobs for an over-graduated elite and a majority of precarious workers whose skills will not pay enough to live. It is also feared a high level of unemployment, created by the substitution of software for lower-skilled jobs.
Both reports question human-machine complementarity in terms of its degree of social acceptability. AI poses a challenge to education and lifelong learning policies. It introduces a potential revolution in the living environment and assistance to people, especially with the appearance of service robots, assistance agents, support and mobility assistance.
Ethical and legal issues related to robotics and artificial intelligences
For example, all domestic or medical applications contain privacy and personal data risks. How to make sure that health data will not be sold to a private company or a future employer? How to combine this massive collection of health data with respect for the “right to be forgotten”? In short, how do you ensure that artificial and robotic intelligence rhymes with ethical data processing?
Faced with a global market for robotics and rapidly growing artificial intelligence technologies and a generalization of robotics in everyday life, the definition of a legal status for robots and other artificial intelligences is debated.
Beyond the “Asimov laws”, which set out general ethical principles, the regulation of supervised or reinforced artificial intelligence systems raises a debate on the development of a law of robotics.
For example, is the European Union considering the possibility of legislation to regulate the rights and duties of this type of autonomous agent on the basis of MEP Mady Delvaux-Stehres’ report on the law of robots? and IAs presented to the European Parliament on 12 January 2017. – * In its March 2017 report, OPECST extends the reflection of the previous May 2016 report on the contours of a robotics law. This right must include the protection of personal data and privacy, intellectual property, liability plans envisaged or possible, the definition of an applicable law depending on the type of autonomous agents (industrial robots, service robots , autonomous cars …).
For its part, the National Commission for Computing and Liberties (Cnil) expressed concern in October 2017 about the protection of personal data in the context of the “smart city” (digital city) and connected vehicles.
Technological and scientific issues
These issues are related to the algorithms used by AI technologies. Algorithms pose four types of questions.
The risk of piracy (drone, autonomous car) or loss of control of AI systems (especially during a military crisis) require to provide a disabling of IA systems.
The introduction of biases upstream of the algorithms, from the data set stage, requires, according to the rapporteurs, to be vigilant especially for the automatic learning algorithms.
The operation of deep learning algorithms through multi-layer artificial neural networks gives, in a certain number of areas, excellent results and very quickly, without it being possible to advance a satisfactory theoretical explanation (for example, Google’s AlphaGo Zero algorithm). This unsupervised treatment of information poses ethical (what human autonomy from a functioning that he does not understand?) And legal (who is responsible?) Problems.
The selection by the algorithms of a content of targeted information, just like the personalized advertising or the logic of construction of the “threads of news” of the social networks, or the false information (fake news), require the setting up verification tools.
The state of artificial intelligence research in France
In its March 2017 report, the Parliamentary Office underlines the place occupied by many public bodies involved in artificial intelligence research. The first is the National Institute for Research in Computer Science and Automation (Inria), the CNRS, the CEA, and various universities and colleges whose work has international visibility. However, OPECST notes that the international recognition of French research is accompanied by a large diaspora of French AI researchers.
OPECST draws the conclusion of a French community of artificial intelligence still insufficiently organized and visible.
The Artificial Intelligence Actors’ Observatory of the Ministry of Higher Education, Research and Innovation lists in an interactive cartography 542 French structures of Artificial Intelligence (research teams and companies).
As part of the mapping carried out for France IA, a work of identification has brought to light a network of more than 250 research teams in pure artificial intelligence as well as in human sciences, for a total of 5,300 researchers, including more 4,000 distributed outside the Paris region.
The development and development of the industry of the IA
In January 2017, the Secretary of State for Digital and Innovation, Axelle Lemaire, launched the “France IA” operation, which brings together 17 working groups bringing together 500 contributors (experts, academic researchers and representatives of the world. private enterprise) responsible for defining a strategy for the development and development of the IA sector in France.
The report of “France IA”, presented to the President of the Republic at the City of Sciences and Industry in Paris on March 21, 2017, confirms the place of France in an area considered essential for the future, but highlights the weakness of large industrial groups in upstream research on artificial intelligence and calls for progress to facilitate technology transfer and experimentation.
The fifty or so “France IA” proposals revolve around five main areas: investment in research and development; training in new issues and trades, starting at elementary school; the transfer of research to industrial and economic applications; the development of an “industrial strategy” to integrate the IA in each sector as well as the continuation of the public debate in order to facilitate the understanding of the societal and economic stakes by the citizens. Under the impetus of this report, President François Hollande announced public funding of € 1.5 billion for the next ten years under the Investments for the Future program.
Six months after the presentation of the “France IA” plan, Mounir Mahjoubi, Secretary of State for Digital, has been entrusted by Prime Minister Philippe to lead a new fact-finding mission on the subject. The conclusions of this mission, entrusted to the mathematician and deputy LREM Yvelines Cédric Villani, are expected in late March 2018. It aims, in the words of the Secretary of State, “to propose a strategy for the next few years (and) define if a national debate is to be conducted and on which topics. “