Does it regulate the artificial Intelligence ?

on October 13, 2017 – Anne MOREAUX – Right

© A. P. – The Club of lawyers has gathered the top experts from the legal issues and ethical posed by the AI.

“The robots do not have ethics. Ethics, it is we who invent “. There is therefore an urgent need to train themselves to the challenges posed by the algorithms in order to regulate them. This is the observation emerged from the debate organized by the Club of lawyers, the 12 October in Paris, entitled ” Law and artificial Intelligence : what are the impacts ? “.

The debate of human Vs machine is not new, but resurfaced with the progress in the field of intelligence artificial (IA). The man against the robots of Bernanos, published in 1947, spring in the bookstore, and the success of Homo Deus of Yuval Noah and the series Westworld show.

The time of an evening, the think tank legal has brought together the cream of experts of legal issues and ethics that arise in the use of AI. An event that is part of the public debate on the ethical issues of the algorithms launched by the CNIL.

Amal Taleb, vice-president of the national Digital Council, introduced the debate by elaborating on the role of this advisory committee to address the digital challenges of the future in their aspects political, social, and legal. It ensures that the instance is ” the promotion of a digital rooted in his european culture “, with a vision oriented towards the constitutional tradition of the member States of the european who defend the freedoms.

Need of training

For Amal Taleb, it is necessary to be able to ” steal the AI, allow workers to understand it, domesticate it and use it.” It is necessary therefore form. While the panel seems unanimous on the need for training.

“It is necessary to design these machines as helpers to humans, but it must be used with attention,” explains Laurence Devillers, a professor of computer science applied to social sciences, a specialist in man-machine interaction. After having detailed the pillars of AI (perception, decision, action, learning), she warned the audience on the ethical implications serious of this interaction and the risk of manipulation (nudging).

“Education is a key lever” according to Marc Mossé, director of public affairs and legal from Microsoft, because ” the real question is that of the uses of the machine “, what one decides to do.

Franca Salis-Madinier, national secretary of the CFDT Cadres member of a committee of the EESC european who has rendered an opinion on the AI last spring which addresses the major question of knowing what we want as a society, confirms by responding ” a society with the Man in charge and not the machines “.

The other question is that of the enormous impact of the IA on the world of work. France strategy predicts a loss of 10 % of current employment. The training is “vital” to adapt to the AI. There are even “emergency” for Eric Bothorel, member of the presidential majority.

Laurence Devillers, Benoît, Georges, Florence Chafiol, Marc Mossé, and Eric Bothorel. © A. P.

Need for regulation

If the stakeholders do not all agree on how to oversee and regulate the algorithms, all agree on this need.

“It already has sufficient legal foundation to successfully manage the AI with what we have,” says Master Florence Chafiol, associated with August & Debouzy, citing our right of accountability, and the protection of personal data.

One can, for example, apply to the robots the responsibility of the things or animals. “There’s the legal side and the ethics of the other “, the distinguished lawyer, for which there is not a need to reform the law, but rather a need of regulatory ethics. “The robots do not have ethics. Ethics, it is we who invent them, ” adds Franca Salis-Madinier.

The u.s. attorney Winston Maxwell, a specialist in the protection of privacy on the internet, which contributed to the report of the Council of State on the fundamental rights in the digital age, sharing this analysis. According to him, it doesn’t make sense to regulate the AI with a new law, it is enough for the moment to treat on a case-by-case basis its legal issues. The case law and some legislation sectoral the law of 2015 on the autonomous cars will do the trick.

The researcher Laurence Devillers does not agree. According to her, ” we need regulation “, therefore it is necessary to work on the ethics and the legal thinking of the guards. She has participated in the drafting of several reports (alliance Allistene, CNRS).

If the parliamentarian Eric Bothorel is ” not obsessed by the idea of making a law to regulate the AI “, because there are ” already many of the texts are satisfactory “, it proposes to adapt the instances to address specific and complex issues that it generates and evokes the creation of a special Court in the digital age.

Need to experiment with it

The lawyer Winston Maxwell puts in advance the need for companies to make impact studies when they create algorithms, to show the legislator and why not to instances of ethical review.

“There are machines which you don’t needs it,” says Franca Salis-Madinier and “things that we should not develop AI,” adds Marc Mossé, which highlights the need for experimentation and collaborative work with the States and public and private laboratories.

Also, Microsoft has created by the end of 2016, with his friends the giants of the tech (Google, Apple, Facebook, IBM, Amazon…), ” the alliance IA “, a sort of permanent forum, open to all (they say), where one reflects on good practice, including ethical, to promote in the AI.

Ethics should be decided at the state level and not in business according to Laurence Devillers, shocked by these practices.

For the member, the concern is that european companies are excluded from the debate. To remedy this discrimination it would be necessary to facilitate the investment of risk capital, that is why the Government wishes to give the ability for businesses to raise funds in order to create its own GAFAM.

All of this is expensive, in both time and money. In fact, Marc Mossé compared with malice the total budget allocated by the former tenure of the project France IA and the annual budget of Microsoft of 10 million euros, more than 10 times higher.

Leave a Reply