Every weekend in the summer, the editorial team of Duty offers you a reflection on the social issues that will shape our world in the coming years. Individual and collective challenges will challenge us constantly on these issues, which we will approach from the angle of solutions as far as possible. Today: artificial intelligence (AI).
In conclusion ofHomo Deus, Yuval Noah Harari was asking a fundamental question. What will happen to society, politics and everyday life when intelligent, but mindless algorithms know us better than we know ourselves?
As early as 2015, the philosopher and historian predicted some of the upheavals we are experiencing. The released algorithm produces a deluge of data with which the organic brain can no longer deal, to the point of threatening obsolescence to humans and the institutions they have created. Even mythomaniacs, full of prejudices and intoxicated by disinformation, these robots endowed with the faculty of continuous learning give rise to the feeling that humanity is overwhelmed. AI is to be placed in the category of revolutionary paradigm shifts, alongside agriculture, printing, steam or the microprocessor. And we are not ready yet.
The panicked reactions of AI luminaries calling for a moratorium on the deployment of systems testify to this crisis of modernity. Changes are happening too quickly for our ability to adapt. In Release, the psychologist Frédéric Tordo puts forward the hypothesis that AI “is embodied for the first time in language, which is specific to man and from which it seems to dispossess us”, hence this “anxiety of replacement” rather ill-founded. always in Release, the mathematician Cédric Villani plays down the use of AI, which is already widespread in a host of applications and systems (GPS navigation, self-correction, work software, etc.). Supporter of a “third way”, in the same way as one of the founding fathers of AI, Yann Le Cun, he calls for “taking the subject head-on”, an avenue rich in possibilities.
Beforehand, democratic nations will have to define, domestically and internationally, the risk factors and the roadmap that will allow AI tools to be deployed in accordance with their values and principles, and that they are subject to a validation mechanism that balances the protection of human dignity and innovation.
Judging by the toll of the platform age, which is coming to its zenith, this will be a huge challenge. In the attention economy, we have become more or less willing guinea pigs for the commercial strategies of the digital giants, resulting in an erosion of privacy and the transformation of our personal data into new black gold, even if it means tolerating pass division, cat videos and misinformation. Provided that we remain glued, docile, to our touch screens. We cannot wait or hope that the libertarian ethic serving as the inner compass of the digital giants will generate, by one of these unexpected miracles, democratic practices in the use of AI.
Democratic nations are particularly at risk from the deployment of AI, which has the potential to undermine the foundation of established institutions and fundamental rights and freedoms if the algorithms are activated by malicious or greedy minds. The risks posed by AI include, among others, the automation of cyber espionage, the deployment of the digital warfare arsenal, disinformation, the perversion of the electoral system, the trivialization of hateful and discriminatory speech aimed primarily at women and people from diverse backgrounds.
The issue of discrimination, bolstered by an evil empire of misinformation, is particularly urgent, as it undermines the very foundations of equality, diversity and human dignity, all of which are concerns of Bill C-27. in Canada. For the time being, the European Union has developed the most comprehensive legislative framework, with the ambition of inspiring international standards. The law voted by MEPs, which is supposed to come into force in 2026, will notably prohibit the use of facial recognition systems in public places by law enforcement, emotion recognition systems and behavior profiling systems ( social rating). Suppliers will be forced to reveal the copyrighted data that was used in the development of the algorithms (texts, images, sound clips, photos, etc.). There will also be an obligation to inform users if they are in contact with a machine and to disclose the creation of artificial images by means of AI. Finally, there must be, at all times, human control over the machine and increased transparency on the functioning of the algorithms in systems deemed “high risk”.
These are sound democratic bases for framing AI. We are the only ones capable of ensuring that it will remain at the service of the values which are the essence of the human experience in what is most precious: mutual aid, equity, social justice and emancipation. , rather than control, lies and technological obscurantism.