Researchers of l’X co-organize a conference on AI at the European Parliament
The conference “AIvolution - Transforming the European Union´s Economy and Society” gathered over 60 scholars and researchers from several universities in Europe, including École Polytechnique and other founding partner institutions of IP Paris, MEPs (Members of the European Parliament) and other EU officials, and experts from industry, at the European Parliament, on November 16, 2023.
Artificial Intelligence (AI), digital technologies, and, more precisely, their underlying trade-offs and their potential - positive as negative - to transform the economy and society were at the center of the discussions introduced by MEP Eva-Maria Poptcheva.
Taking the presentation of several technical and scientific aspects of the latest AI applications as a starting point, the conference AIvolution addressed the underlying foundations of digital technologies that allow massive data collection, transmission, and computing - at the core of AI and digital services -, and subsequently the political, social, and environmental implications of the successive technological innovation cycles. The participants stressed that increased dependency on technology requires collective deliberation and a comprehensive democratic regulatory response on the EU level.
Several researchers and professors of École Polytechnique, Télécom Paris, and Télécom SudParis took part in the debate. Other than Juan-Antonio Cordero-Fuertes, associate professor at École Polytechnique who co-organized the conference, Hervé Debar, professor and researcher at Institut Polytechnique de Paris, Marceau Coupechoux, professor at Institut Polytechnique de Paris, Ruta Binkyte, researcher at École Polytechnique (LIX) and doctoral Machine Learning Ethics researcher at Inria, Sonia Vanier, researcher at École Polytechnique and professor at Paris1 Panthéon-Sorbonne University, Oana Balalau, associate professor at École Polytechnique and researcher at Inria, and Oana Goga, researcher at École Polytechnique (LIX) and CNRS.
Along four thematic sessions, the conference allowed EU officials, researchers, and industry experts to discuss opportunities, challenges, and potential threats - both political and technical – that digital technologies and AI entail. Advances in Large Language Models (LLMs), progress in robotics, or the possibilities for rapid information retrieval and processing indeed represent significant opportunities. However, digital technologies and AI also bring many challenges with them.
Challenges linked to digital technologies and AI
Hence, the second session focused on two challenges digital technologies entail: the first linked to big data and digital rights and the second to social networks' role in triggering disinformation. The exponential increase in data collection and traffic enables powerful AI-enabled technologies and services but requires, on a political and institutional level, the enforcement of legal mechanisms to provide state or EU-level protection of digital rights and personal data privacy.
Chaired by Juan-Antonio Cordero-Fuertes, this second panel also addressed the potential of AI as information retrieval technology for democracy. On the one hand, it represents a powerful tool to increase the quality of public debate (e.g., through mechanisms for fact-checking, automatic disclosure of conflicts of interest, and fallacy detection). Still, it is also dangerous as it can influence and even distort or destabilize the conditions in which deliberation and public decisions should occur, e.g., through political-motivated advertising campaigns or deliberate misinformation attacks over social platforms.
The researchers stressed the need to make disclosure and deeper transparency about algorithms and personal data collection binding for social platform operators. They also highlighted the importance of providing less limited access to data and reinforcing legal protection for scientists working on these matters. The panel discussion identified these measures as critical for a better understanding of the effects of data- and AI-driven algorithms in social networks, ensuring data regulation compliance, and protecting user digital rights in the EU. More generally, the session illustrated the relevance of a broader discussion, technical and political, to determine the necessary European regulation to protect fundamental digital rights (personal data privacy, children rights), to assess digital sovereignty in the EU, and to provide protection against novel disinformation threats on social networks.
Energy, sustainability, and cybersecurity
During the third session, which further analyzed challenges, opportunities, and trade-offs linked to digital infrastructures supporting interpersonal communication, data collection, transfer, and processing, the speakers focused on energy, sustainability and cybersecurity aspects.
Cybersecurity risks increase as essential services such as healthcare systems, transportation, or energy facilities become digitized. Researchers and professionals who participated in the panel elaborated on how AI and digital technologies may help respond to the security threats against digital sovereignty.
This session also addressed a problematic side effect of digitization: the increasing energy consumption due to data traffic for digital services. Participants argued that the energy consumption it entails put digitization at odds with the objectives of energy sobriety of the ecological transition envisioned by the EU. Furthermore, the panel participants also discussed the context and dynamics linked to procuring the raw materials needed to manufacture the devices and maintaining the infrastructures enabling digital services. The global supply chains, which imply mining, production, and distribution of critical raw materials, are subject to increasing geopolitical tensions, affecting also the EU.
Scholars in history, philosophy, law, and computer science participated in the last panel dedicated to examining the social impact of technological change. Technology – and thus AI technology and the intertwined Internet-based technologies - drives social transformation. However, it is also a product of the societal conditions in which it has been developed. As such, it strongly depends on the expectations and optimization objectives that dominate society and the public discussion. From a more philosophical and sociological perspective, the speakers also discussed how current technology trends modify democracy and encourage a certain kind of citizenship, subject to specific, data-driven policies and incentives.
As digital technology becomes more pervasive and is increasingly deployed for essential services, its progress, insights, trade-offs, and implications for society must be examined and discussed on a broader level as part of a multi-disciplinary and steady consultation among scientists and engineers, researchers and professionals, policymakers, lawmakers, and the civil society. The conference “AIvolution”, hosted at the European Parliament, is part of that effort for scientific and technical dissemination and discussion, to which École Polytechnique is highly committed.