"Reclaiming our Humanity in the Age of AI"

Artificial Intelligence has quite suddenly become a technology that is invading all aspects of our lives. In many cases this has very beneficial effects. For example, to battle the Coronavirus epidemic, AI is used for early warnings, analyzing and predicting trends in infections, decoding the genome of the virus and how it evolves, tracking and tracing contacts of infected patients, synthesizing and retrieving information from the rapidly growing scientific literature, the search for vaccines, management and provisioning of medical stocks, and prediction of future infection waves.

But questions are also raised about the impact of this technology on humans. A contact tracing app can also be used to violate our privacy, the AI behind social media can be used to spread disinformation and engage in political manipulation,

recommender systems can be biased in favor of sellers, the dominance of English-speaking media in the digital sphere can lead to the impoverishment of other human languages, particularly if spoken by smaller communities. There is an urgent need for humanists to get involved with the ethical, social and cultural issues raised by AI.

In addition there is a fundamental issue with the core technology now used as the basis for AI, namely machine learning with big data. This technology has its roots in behaviorist psychology, which was hotly contested in the nineteen sixties because it was ignoring what was making humans unique, namely meaning. The translation systems currently available (such as Google Translate) translate on the basis of shallow mappings between source and target language, not by first reading the text, analyzing the meaning, and reproducing the content of the text. Image recognition is based on pattern recognition after seeing millions of images but the AI system has no clue what it is seeing, the context, or why the image was made. There are other approaches in AI which have stronger roots in cognitive psychology and other humanities disciplines. They have grappled with the question of meaning but results of this research is more difficult to put into practical usage. To achieve meaningful (and therefore more human-centric and robust) AI humanists should get involved on a much larger scale than is currently the case. Seeking meaning rather than shallow pattern recognition is the hallmark of human intelligence, and AI can only fulfill its promise when it finds methods how to do this.

Luc Steels is an ICREA research professor at the Institute for Evolutionary Biology (UPF/CSIC) in Barcelona and visiting professor at University Ca’Foscari in Venice. He studied linguistics at the University of Antwerp and computer science (with specialization in AI) at MIT (United States). His work has spanned many areas of Artificial Intelligence, from knowledge representation to expert systems, cognitive robotics, and natural language processing.

He is particularly well known for modeling of the cognitive and social processes that underlay the emergence and evolution of language using robotic agents. Steels was founding director of the Artificial Intelligence Laboratory at the University of Brussels (VUB) and founding director of the Sony Computer Science Laboratory in Paris. He published a dozen books on AI and hundreds of papers in top level journals such as Behavioral and Brain Science, Cognitive Science, Constructions and Frames, Artificial Intelligence, Nature Physics, etc., with an H-index of impact > 70. He is an elected member of the Royal Academy of Belgium and the European Academy. Besides his scientific work, he has also been active in the arts. He created works for various exhibitions such as Laboratorium in Antwerp, wrote a theater play for the Avignon Theatre Festival with Jean-Francois Peyret, and composed two operas (Casparo and Fausto) that were performed in a dozen venues including the Monnaie Opera house in Brussels, Gaité Lyrique in Paris, and the Sony Concert Hall in Tokyo.

Please comment and discuss below.

To return to the conference overview page, click here.