Luc Leo Steels (born 22 January 1952 in Belsele, Sint-Niklaas, Belgium) is a Belgian scientist and artist.^[4] Steels is considered a pioneer in artificial intelligence in Europe who has made important contributions to expert systems, behavior-based robotics, artificial life, and evolutionary computational linguistics.^[1]

He was educated at the University of Antwerp and MIT, where his 1977 dissertation was titled "Aspects of a modular theory of language", under advisors including Marvin Minsky and Hugo Brandt Corstius. Catalan Wikipedia states that he is director of the Artificial Intelligence Laboratory at the Vrije Universiteit Brussel, while English Wikipedia indicates he was formerly its founding director. He is also the founder and former founding director of the Sony Computer Science Laboratory in Paris, currently directed by François Pachet.^[1] Steels has also been affiliated with Schlumberger and has worked with researchers such as Pattie Maes and Pierre-Yves Oudeyer. In Barcelona, he has served as a visiting researcher at Pompeu Fabra University and as a fellow of the Catalan Institution for Research and Advanced Studies (ICREA) associated as a research professor with the Institute of Evolutionary Biology (UPF/CSIC).^[1]

Steels was born in the town of Belsele near Antwerp, Belgium.^[5] He studied German philology at the University of Antwerp, where he obtained an M.A. in 1974 and a Ph.D. in linguistics in 1977 under the supervision of Hugo Brandt Corstius, with a thesis in computational linguistics on a parallel model of parsing. Steels pursued another master's degree in Computer Science at MIT, specializing in AI under the supervision of Marvin Minsky and Carl Hewitt, and graduated in 1979.

In 1980, Steels joined the Schlumberger-Doll Research Laboratory in Ridgefield, Connecticut to work on knowledge-based approaches to the interpretation of oil well logging data and became leader of the group who developed the Dipmeter Advisor, which he transferred into industrial use while at Schlumberger Engineering, Clamart. In 1983, he was appointed as a tenured professor in Computer Science with a chair in AI at the Free University of Brussels (VUB). The same year he founded the VUB Artificial Intelligence Laboratory and became the first chairman of the VUB Computer Science Department from 1990 to 1995. The VUB AI Lab focused initially on knowledge-based systems for various industrial applications but gradually focused more on basic research in AI.

In 1996, Steels founded the Sony Computer Science Laboratory (CSL) in Paris as a spin-off from the Sony Computer Science Laboratory in Tokyo directed by Mario Tokoro and Toshi Doi, and became its acting director. The laboratory targeted research in AI, particularly on the emergence and evolution of grounded language and ontologies on robots, the use of AI in music directed by Francois Pachet, and contributions to sustainability directed by Peter Hanappe. In 2011, Steels became a fellow at the Institute for Research and Advanced Studies (ICREA) and research professor at the Universitat Pompeu Fabra (UPF) in Barcelona, embedded in the Evolutionary Biology Laboratory (IBE), where he pursued further fundamental research in the origins and evolution of language through experiments with robotic agents. From 2018 he started to work in Venice within various European projects, first at Ca' Foscari University within the Odycceus and AI4EU projects and then at Venice International University within the MUHAI project.^[6,7]

Throughout his career, Steels undertook research and educational visits to other institutions, serving as a lecturer at the Theseus International Management Institute in Sophia Antipolis, developing courses for the Open University in the Netherlands, being a Fellow at the Wissenschaftskolleg in Berlin (2009–2010 and 2015–2016) and Goldsmiths College London from 2010, and acting as a visiting scholar or lecturer at La Sapienza University Rome, Politecnico di Milano, the University of Ghana, and Beijing Jiaotong University, among others. Steels was a member of the New York Academy of Sciences, and is an elected member of the Academia Europaea and the Royal Belgian Academy of Arts and Sciences (Koninklijke Vlaamse Academie voor Wetenschappen en Kunsten), where he serves as chairman of the Natural Science section.^[8] Steels received several awards including the best paper award at the European Conference in AI in 1982, the Franqui chair at the University of Leuven in 2018, the EurAI Fellowship, the Calewaert chair at the Free University of Brussels in 2024, and the EurAI Distinguished Service Award, presented every two years to an individual who has made exceptional contributions to the AI community in Europe.^[9,10,11,12]

The scientific work of Steels is trans-disciplinary, focusing on finding conceptual breakthroughs in AI, building the technical tools to work out and develop these breakthroughs, and developing concrete experiments to turn the breakthroughs into viable new AI paradigms. Since the early 1980s and using this approach, Steels has played a role in four profound conceptual shifts: from heuristic rule-based systems to model-based knowledge systems, from model-based to behaviour-based, artificial Life inspired robots, from static, engineered language systems to dynamic, evolving emergent communication systems with features of human languages, and most recently from data-driven AI to AI capable of understanding and forms of awareness.

The early 1980s saw a period of interest in the use of the rule-based paradigm for building expert systems. Expert systems are intended to assist human experts in tackling challenging problems, such as medical diagnosis (e.g. MYCIN) or the configuration of complex technical equipment (e.g. R1). By the mid-1980s these techniques became used in industry and integrated in software engineering practice, but the exclusive focus on heuristic rules was limiting, primarily because of the efforts involved in finding an adequate set of rules and because of brittleness seen when there were cases that fell outside the scope of the rules.

From 1985 a trend among AI researchers, including Balakrishnan Chandrasekaran, William Clancey, Doug Lenat, John McDermott, Tom Mitchell, Bob Wielinga, and others, arose to capture human expertise in more depth. Triggered by Allen Newell's paper on the need to adopt a 'knowledge-level' analysis and design strategy, the new generation of knowledge systems used models of the problem domain based on an explicitly represented ontology and employing problem solving strategies to compose tasks into subtasks and solving them.^[14,13] Heuristic rules were still relevant but they would now be learned by first solving a problem using models and inference strategies and by then storing the solution, after some degree of abstraction.^[15] The key advantages of this knowledge level approach are more robustness, because the system can fall back on deeper reasoning when heuristic rules are missing, a richer explanation facility because of the use of deeper models, and a more methodical design process including techniques for verification and validation.^[16]

Steels played a role in establishing this new paradigm in the 1980s, organising a number of workshops and tutorials, helping to develop knowledge level design methodologies, particularly in collaboration with Bob Wielinga and the CommonKADS approach developed at the University of Amsterdam, and publishing papers outlining the knowledge level approach.^[19,17,18] With his team at the AI Lab of the Vrije Universiteit Brussel, he developed various tools, such as the knowledge representation system KRS, which was a frame-based object-oriented extension of LISP with facilities for truth maintenance, meta-level inference and computational reflection.^[22,20,21] The team applied the approach for building challenging operational expert systems in various technical domains (electronic circuit design for digital telephone, scheduling of Belgian railway traffic, monitoring of subway and diagnosis of nuclear power stations).^[23] These systems became used in real operation and ran on the innovative Symbolics LISP machines. It all lead to the creation of a spin-off company Knowledge Technologies (with Kris Van Marcke as CEO) to further channel these developments into practical industrial use. The company was active from 1986–1995.

Around 1986, Steels opened a second research line in his VUB laboratory to develop a new paradigm for AI inspired by living systems. Because this paradigm arose as part of the movement towards Artificial Life, it became known as the Artificial Life approach to AI or, due to the focus on behavior, as the behavior-based approach to AI and robotics, as well as the animat approach.^[25,24] The behavior-based paradigm was intended to be complementary to the knowledge-based paradigm, which targets deliberative intelligence, by tackling reactive intelligence for real-time adaptive behavior of autonomous robotic agents embodied in real-world environments.^[26] This new research line stood at the confluence of several emerging trends in the late nineteen-eighties and nineteen-nineties: a revival of cybernetic reactive robots spearheaded by Rodney Brooks, the establishment of Artificial Life as a new discipline shaped by Chris Langton, a renewed focus on emergent computation through self-organisation using cellular automata, models from chaos theory, and genetic algorithms, and the rise of multi-layered neural networks initiated by David Rumelhart and James McClelland.^{[30,27,28,29]}

Steels actively helped establish the new paradigm by organising a series of workshops, conferences, and summer and spring schools, as well as by writing influential papers defining it.^[33,31,32] With his team in Brussels, he developed hardware platforms (utilizing self-designed processing boards, Lego, and simple electronics parts, led by Tim Smithers) and software platforms, including PDL (Process Description Language).^[35,34]

He also established various robotic experiments, most notably the self-sufficiency experiment initiated with ethologist David McFarland.^[36,57] The self-sufficiency experiment was based on Walter Grey's electric tortoise experiment from the 1950s. This experiment featured simple automatons (animats) capable of wall following, phototaxis, and locating and using a charging station. The McFarland-Steels experiment introduced the challenge of multiple robots competing for energy at the charging station, requiring the robots to do work.^[37] The experimental setup served for 10 years as a framework for experiments in adaptive behavior, genetic algorithms, and reinforcement learning by students at the VUB AI Lab under the leadership of Andreas Birk.

In 1995, after visiting the Sony Computer Science Laboratory in Tokyo at the invitation of Mario Tokoro, Steels initiated a new research direction, applying evolutionary concepts from Artificial Life and developments in behavior-based robotics to investigate how a population of agents could self-organise an evolving adaptive language to communicate about the world perceived via their sensory-motor apparatus. A new team of collaborators was formed across the VUB AI lab and the Sony Computer Science Laboratory in Paris, conducting research on this topic from 1995 to 2015.

Initial breakthroughs occurred around 1996 in phonetics and phonology, where Steels introduced a self-organisation approach to explain the origins of speech sounds and phonetic structures. Experiments were conducted in which agent populations with basic vocal and auditory apparatuses developed a shared repertoire of speech sounds by playing imitation games, introducing sound variations, and adapting to one another's sounds. These experimental setups were elaborated in the Ph.D. dissertations of Bart de Boer and Pierre-Yves Oudeyer.^[39,38]

Concurrently in 1995, Steels introduced the Naming Game to investigate the formation of lexicons and the emergence of linguistic conventions in general.^[40] In this language game, a speaker selects a topic and utilizes one or more words to direct a listener's attention to that topic during each interaction. The game succeeds if the reader pays attention to the listener's chosen topic, leading both agents to reinforce their existing word inventory; otherwise, speakers invent new terms, listeners adopt them, and both update the associative scores linking words and meanings in their inventories. In concrete experimental settings, agents begin without an initial vocabulary, progressively inventing words and coordinating their usage through local interactions, which leads to a coherent emerging vocabulary that persists despite population turnover or the introduction of new topics.^[41]

Steels introduced the Discrimination Game in 1996 to examine the origins of meaning, followed by the Syntax Game in 2014 to study syntactic emergence.^[43,42] The Language Game paradigm proved effective for analyzing numerous aspects of language emergence and evolution, starting with theoretical proofs by Bart de Vylder and Karl Tuyls in 2005 showing that populations can achieve coherence, and the 2007 identification of scaling laws relating population growth to topic sets by Andrea Baronchelli and Vittorio Loreto.^[45,44] As evolving languages expanded to encompass morphology and syntax across broader conceptual domains, Steels conducted in-depth studies on color languages (with Tony Belpaeme and Joris Bleys), case systems (with Remi van Trijp and Pieter Wellens), spatial language (with Martin Loetzsch and Michael Spranger), agreement systems (with Katrien Beuls), determiners (with Simon Pauw), and action languages (with Martin Loetzsch, Michael Spranger, and Sebastian Höfer).^{[46,54,56,48,51,52,53,55,58,59,60,61]} Many of these developments were validated in robotic platforms, starting with simple Lego vehicles, progressing to vision-based agents in the 'Talking Heads Experiment', and eventually using the four-legged Sony AIBO robot and the Sony humanoid robot QRIO.^{[54,56,48,51,52,53,55,58,59,60,61]}

Steels promoted the language game paradigm through summer schools (Erice in 2004 and 2006, Cortona in 2009 and 2013, and Como in 2016), founding the Evolution of Communication journal, and publishing significant papers and edited volumes on language evolution.^[62] He also advanced experimental tools, notably the BABEL software platform for language emergence research and Fluid Construction Grammar (FCG), a formalism designed for representing emergent grammars.^[63,64] Beginning in 2000, Fluid Construction Grammar underwent multiple design iterations to become the primary operational framework for computational construction grammar.^[65]

From around 2018, at the peak of advancements and applications in data-driven neural network style AI, Steels began participating in efforts to create a human-centric form of AI. Together with Ramon Lopez de Mantaras, he launched the Barcelona declaration for the proper development and usage of artificial intelligence in Europe in 2018, which influenced the European Ethical Guidelines for Trustworthy AI published in 2019.^[67,24] He also initiated the ethical AI workpackage in the European Commission's large-scale AI4EU coordination project.

Steels argued that more than regulations is needed to make AI more human-centered and launched a number of projects to combine reactive intelligence, captured through neural network style systems, with the deliberative intelligence that was the focal point of earlier AI research.^[68] Concretely, the EU project MUHAI focuses on how the level of understanding in AI systems could be increased by building rich models of problem domains and problem situations and integrating a variety of knowledge sources (ontologies, language, vision and action, mental simulation, episodic memory, and context models), and the EU project VALAWAI focuses on how AI systems can be made 'value-aware' by introducing attention mechanisms to deal with highly complex, uncertain fragmented inputs, and a component implementing 'moral intelligence'.^[69,70]

In the early 1970s, Luc Steels became active in performance art and avant-garde electro-acoustic music. In 1972 he founded the collective 'Dr. Buttock's players pool', and went on to participate in the Welfare State theatre in 1977 as well as with performance artist Hugo Roelandt.^[71]

In music, he was part of the 1970s Antwerp Free Music scene, playing guitar in a style created by Derek Bailey. In 1971 he co-founded the ensemble Mishalle-Geladi-Steels (MGS) with saxophonist Luc Mishalle and electronic musician Paul Mishalle. The ensemble performed with the Studio for New Music set up by Joris De Laet, particularly at the ICC in Antwerp.^[72] Lifelong interactions originated from this period with artist Anne-Mie Van Kerckhoven, who Steels had invited as artist in residence at the University of Antwerp and later at the VUB AI laboratory in Brussels, and Peter Beyls, who was also artist in residence at the VUB AI Lab.^[73,75]

After a period of total focus on scientific work while in the United States, Luc Steels returned to artistic activities from the 1980s onwards. Thanks to an encounter with H-U Obrist at the Burda Akademie symposium in Munich in 1995, he came into contact with artists, resulting in public presentations in art contexts such as at the Bridge the Gap encounters (2001 Kitakyushu), the Memory Marathon (Serpentine Gallery, London, 2007 & 2012), and the Experiment Marathon (Reykjavik 2008).^{[80,76,77,78,79]} Within this artistic network, Steels collaborated with several artists for the co-creation of new works, including with Carsten Holler (for the CapC Musee in Bordeaux and the Koelnerische Kunstverein); with Olafur Eliasson for a piece 'Look into the box' for the Musee d'art moderne in Paris in 2002 and later shown at the Festival dei 2 Mondi (Spoleto, 2003), the ExploraScience Museum (Tokyo, 2006), and other locations; with Sissel Tolaas for work shown at the Berlin Biennale; with Anne-Mie van Kerckhoven at the NeuerAachenerKunstverein; and with Armin Linke and Giuliana Bruno for the New Alphabeth (Stop Making Sense) exhibition at the Haus der Kulturen der Welt (Berlin).^{[81,82,83,84,85]} Steels also participated with his own installations in various art-science exhibitions, the most important ones being Laboratorium curated by H-U Obrist and B. Vanderlinden in Antwerp in 1999, and N01SE in Cambridge (Kettle's Yard) and London (Wellcome Gallery) in 2000, curated by Adam Lowe and Simon Schaffer.^{[81,82,83,84,85]}

Beginning in 2010, music and theatre converged in two opera projects featuring Luc Steels as composer and neuroscientist Oscar Vilarroya as librettist.^[88] The first opera, entitled Casparo, premiered at the Palau de la Musica in Barcelona in 2011 and was subsequently performed in Brussels at Theatre Moliere (2013), Tokyo at the Sony Concert Hall (2013), Leuven at Iers College (2014), and Paris at Jussieu Theatre (2014).^[89,90] The second opera, titled Fausto, had avant-premiere performances at La Gaite Lyrique in Paris in 2016 and at the Monnaie Opera House in Brussels in 2017, followed by full performances at the And&MindGate Festival in Leuven in 2018 and at the Homo Roboticus event at the Brussels Monnaie Opera House in 2019.^[93,91,92] Most of these performances were conducted by Kris Stroobants with the Frascati Symphonic Orchestra, the choir La Folia, and various soloists, including Reinoud van Mechelen and Pablo Lopez Martin of the Mallorca opera. Written in a neoclassical, postmodern musical style, the operas explore societal and transhumanistic issues raised by the use of artificial intelligence, including the occurrence of a singularity and the possibility of achieving immortality through virtual agents.

Luc Steels curated a number of international exhibitions, including Intensive Science at La Maison Rouge in Paris (in 2006 and 2008), artes@ijcai at the Centro Borges in Buenos Aires (Argentina) in 2015, and the 'Aqua Granda. Una Memoria Digitale' exhibition at the Science Gallery Venice in 2021.^[94,95] He contributed essays on art and music for journals such as KunstForum and Janus Magazine (Issue 20) and for exhibition catalogs, and also wrote academic papers on computer music and art interpretation.^{[101,96,97,100]}

In 2020, Steels was S+T+ARTS 'scientist in residence' at the Luc Tuymans art studio, 'Studio Tuymans' in Antwerp, which resulted in an exhibition at the BOZAR museum in Brussels based on the use of AI methods to interpret a single art work by painter Luc Tuymans called 'Secrets'.^[102] In 2023, Luc Steels curated an exhibition 'Science on the Edge of Chaos' at the Royal Library in Brussels, focused on research during the 1980s and 1990s about chaos theory and its application in different sciences.^[103,104]

Luc Steels's publications include the 1990 article "Components of Expertise" in AI Magazine (volume 11, issue 2, pages 28–49). In 2001, he authored "Grounding Symbols through Evolutionary Language Games," included in Simulating the Evolution of Language, edited by A. Cangelosi and D. Parisi and published by Springer. In 2011, he published "Design Patterns in Fluid Construction Grammar" with John Benjamins Pub in Amsterdam, as well as "Modeling the cultural evolution of language" in Physics of Life Reviews.