Ateliers/Séminaires/Colloques

While language models provide a working implementation that surpasses symbolic theories in many respects, Piantadosi also presents work based in early computer science that seeks to formalize what latent structures must be present in a system in order to generate its observed behavior. This approach is the topic of a forthcoming open textbook, and the approach holds promise for understanding if any grammar-like structures are necessarily present in, for instance, statistical language models. This line of work also points to ways we can rigorously connect neuroscience to behavior.

Steven T. Piantadosi, Professor in the Psychology Department and Helen Wills Neuroscience Institute, University of California, Berkeley, leads the Computation and Language Lab (CoLaLa). His computational and behavioral research is on the learning of language and concepts, the evolution of human-like cognition, and how ambiguity can serve communicative functions. In his critique of Noam Chomsky’s theory of Universal Grammar, Piantadosi argues that the success of large language models (LLMs) challenges most assumptions of standard linguistic theories.

As large language models (LLMs) increasingly permeate scientific research, their use for generating or analyzing data often relies on ad-hoc decisions, raising concerns about transparency, objectivity, and rigor. This talk introduces a methodology inspired by qualitative codebook construction to systematize prompt engineering. By integrating humans in the loop and a multi-phase verification process, this approach enhances replicability and trustworthiness in using LLMs for data analysis. Practical examples will illustrate how rigorous labeling, deliberation, and documentation can reduce subjectivity and ensure more robust and generalizable research outcomes.

Chirag Shah is Professor in the Information School at the University of Washington, where he conducts research at the intersection of information retrieval, human-computer interaction, and artificial intelligence. His recent work focuses on prompt engineering for optimizing interactions with large language models, exploring both theoretical underpinnings and practical applications. Dr. Shah has authored numerous publications and is actively involved in advancing the understanding of how humans and AI systems can collaborate more effectively.

Generative AI introduces a new computing paradigm, serving as an interface that translates human goals into computational actions. This talk examines three core aspects: generating computation from prompts, optimizing code, and coordinating API integration. A key shift is the emergence of prompt engineering, which enables users to specify complex tasks in natural language, differing from traditional programming. These new abstractions redefine how computation is conceived and executed. Generative AI also reshapes software development, empowering domain experts while computer scientists focus on scalable frameworks. This shift fosters collaboration between human creativity and machine intelligence, expanding computational accessibility.

Jules White, Professor of Computer Science at Vanderbilt University and Senior Advisor to the Chancellor for Generative AI, directs Vanderbilt’s Initiative on the Future of Learning & Generative AI. His research spans cybersecurity, mobile/cloud computing, and AI, with over 170 publications and multiple Best Paper Awards. He is a National Science Foundation CAREER Award recipient. He created one of the first online classes for Prompt Engineering.

La motivation et l’interaction sociale sont deux concepts fondamentaux en psychologie qui sont rarement considérés ensemble. Martin présente des recherches interdisciplinaires visant à concevoir des interactions homme-machine qui motivent les utilisateurs à interagir avec autrui (par exemple, des interactions tangibles et virtuelles pour les utilisateurs autistes) ou soutiennent des interactions destinées à motiver les utilisateurs (par exemple, des technologies mobiles motivationnelles pour adopter de meilleurs modes de vie).

Jean-Claude Martin est professeur à l’Université Paris-Saclay, en France, dans le domaine de l’interaction homme-machine. Il dirige l’équipe de recherche « Cognition, Perception et Usages » au Laboratoire Interdisciplinaire des Sciences du Numérique. Ses recherches portent sur l’adaptation et la combinaison des théories psychologiques avec des approches de conception centrée sur l’utilisateur afin de concevoir des interactions homme-machine pour la formation aux compétences sociales et la motivation à l’activité physique.

Cognitive robotics aims to develop robots capable of human-like learning, interaction, and behavior by grounding abstract concepts in sensorimotor experiences and social interactions. This talk explores how principles like “starting small” and “super-embodiment” can address the limitations of AI tools, such as large language models (LLMs), which rely heavily on large datasets and static learning protocols. By integrating incremental, multimodal learning and redefining embodiment to encompass physical, mental, and social processes, we can enable robots to better understand and utilize abstract concepts. These advancements hold promise for applications in caretaking, education, and beyond, while advancing the intersection of AI, grounded intelligence, and human development.

Angelo Cangelosi, Professor of Machine Learning and Robotics at the University of Manchester and co-directs the Manchester Centre for Robotics and AI. His research focuses on cognitive and developmental robotics, neural networks, language grounding, human-robot interaction, and robot companions for health and social care. He is the author of Developmental Robotics: From Babies to Robots (MIT Press, 2015) and Cognitive Robotics (MIT Press, 2022), co-edited with Minoru Asada and Editor-in-Chief of Interaction Studies and IET Cognitive Computation and Systems.

The ability to continuously learn and accumulate knowledge over a lifetime is a hallmark of human intelligence. However, this essential capability is missing in current machine learning paradigms. This talk explores continual learning in machine learning, with a focus on the challenges of catastrophic forgetting and inter-task class separation. These issues have prevented existing methods from reaching the theoretical upper-bound performance, often with a significant gap. Our recent work demonstrates that achieving this upper bound is indeed possible, offering intriguing insights into both cognition and the foundations of AI.

Bing Liu is a Distinguished Professor and Peter L. and Deborah K. Wexler Professor of Computing at the University of Illinois Chicago. He earned his Ph.D. from the University of Edinburgh. His current research interests include continual or lifelong learning, continual learning dialogue systems, sentiment analysis, machine learning, and natural language processing. He is the author of several books on these topics and has also received multiple Test-of-Time awards for his research papers. He is a Fellow of ACM, AAAI, and IEEE.

GenAI models are computationally complex, but conceptually simple. They are trained to fill in the blanks. Model semantics is limited to word distribution patterns (Harris, 1956), yet many claim that GenAI models are capable of deep cognitive processes (such as reasoning and understanding). These assertions imply that cognitive processes can spontaneously emerge from these behavioral patterns—that a theory of cognition can be constructed at the purely behavioral level of word use patterns. We have seen that movie before, but Chomsky (1959/1967) and a good deal of research with humans and animals, peaking in the 1980s, have demonstrated that a purely behavioral theory of cognition is not viable. Those same research methods could be applied to the analysis of the latest forms of artificial intelligence, but their relevance is rarely recognized. Instead, much of what passes for theoretical analysis of GenAI models is based on the logical fallacy of “affirming the consequent.” The models behave as if they had underlying cognitive processes, but their proponents fail to consider whether other explanations (e.g., stochastically parroting training data) could also explain the observations. Roitblat discusses the internal structure of GenAI models and how to understand them. He also offers some theoretical suggestions for approaching understanding and artificial general intelligence.

Herbert Roitblat, is lead data scientist for Egnyte’s research and development in artificial intelligence. Formerly professor of Psychology, Marine Biology, and Second Language Acquisition, University of Hawaii, Roitblat’s work on how dolphins recognize targets underwater with biosonar led to significant contributions to early neural network research and a patent on a binaural sonar. His more recent work is on artificial intelligence: “Algorithms Are Not Enough: Creating General Artificial Intelligence” (MIT Press, 2020) argues that algorithms and neural network models cannot fully capture the complexity of human cognition or animal intelligence and suggests what is needed to achieve artificial general intelligence.

La recherche en intelligence artificielle développementale puise certaines inspirations dans les théories du comportement animal de von Uexküll, l'épistémologie génétique et la psychologie développementale de Piaget, et l'épistémologie constructiviste de von Glasersfeld. Depuis la première implémentation d'un "schema mechanism" Piagétien par Drescher en 1991, ces recherches continuent à investiguer les questions clés associées : la motivation intrinsèque, la construction de sens, l'autonomie constitutive, le passage à l'échelle en monde ouvert, tout en tentant de les appliquer à la robotique autonome. Cette communication présente notre contribution à ces questions avec la plateforme de robotique libre PetitCat. Le robot PetitCat implémente l'hypothèse de l'ancrage de la connaissance dans des boucles d'interaction moto-sensorielles situées dans le temps et l'espace physique. Nous utilisons "moto-sensoriel" plutôt que "sensorimoteur" pour souligner le rôle clé de l'action, et l'exploitation de certains signaux sensoriels comme des feedbacks pas nécessairement représentationnels--idées déjà présentes chez Piaget et Wiener, puis approfondies par la théorie de l'énaction.

Olivier Georgeon est maître de conférences à l’Université Catholique de Lyon. Ses recherches portent sur l’apprentissage artificiel développemental, notamment les approches constructivistes appliquées à la robotique autonome. Il a développé des algorithmes pour des agents autonomes motivés intrinsèquement et a exploré la construction autonome de la mémoire spatiale et l’inférence énactive dans des environnements tridimensionnels. Georgeon a reçu une formation interdisciplinaire avec un diplôme d’ingénieur en informatique de l’École Centrale de Marseille et un doctorat en psychologie cognitive de l’Université Lumière Lyon.

Humans, like animals, have perceptions, actions, and thoughts that they cannot put into words. The challenge is to understand how humans gained the ability to put so much into words, describing not only what is but also what is not and what could possibly be, and in this way came to construct a diversity of physical and symbolic worlds (Arbib et al., 2023, 2024). Building on modeling brain mechanisms for sensorimotor interaction with the world, we have explored both “how the brain got language” (Arbib, 2012) and what happens “when brains meet buildings” (Arbib, 2021). The little explored relation between these two studies is rooted in the notion of “construction” as used in a physical sense in architecture and in a symbolic sense as the tool for assembling utterances hierarchically. This talk offers hypotheses that address the question of how biological evolution yielded humans with the “construction-ready brains” and bodies that made us capable of the cultural evolution that created the diversity of our mental and physical constructs that we know today. The framework for all this is EvoDevoSocio – the idea that biological evolution yields biological mechanisms for both development and adult function of members of a species, but that social interaction is an important part of that environment, and that in humans cultural evolution has played the crucial role in changing the social, physical and increasingly symbolic and technological environments in which most humans now develop. The bridge between the two forms of construction is provided by the rooting of pantomime in manual action.

Michael A. Arbib, University Professor Emeritus and Professor Emeritus of Computer Science, Biomedical Engineering, Biological Sciences, and Psychology at the University of Southern California (USC). He is currently also an Adjunct Professor of Psychology at UCSD. Arbib’s research bridges neuroscience, computer science, and cognitive science, with a focus on the coordination of perception and action in frogs, rats, monkeys and humans. He applies schema theory and neural network analysis to study brain function, robotics, and machine vision. Known for the Mirror System Hypothesis, he explores language evolution through neural mechanisms for action understanding. More recently, he has explored the neuroscience of the experience of buildings, the design of buildings, and what it might mean for buildings to have “brains.

Les grands modèles de langage (LLM) tels que GPT ou BERT, bien que puissants, souffrent d'une opacité importante qui rend leur fonctionnement interne difficile à comprendre. Leur complexité provient de millions, voire de milliards de paramètres, ce qui limite leur explicabilité et complique l'interprétation de leurs décisions. Cette "boîte noire" pose des défis éthiques et pratiques. Comprendre pourquoi et comment un modèle génère une réponse devient crucial pour favoriser la confiance et permettre une adoption responsable de ces technologies. Des approches comme les techniques d'explicabilité ou les modèles hybrides tentent d'atténuer ces problèmes en combinant puissance et interprétabilité. Comment peut-on répondre au problème d'opacité et d'inexplicabilité des grands modèles de langage en proposant l'injection de connaissances syntaxiques et visuelles, tout en améliorant l'encodage du sens des mots pour les rapprocher de leurs représentations conceptuelles et contextuelles ?

Toufik Mechouma est un professionnel des TI, avec dix ans d'expérience dans le domaine du pétrole et du gaz. Il a débuté sa carrière dans l'apprentissage automatique en 2020 au sein d'une entreprise canadienne, après avoir rejoint le programme d'informatique cognitive de l'UQAM en tant que doctorant en 2019. Il travaille pour le Collège LaSalle en tant que professeur d'IA. Les recherches de Toufik portent sur le traitement automatique du langage naturel d'un point de vue cognitif et computationnel. Il travaille sur l'interprétabilité des LLM et l'encodage du sens des mots à travers des architectures cognitives multimodales.

Lorsque nous lisons ou entendons un mot dont le sens nous échappe, nous pouvons chercher sa définition dans un dictionnaire. Si dans la définition nous lisons un autre mot au sens inconnu, nous pouvons chercher sa propre définition, répétant le processus autant que nécessaire. Cependant, une langue ne peut être apprise en utilisant que la recherche de définitions dans un dictionnaire : le sens de certains mots doit être acquis au préalable. Combien de mots doivent alors être connus pour pouvoir apprendre tous les autres? Trouver le plus petit ensemble de tels mots - un ensemble d’ancrage minimal - est formellement équivalent à l'identification d’un transversal de circuits de cardinalité minimale d’un graphe de dictionnaire. Quels sont les mots retrouvés dans ces ensembles, et comment et quand apprenons-nous leurs sens?

Marcos Lopes Professeur de linguistique, Université de São Paulo, est spécialisé en syntaxe, sémantique et linguistique informatique. Ses recherches portent sur les cadres grammaticaux formels, le traitement du langage naturel et l’interface entre la théorie linguistique et l’IA. Lopes contribue activement à faire progresser notre compréhension de la structure du langage et de l'intégration du langage machine.

Nicolas Goulet complète sa maîtrise en informatique à l'UQAM. Ses recherches portent sur l'analyse théorique des graphes des dictionnaires numériques.

L'IA est de plus en plus présente dans nos quotidiens. Si elle force nos cerveaux adultes à s'adapter, elle soulève des questions quant à ses effets potentiels sur les cerveaux des enfants et des adolescents en plein développement. L'enjeu de cette conférence est de présenter les mécanismes du développement neurocognitif de l'enfant et de l'adolescent et d'interroger comment l'IA peut affecter ces mécanismes.

Grégoire Borst est professeur de psychologie du développement et de neurosciences cognitives de l’éducation à l’Université Paris Cité. Ses recherches portent sur les fonctions cognitives de haut niveau telles que la métacognition, la régulation émotionnelle et le contrôle cognitif. Il a reçu le Prix Dagnan-Bouveret de l’Académie des Sciences Morales et Politiques pour ses recherches en psychologie cognitive et éducation.

L’Apprentissage Perceptif Visuel (APV) améliore la sensibilité visuelle grâce à la pratique, mais il se généralise rarement à de nouvelles conditions de test, un phénomène semblable au surapprentissage en apprentissage automatique. Le degré de généralisation de l’APV varie selon les tâches, limitant son application pratique en rééducation visuelle et formation d’experts (radiologues, etc.). Je discuterai des théories expliquant cette généralisation et présenterai les recherches de mon laboratoire sur la NeuroIA, un modèle computationnel visant à prédire et améliorer la généralisation de l’APV.

Shahab Bakhtiari est professeur adjoint au département de psychologie de l’Université de Montréal et membre associé de l’Institut d’intelligence artificielle Mila Québec. Ses recherches portent sur les réseaux neuronaux et leur rôle dans l’apprentissage perceptif humain, en explorant les liens entre l’intelligence artificielle et les mécanismes cérébraux. Bakhtiari détient un doctorat en neurosciences de l’Université McGill, avec une expertise en perception visuelle, apprentissage profond et apprentissage de représentation dans les systèmes neuronaux.

Il n’y a pas si longtemps, la communauté scientifique considérait l’ingénierie des prompts comme une forme de “bidouillage”. La véritable science se situait au niveau des modèles. Deux ans après le lancement de ChatGPT, il semble toutefois que les modèles de langage deviennent une commodité, tandis que les prompts se transforment en une nouvelle couche logicielle. Pourtant, la programmation pilotée par des prompts en est encore à ses débuts. Concevoir des logiciels à grande échelle, robustes et maintenables sur la base de prompts est une tâche ardue, même pour des applications tolérantes aux erreurs, comme les chatbots. Cette nouvelle couche logicielle nécessite de toute urgence une approche rigoureuse, semblable à celle que nous avons appliquée au reste de la pile de développement grâce à l’ingénierie logicielle. Dans cette présentation, nous aborderons l’approche interne de Numeno.ai pour une programmation pilotée par des prompts de haute qualité. Nous parlerons de la couche intermédiaire de primitives fonctionnelles basées sur des prompts que nous avons développée, ce qui nous a permis de multiplier par 10 la vitesse de mise en production, de test et de maintenance de notre logiciel. Nous présenterons également certains des outils que nous utilisons quotidiennement et en quoi ils diffèrent de leurs équivalents dans la pile d’ingénierie logicielle traditionnelle.

Philippe Beaudoin est cofondateur de numeno, une plateforme basée à Montréal utilisant l'IA pour aider les utilisateurs à personnaliser leur contenu. Ancien chercheur chez Google DeepMind, il se concentre sur l'utilisation de l'IA pour un engagement plus significatif, donnant aux utilisateurs plus de contrôle sur leur expérience numérique. Son travail à numeno met l'accent sur la curation basée sur des principes éthiques comme la transparence et l'autonomie. Beaudoin s'intéresse également aux implications éthiques de l'IA, cherchant à concevoir des systèmes qui favorisent le bien-être humain plutôt que des algorithmes d'optimisation axés sur le profit.

Neural networks can approximate a wide range of functions given appropriate credit assignment. Learning theories have focused on adjusting a network’s parameters, such as connectivity weights, to minimize error on specific tasks. In the brain, such modifications are facilitated by synaptic plasticity mechanisms, whereas deep networks use direct gradient descent. Some learning processes in the brain do not rely solely on these parametric updates. They leverage activity-based credit assignment mechanisms to achieve fast, input-driven learning, non-parametrically. In AI, large foundation models pre-trained on simple predictive goals show remarkable expressivity in in-context learning, using only activations evoked by inputs. I will present results disentangling the roles of parametric and non-parametric learning. I will explore the expressivity of neural networks learning solely from inputs and examine the impact of parametric meta-training objectives on this expressivity. This research aims to provide tools for better understanding input-driven learning in the brain and in-context learning in large pre-trained AI systems.

Guillaume Lajoie est professeur associé à l’Université de Montréal et chercheur au Mila (Institut québécois d’intelligence artificielle). Ses travaux se situent à l’intersection des neurosciences et de l’apprentissage automatique, en particulier sur la compréhension des dynamiques neuronales et comment les modèles inspirés du cerveau peuvent améliorer les systèmes d’IA.

Les humains donnent du sens au langage en contexte, en s’appuyant sur leur propre compréhension du monde, y compris leur propre modèle de la compréhension du monde de leur interlocuteur. Dans cette présentation, j’examinerai divers risques potentiels qui surviennent lorsque les humains appliquent cette capacité de construction de sens aux interactions avec les interlocuteurs artificiels. Que se passe-t-il dans les conversations où un des interlocuteurs n’a pas accès (ou a un accès extrêmement limité) au sens, et toute la tache interprétative repose sur l’autre interlocuteur? J’explorerai brièvement les implications que cela peut avoir pour la conception des technologies linguistiques.

Emily M. Bender est professeure de linguistique et professeure associée à l’École d’informatique et à l’École d’information de l’Université de Washington. Ses recherches portent sur l’ingénierie grammaticale multilingue, la sémantique computationnelle et les impacts sociétaux des technologies linguistiques. En 2022, elle a été élue Fellow de l’Association américaine pour l’avancement des sciences (AAAS).

Après avoir rappelé ce qu'est la génération de texte, nous présenterons une méthode pour créer un texte en anglais et un autre en français à partir d'une unique source de données. Les étapes cruciales telles que la sélection des informations pertinentes et l'organisation du texte sont effectuées conjointement, tandis que le choix des termes et la structuration des phrases varient selon la langue. Les deux textes transmettront donc le même message. Cette approche s'appuie sur pyrealb, un réalisateur de phrases bilingue anglais-français. Son utilisation sera illustrée par quelques applications, p.ex. bulletins météo ou descriptions de matches sportifs. Nous terminerons en comparant ces résultats avec ceux obtenus avec un "grand modèle de langue".

Guy Lapalme est professeur à l’Université de Montréal, spécialisé en traitement automatique du langage naturel (TALN). Ses recherches portent sur la génération de texte, la traduction automatique et l’interaction homme-machine dans les technologies linguistiques. Il a contribué au développement de systèmes de TALN et est reconnu pour son travail sur les outils facilitant la communication entre humains et ordinateurs.

A new perspective has taken hold in neuroscience – brains are complex networks, and the structure of their connectivity (the connectome) is critical for how they function. Understanding the organization of brain graphs (networks), their multiscale architecture and communities, as well as the time-varying topology of functional connectivity have become key research objectives in connectomics and network neuroscience. In this presentation I will review a few current themes and future directions. These include comparative connectomics across species, structure/function relationships as disclosed in neuroimaging data, the growing use of computational models to map information flow and communication dynamics, and a novel edge-centric approach to track dynamic functional connectivity. I will connect to challenges facing modern AI and identify some key issues along three main dimensions: networks, dynamics, and information.

Olaf Sporns est professeur distingué à l’Université de l’Indiana, connu pour ses travaux en neurosciences, en particulier dans le domaine de la connectomique—l’étude de la connectivité cérébrale. Il a contribué de manière significative à la cartographie du connectome cérébral, en soulignant son rôle dans la cognition et le comportement. Sporns est l’auteur de Networks of the Brain et cofondateur du Human Connectome Project.

Despite major breakthroughs in chatbot and virtual assistant technology, the use of chatbots in education is still insubstantial. In my talk, based on my thesis, I will delve into chatbot technology’s under-explored potential in education: What are educational chatbots’ advantages and limitations? What pedagogical approaches emphasize their capabilities and help overcome their shortcomings? What are the best scenarios for their integration into a digital educational platform? To explore answers to these questions, I will present a prototype of a crowd-learning platform with integrated peer-chatbots that has a potential to contribute to a more successful educational process.

Gulnara Shaydullina vient de soutenir sa thèse de doctorat au Département d’informatique cognitive de l’UQAM. Elle est diplômée de l’Université d’État de Bachkirie (aujourd’hui l’Université des sciences et technologies d’Oufa) avec des diplômes d’études supérieures en traduction, en enseignement de l’anglais langue seconde et en philologie et de l’Université McGill avec un diplôme d’études supérieures en traduction. Elle a obtenu sa maîtrise à l’UdeM, où sa thèse était une étude de cas sur le crowdsourcing en traduction. Ses intérêts actuels comprennent les méthodes d’enseignement innovantes, l’apprentissage collaboratif avec les chatbots et les nouvelles technologies en éducation.

Large language models (LLMs) excel at few-shot in-context learning (ICL): learning from a few examples provided in context at inference, without any weight updates. Newly expanded context windows allow us to investigate ICL with hundreds or thousands of examples: many-shot learning. Going from few-shot to many-shot yields significant performance gains across a wide variety of generative and discriminative tasks. Many-shot ICL can be bottlenecked by the available number of human-generated examples. To mitigate this, we explore Reinforced and Unsupervised ICL. (1) Reinforced ICL uses model-generated chain-of-thought rationales in place of human examples. (2) Unsupervised ICL removes rationales from the prompt altogether and prompts the model only with domain-specific questions. Both Reinforced and Unsupervised ICL can be quite effective in many-shot regime learning, particularly on complex reasoning tasks. Unlike few-shot learning, many-shot learning is effective at overriding pretraining biases; it can learn high-dimensional functions with numerical inputs, and it performs comparably to fine-tuning. Our analysis also reveals the limitations of next-token prediction loss as an indicator of downstream ICL performance.

Rishabh Agarwal est chercheur principal chez Google DeepMind à Montréal et professeur adjoint à McGill. Après une année avec Geoffrey Hinton à Google Brain, il a fait son doctorat à Mila sous la direction d’Aaron Courville et Marc Bellemare. Ses recherches portent sur l’apprentissage par renforcement profond, avec des applications aux problèmes réels.

Piek Vossen is Professor of Computational Lexicology at the Vrije Universiteit Amsterdam, where he directs the Computational Linguistics and Text Mining Lab. His research focuses on modeling understanding of language by machines. Within the Hybrid Intelligence program, he currently investigates how human and AI memories can be aligned through communication and their differences can be leveraged for collaborative tasks.

Matt Fredrikson’s research aims to enable systems that make secure, fair, and reliable use of machine learning. His group group focuses on finding ways to understand the unique risks and vulnerabilities that arise from learned components, and on developing methods to mitigate them, often with provable guarantees.

Pierre-Yves Oudeyer and his team at INRIA Bordeaux study open lifelong learning and the self-organization of behavioral, cognitive and language structures, at the frontiers of AI and cognitive science. In the field of developmental AI, we use machines as tools to better understand how children learn, and to study how machines could learn autonomously as children do and could integrate into human cultures. They study models of curiosity-driven autotelic learning, enabling humans and machines to set their own goals and self-organize their learning program. They also work on applications in education and assisted scientific discovery, using AI techniques to serve humans, and encourage learning, curiosity, exploration and creativity.

Andy Lücking is Privatdozent at Goethe University Frankfurt. His work contributes to theoretical linguistics and computational semantics, in particular to a linguistic theory of human communication, that is, face-to-face interaction within and beyond single sentences. Besides publishing on deixis and iconicity in manual gesture, Andy is the main author of Referential Transparency Theory, the current semantic theory of plurality and quantification. His work on the perception of iconic gestures received an IEEE best paper award.

Alessandro Lenci is Professor of linguistics and director of the Computational Linguistics Laboratory (CoLing Lab), University of Pisa. His main research interests are computational linguistics, natural language processing, semantics and cognitive science.

Gary Lupyan is Professor of Psychology at the University of Wisconsin-Madison. His work has focused on how natural language scaffolds and augments human cognition, and attempts to answer the question of what the human mind would be like without language. He also studies the evolution of language, and the ways that language adapts to the needs of its learners and users.

Robert Goldstone is a Distinguished Professor in the Department of Psychological and Brain Sciences and Program in Cognitive Science at Indiana University. His research interests include concept learning and representation, perceptual learning, educational applications of cognitive science, and collective behavior.

Raphael Millière is Lecturer in Philosophy of Artificial Intelligence at Macquarie University in Sydney, Australia. His interests are in the philosophy of artificial intelligence, cognitive science, and mind, particularly in understanding artificial neural networks based on deep learning architectures such as Large Language Models. He has investigated syntactic knowledge, semantic competence, compositionality, variable binding, and grounding.

Ben Goertzel is a cross-disciplinary scientist, entrepreneur and author. He leads the SingularityNET Foundation, the OpenCog Foundation, and the AGI Society which runs the annual Artificial General Intelligence conference. His research work encompasses multiple areas including artificial general intelligence, natural language processing, cognitive science, machine learning, computational finance, bioinformatics, virtual worlds, gaming, parapsychology, theoretical physics and more. He has published 25+ scientific books, ~150 technical papers, and numerous journalistic articles, and given talks at a vast number of events of all sorts around the globe.

Stevan Harnad is Professor of psychology and cognitive science at UQAM. His research is on category-learning, symbol-grounding, language-evolution, and Turing-Testing.

Jake Hanson is a Senior Data Scientist at a financial tech company in Salt Lake City, Utah. His doctoral research in Astrophysics from Arizona State University focused on the origin of life via the relationship between information processing and fundamental physics. He demonstrated that there were multiple foundational issues with IIT, ranging from poorly defined mathematics to problems with experimental falsifiability and pseudoscientific handling of core ideas.

Christoph Durt is a philosophical and interdisciplinary researcher at Heidelberg university. He investigates the human mind and its relation to technology, especially AI. Going beyond the usual side-to-side comparison of artificial and human intelligence, he studies the multidimensional interplay between the two. This involves the study of human experience and language, as well as the relation between them.

Anders Sogaard is University Professor of Computer Science and Philosophy and leads the newly established Center for Philosophy of Artificial Intelligence at the University of Copenhagen. Known primarily for work on multilingual NLP, multi-task learning, and using cognitive and behavioral data to bias NLP models, Søgaard is an ERC Starting Grant and Google Focused Research Award recipient and the author of Semi-Supervised Learning and Domain Adaptation for NLP (2013), Cross-Lingual Word Embeddings (2019), and Explainable Natural Language Processing (2021).

Usef Faghihi est professeur adjoint à l'Université du Québec à Trois-Rivières. Auparavant, Usef était professeur à l'Université d'Indianapolis aux États-Unis. Usef a obtenu son doctorat en Informatique Cognitive à l'UQAM. Il est ensuite allé à Memphis, aux États-Unis, pour effectuer un post-doctorat avec le professeur Stan Franklin, l'un des pionniers de l'intelligence artificielle. Ses intérêts de recherche sont les architectures cognitives et leur intégration avec les algorithmes d'apprentissage profond.

Casey Kennington is associate professor in the Department of Computer Science at Boise State University where he does research on spoken dialogue systems on embodied platforms. His long-term research goal is to understand what it means for humans to understand, represent, and produce language. His National Science Foundation CAREER award focuses on enriching small language models with multimodal information such as vision and emotion for interactive learning on robotic platforms. Kennington obtained his PhD in Linguistics from Bielefeld University, Germany.

Eric Schulz, Max-Planck Research Group Leader, Tuebingen University works on the building blocks of intelligence using a mixture of computational, cognitive, and neuroscientific methods. He has worked with Maarten Speekenbrink on generalization as function learning and Sam Gershman and Josh Tenenbaum.

Dor Abrahamson is Professor in the Graduate School of Education at the University of California Berkeley, where he established the Embodied Design Research Laboratory devoted to pedagogical technologies for teaching and learning mathematics. He is particularly interested in relations between learning to move in new ways and learning mathematicaal concepts. His research draws on embodied cognition, dynamic systems theory, and sociocultural theory.

Melanie Mitchell is Professor at the Santa Fe Institute. Her current research focuses on conceptual abstraction and analogy-making in artificial intelligence systems. Melanie is the author or editor of six books and numerous scholarly papers in the fields of artificial intelligence, cognitive science, and complex systems. Her 2009 book Complexity: A Guided Tour (Oxford University Press) won the 2010 Phi Beta Kappa Science Book Award, and her 2019 book Artificial Intelligence: A Guide for Thinking Humans (Farrar, Straus, and Giroux) is a finalist for the 2023 Cosmos Prize for Scientific Writing.

Paul S. Rosenbloom is a Professor Emeritus of Computer Science in the Viterbi School of Engineering at the University of Southern California (USC). His research has focused on cognitive architectures (models of the fixed structures and processes that together yield a mind), such as Soar and Sigma; the Common Model of Cognition (a partial consensus about the structure of a human-like mind); dichotomic maps (structuring the space of technologies underlying AI and cognitive science); “essential” definitions of key concepts in AI and cognitive science (such as intelligence, theories, symbols, and architectures); and the relational model of computing as a great scientific domain (akin to the physical, life and social sciences).

Ellie Pavlick is an Assistant Professor of Computer Science at Brown University. She received her PhD from University of Pennsylvania in 2017, where her focus was on paraphrasing and lexical semantics. Ellie’s research is on cognitively-inspired approaches to language acquisition, focusing on grounded language learning and on the emergence of structure (or lack thereof) in neural language models. Ellie leads the language understanding and representation (LUNAR) lab, which collaborates with Brown’s Robotics and Visual Computing labs and with the Department of Cognitive, Linguistic, and Psychological Sciences.

David Chalmers is University Professor of Philosophy and Neural Science and co-director of the Center for Mind, Brain, and Consciousness at New York University. He is the author of The Conscious Mind (1996), Constructing The World (2010), and Reality+: Virtual Worlds and the Problems of Philosophy (2022). He is known for formulating the “hard problem” of consciousness, and (with Andy Clark) for the idea of the “extended mind,” according to which the tools we use can become parts of our minds.

Dimitri Coelho Mollo is Assistant Professor with focus in Philosophy of Artificial Intelligence at the Department of Historical, Philosophical and Religious Studies, at Umeå University, Sweden, and focus area coordinator at TAIGA (Centre for Transdisciplinary AI), for the area 'Understanding and Explaining Artificial Intelligence'. He is also an external Principal Investigator at the Science of Intelligence Cluster, in Berlin, Germany. His research focuses on foundational and epistemic questions within artificial intelligence and cognitive science, looking for ways to improve our understanding of mind, cognition, and intelligence in biological and artificial systems. His work often intersects issues in Ethics of Artificial Intelligence, Philosophy of Computing, and Philosophy of Biology.

Ben Bergen is Professor of Cognitive Science at UC San Diego, where he directs the Language and Cognition Lab. His research focuses on language processing and production with a special interest in meaning. He’s also the author of 'Louder than Words: The New Science of How the Mind Makes Meaning' and 'What the F: What Swearing Reveals about Our Language, Our Brains, and Ourselves.’