Home

Pierre-Yves Oudeyer

I am research director (DR1) at Inria Center of University of Bordeaux, heading the Flowers team (see PhD students). I was also a research visitor at Microsoft Research Montreal (2021-22), and permanent researcher in Sony Computer Science Laboratory for 8 years (1999-2007). 

Together with a great team, I study lifelong autonomous learning, and the self-organization of behavioural, cognitive and language structures, at the frontiers of artificial intelligence and cognitive sciences. I use machines as tools to understand better how children learn and develop, and I study how one can build machines that learn autonomously like children, as well as integrate within human cultures, within the new field of developmental artificial intelligence.

We work on applications in education, aiming to use AI techniques as tools in the service of humans, fostering learning, curiosity and creativity (e.g. our algorithms are used in Adaptiv’Maths).  We also do outreach for democratizing access and understanding of AI, in diverse contexts ranging from schools to artistic projects. These projects led to several spin-off from the team, e.g. start-ups or NGOs.

For both fundamental and applied projects, we take an open science and open-source approach (e.g. see our team’s GitHub, and the Poppy open-source educational robotic kits with the associated community, used by >30k children in world in several countries).

We collaborate with many academic labs (e.g. Sorbonne-Université, MILA, Inserm) and industry (e.g. HuggingFace, Microsoft Research, evidenceB, Ubisoft, Poïetis, OnePoint).

Important note: If you wish to contact me for an application, please send an email including [application] in the object. For any other reason, please include [information] in the object. As I receive many emails, I may not be able to see your email if you do not include one of these tags.

Quick links:

In more details:

In AI, we study how machines can efficiently acquire world models and open-ended repertoires of skills over an extended time span. We develop autotelic deep reinforcement learning agents able to learn to represent and generate their own goals, leveraging language and culture as cognitive tools for creative exploration, planning and abstraction (see our recent perspective paper discussing new opportunities at the crossroads of Deep RL and large language models, where llms are seen as culture models). This research also involves automatic curriculum learning.

With colleagues, I develop computational theories of curiosity and intrinsically motivated learning (with AI, neuroscience and developmental psychology perspectives). We proposed the learning progress hypothesis to explain key aspects of human spontaneous exploration. We showed how curiosity-driven exploration can self-organize long term developmental trajectories, accouting for how infants progressively develop vocal skills, tool use and language.

I also work on theoretical models of the origins and evolution of speech and language, studying the role of self-organization in neural networks and agents dynamical coupling. In the new edition of my book “Self-organization in the evolution of speech” (to appear in 2020 at OUP, CC-BY), I present an integrated view of the roles of self-organization and intrinsic motivation in the origins of language.

Educational applications. We use curiosity-driven learning algorithms to personalize sequences of exercises for human learners, maximising learning efficiency and motivation: after an initial series of experiments with >1000 children in primary schools, we are now working with a consortium of edTech companies and the support of French ministry of Education to integrate this approach in an educational software for wide dissemination. We developed open-source educational robotics kits (some used by dozens of thousands of children in the world and adapted in largely disseminated educational books e.g. from Main à la Pâte), now disseminated by a non-governmental organization and a start-up.

AI for science: Recently, we started exploring the new area of automated discovery of self-organized patterns in complex systems, leveraging intrinsically motivated goal exploration and unsupervised representation learning.

Robotics applications. In robotics, our curiosity algorithms have been used within the Sony Aibo and Qrio humanoid entertainment robots. I also worked on emotional speech synthesis technologies used in some Playstation video games, and various forms of adaptive human-computer interfaces.

AI and the arts: We collaborate with artists through diverse projects aiming to explore the links between AI and society (e.g. here, here or here), in particular to facilitate understanding of AI and its societal dimensions in the general public.

Selected publications:

Carta, T., Romac, C., Wolf, T., Lamprier, S., Sigaud, O., & Oudeyer, P. Y. (2023). Grounding large language models in interactive environments with online reinforcement learning. ICML 2023.
Abdelghani, R., Wang, Y. H., Yuan, X., Wang, T., Lucas, P., Sauzéon, H., & Oudeyer, P. Y. (2024). GPT-3-driven pedagogical agents to train children’s curious question-asking skills. International Journal of Artificial Intelligence in Education, 34(2), 483-518.
Forestier, S., Portelas, R., Mollard, Y., & Oudeyer, P. Y. (2022). Intrinsically motivated goal exploration processes with automatic curriculum learning. Journal of Machine Learning Research (JMLR), 23, 1-41.
Colas, C., Karch, T., Sigaud, O., & Oudeyer, P. Y. (2022). Autotelic agents with intrinsically motivated goal-conditioned reinforcement learning: a short survey. Journal of Artificial Intelligence Research (JAIR), 74, 1159-1199
Ten, A., Kaushik, P., Oudeyer, P. Y., & Gottlieb, J. (2021). Humans monitor learning progress in curiosity-driven exploration. Nature communications, 12(1), 5972.
Colas, C., Karch, T., Lair, N., Dussoux, J. M., Moulin-Frier, C., Dominey, P., & Oudeyer, P. Y. (2020). Language as a Cognitive Tool to Imagine Goals in Curiosity Driven Exploration. Advances in Neural Information Processing Systems (Neurips 2020), 33.
Etcheverry, M., Moulin-Frier, C., & Oudeyer, P. Y. (2020). Hierarchically organized latent modules for exploratory search in morphogenetic systems. Advances in Neural Information Processing Systems (Neurips 2020), 33, 4846-4859
Colas, C., Fournier, P., & Chetouani, M., Sigaud, O., Oudeyer, P. Y., (2019). CURIOUS: Intrinsically Motivated Modular Multi-Goal Reinforcement Learning. In International Conference on Machine Learning (ICML 2019).
Gottlieb, J. and Oudeyer, P-Y. (2018) Towards a Neuroscience of Active Sampling and Curiosity, Nature Reviews Neuroscience, 19(12), 758-770.
Oudeyer, P-Y. and Smith. L. (2016) How Evolution may work through Curiosity-driven Developmental Process , Topics in Cognitive Science, 1-11.
Baranes, A., Oudeyer, P-Y. (2013) Active Learning of Inverse Models with Intrinsically Motivated Goal Exploration in Robots, Robotics and Autonomous Systems, 61(1), pp. 49-73.
Oudeyer, P. Y., & Kaplan, F. (2007). What is intrinsic motivation? A typology of computational approaches. Frontiers in neurorobotics, 1, 108.
Oudeyer, P-Y., Kaplan F., and Hafner V. (2007) Intrinsic motivation systems for autonomous mental development. IEEE transactions on evolutionary computation, 11.2, 265-286.
Oudeyer, P. Y. (2006, updated 2020 version). Self-organization in the evolution of speech (Vol. 6). Oxford University Press.