Responsible Intelligent agents & Intelligent human/agent-agent interaction
Our mission is to understand and engineer collaboration between humans & agents. We develop theories, models and methods of interactive intelligence by combining methods from AI and the social sciences.
About the group
The Interactive Intelligence (II) section focusses on socially interactive, intelligent agents. We research the intelligence that underlies and co-evolves during the repeated interactions of human and technology “agents” who cooperate to achieve a joint goal. Our research program aims for synergy and social interaction between humans and technology, to empower humans in their social context. The new technological challenges we face arise from the need to integrate Artificial Intelligence, Cognitive Engineering, and behavioural sciences. In particular, the technological challenge is to develop socially aware agents that in interaction with humans co-adapt and co-learn over time. Social awareness implies context-awareness with the knowledge to interpret the physical situation in social terms and the knowledge to behave in a distinctive individual way that is personalized towards those the agent interacts with. In this manner, we endeavour to develop the interactive agent technology that empowers humans and groups of humans to deal with the societal and individual challenges such as the increasing need for sustained self-management for healthy ageing, safety and life-long education.
Crosscutting aspects:
Behaviour modelling, Mental models/ToM, Responsible AI/EXAI, Intelligent agents, Multi-modal perception systems, Knowledge representation x ML, Socio-cognitive engineering, Humane/Hybrid AI, Long-term interaction, Situated awareness.
Embedding
We are part of the Department of Intelligent Systems.
We are participating in:
01 november 2022
Frans Oliehoek new Director ELLIS Delft Unit
After almost three years of guidance by Robert Babuska, Frans Oliehoek is taking over the lead of the ELLIS Delft Unit. He will be continued to be advised by Robert, as well as Jens Kober. We thank Robert for all his work and efforts.
04 oktober 2022
Drivers of partially automated vehicles are blamed for crashes that they cannot reasonably avoid
People seem to hold the human driver to be primarily responsible when their partially automated vehicle crashes. But is this reasonable? Researchers Niek Beckers, Luciano Cavalcante Siebert, Merijn Bruijnes, Catholijn Jonker & David Abbink from the AiTech initiative investigated the apparent mismatch between the public’s attribution of blame and finding from the human factors literature regarding human’s ability to remain vigilant in partially automated driving.
20 september 2022
Adding social awareness to conversational agents
Catharine Oertel from the Interactive Intelligence Group develops conversational agents that are socially aware, allowing them to engage with people in a much more human-like manner, even going as far as holding their own in a group setting.
20 september 2022
Keynote lecture at IC3K 2022
Catholijn Jonker will be giving a keynote lecture at the 14th International Joint Conference on Knowledge Discovery in October 2022.
05 juli 2022
Interactive Intelligence Group presents two papers at ICML'22
We will be presenting 2 papers at ICML’22, one of the top-tier international conferences in machine learning.
27 juni 2022
Waar blijft de planologie van AI?
Files, opstoppingen op sluiproutes waar de omgeving niet op is bedacht en waar omwonenden niet op zijn voorbereid, zijn aan de orde van de dag. Toch ligt de oorzaak niet per se bij het autogebruik dat in de afgelopen decennia is gestegen, maar eerder door routeplanners. Hoe veroorzaken zulke appjes dat? Dat is de schuld van onze insteek en mentaliteit als informatici. Sietze Kuilman van de Interactive Intelligence Group schreef een stuk voor Computable.
27 juni 2022
Best paper and poster at HHAI
Our team led by Michiel van der Meer and Enrico Liscio won best paper award and best poster award for their works. Michiel presented HyEnA, a hybrid method for extracting arguments from opinions. Enrico presented Axies, a hybrid methodology for identifying context-specific values.