The chair team intends to study human-AI interactions with the optimization of human-technology teaming as ultimate goal. Indeed, the development of artificial intelligence-based technology (e.g. automated cars, aircraft, virtual assistants) are becoming increasingly present in a wide variety of operational contexts and in everyday life situations.
Most scientific and technical efforts have focused on the implementation of AI and smart sensors. However, these developments are generally achieved without questioning the integration of the human in the control/decision loop.
Safety analyses in many critical domains (e.g. aviation, nuclear power plant, high frequency trading) highlight that human-artificial agent interactions breakdown represents one major contributive factor to recent industrial disasters. A promising avenue to deal with these issues is to consider that humans and artificial agents have complementary skills/abilities and are likely to provide better performance when joined efficiently than when used separately.
A first step to enhance human-AI interactions is to design monitoring technology to “sense” the cognitive state of the human (e.g. degraded attentional abilities). To that end, we will implement passive Brain Computer Interfaces (pBCIs) dedicated to monitoring brain activity while performing complex real-life tasks. We will especially focus on the use of advanced artificial intelligence techniques dedicated to better understand brain dynamics and measure one and/or multiple users’ brains while interacting with each other and with artificial agents. The outputs of this mental state estimator will feed a decision system to close the loop.
We will then develop a decisional unit that considers uncertainties on actions, partially observable states (e.g. states of the humans and states/failures of the artificial agent) or potentially non-deterministic human behavior. Such decision-making system will be governed by a policy to adapt and to enhance human-machine teaming. One originality of this Chair will be to examine such human-technology interaction with (i) multiple agents and (ii) in increasingly naturalistic settings representative of work and in everyday life situations.
Programme : Collaborative AI
Themes: : Neuroscience et IA, Robotique et IA
Porteur : Frédéric Dehais, PR ISAE-Supaero