About this project
In progress 2023 - 2027
Counterfactual thinking corresponds to the ability to (mentally) simulate alternate scenarios, e.g., for introspecting about the past or for anticipating the future. Humans are adept at such commonsense counterfactual thought in their everyday perception and interaction. This research centralizes (human-like) counterfactual commonsense reasoning as a foundation for the design and engineering of next-generation (human-centred) AI methods. Key questions pursued in the project focus on the theoretical, methodological, and robust algorithmic foundations of computational counterfactual commonsense with a specific emphasis on commonsense reasoning about space-time dynamics. The project also emphasise application and evaluation of methods developed from the standpoints of foundational generalisation, and real-world usability in domains such as self-driving vehicles, assistive robots.
More broadly, through a confluence of (analytical and empirical) research methods in artificial and human intelligence, the project advances an interdisciplinary methodology and mindset -encompassing Cognition, AI, Interaction, and Design- for the development of next-generation human-centred AI and computational cognitive systems.