The Future of Human-AI Interaction
Imagine yourself behind a steering wheel of a car, stuck in traffic on your way out of the city for a weekend holiday outdoors. A few metres ahead of you someone is trying to join the traffic from a side road. Will you stop and let them in or push on, hoping that someone else will let them in behind you? Whatever you’re thinking, will you do the same if the other is a driverless car?
Humans will soon have to switch from being mere users of machines (think, for instance, of Google Translate) to being their co-players in strategic social settings with artificial agents increasingly endowed with their own autonomous decision-making capacities. The impact of this on people’s choice behaviour and, in turn, on the desirability of outcomes of human-AI interactions is yet unknown. In this project we investigate whether and how patterns of cooperation and coordination between humans and AI systems will emerge in our day-to-day interactions with one another.
Using the methods of behavioural game theory, we are currently conducting a series of empirical studies to see whether people are as likely to trust, take risks, and cooperate with AI systems as they do with other humans in a variety of social scenarios.
- Jurgis Karpus
- Ophelia Deroy
- Bahador Bahrami
- Adrian Krüger