AI in Motion: Studying the Social World of Autonomous Vehicles

As autonomous and semi-autonomous vehicles have grown in sophistication and ability, they have been deployed on the road in increasing numbers. While these technologies have driven millions of miles, they have also arrived in a partially broken form – and involved in a number of fatal accidents.

Recent research has documented the ‘sociality of traffic’ – the taken for granted unseen order that makes public space effective and safe. If our public spaces are increasingly going to become a mix of ‘AI’ and ‘human’ vehicles, what will happen to this ‘social order’ of traffic? Will the burden of making autonomous vehicles work fall onto those who do not have access to these vehicles – pedestrians, cyclists and other road users? 

This project will document how humans change their expectations and behaviours when meeting AI-controlled vehicles, but also how those ongoing interactions affect other road users. We will collect video recordings in a range of different contexts – from existing semi- autonomous systems on the road, experimental test vehicles and current deployments on Swedish roads.

More broadly, autonomous vehicles, and our interactions with them, present a new model of thinking about AIs as systems in ‘continual interaction’ with humans. In our conceptual work we will engage with recent discussions of interactive machine learning, exploring real-time continual dialogue, with adjustments of both human and system. This has interesting implications for concepts such as the transparency and explicability of AI. Our goal is to broadly think through AI in-action in dialogical ways.