Skip to main content Skip to navigation

Professor Noortje Marres, Leverhulme Research Fellowship

Beyond the Lab: An Empirical Philosophy of Intelligent Vehicle Testing

Intelligent technologies used to be tested in laboratories, and in remote environments such as deserts. Yet today artificial intelligence (AI) testing happens in schools, hospitals, city streets, and other everyday places in society.

The introduction of new technologies into society at an early stage of their development poses challenges for democracy, and for the relations between innovation and society. Back when technologies were developed in-house, in R&D laboratories, they could be assessed for safety and quality before their introduction into society, but this is no longer necessarily the case.

A shift from industrial to experimental innovation

As technology testing increasingly takes place “beyond the laboratory”, in environments in society, a fundamental shift may be underway in the relations between innovation and society.

Sociologists of technology have argued that we are moving from a regime of industrial innovation to a new paradigm of experimental innovation. Innovation increasingly involves creating test environments in society, such as testbeds (platforms for conducting replicable tests of scientific theories and new technologies), living labs and on social media platforms.

This requires the development of new frameworks for the regulation and governance of innovation.

Street trials of autonomous vehicles offer a manifestation of AI beyond the lab.

The Greenwich Gateway trial route (Gross, 2016).

Driverless pods test route, Milton Keynes (Marres, 2018).

Intelligent technologies pose special problems in this regard, as they are said to inherently resist oversight. Machine learning systems, such as those built into computer vision systems used by intelligent vehicles, are not only highly complex, they are also dynamic and opaque, rendering their operations inaccessible to scrutiny by experts and non-experts alike.

Technology testing in the street: New problems, forms and sites for accountability

However, paradoxically, technology testing in real-world settings also offers opportunities for examining and perhaps addressing these challenges that AI poses for society and democracy.

Whilst the AI that is used in transport systems is often invisible, street tests of intelligent vehicles offer a prominent manifestation of intelligent technology beyond the lab.

As testing here takes place in publicly accessible environments, such as city squares, it's now possible for everyday people to become more actively involved in the evaluation of innovation.

An example of this occurred when citizen journalists reported on the sighting of autonomous vehicles in Coventry in 2017.

Professor Noortje Marres from the Centre for Interdisciplinary Methodologies is undertaking a Fellowship to investigate these problems and emerging forms of accountability.

She will conduct field research on AI testing in society in three different environments:

  • A regional testbed (West Midlands)

  • A capital city (London)

  • A laboratory town (Milton Keynes).

The street as a theatre of accountability

Engineers and computer scientists seem to be increasingly interested in “taking the social world into account” in technology development.

However, accountability is still often framed in a rather technology-centric way, with all the attention going to how machines perform their tasks. To develop a more society-centric approach, this Fellowship will study street trials of intelligent technologies as “theatres of accountability”: choreographed situations where the ability to take into account, to give an account, and to call to account, is shared, contested and negotiated between different road users.

To be accountable means not just performing well. It means being able to adopt the perspectives of others. Taking this into account will require the development of new understandings and models of the interaction between technology and society.


Find out more about Professor Marres' research