AI as super-controversy

Eliciting AI and society controversies with an extended expert community in the UK

Bibliographic Data

ID5260461
AuthorsN Marres (0000-0002-8237-6946, University of Warwick, corresponding author), Michael Castelle (0000-0001-6754-6516, University of Warwick), Beatrice Gobbo (0000-0001-7085-0017, Politecnico di Milano), Cristian Poletti (0009-0003-4402-1137, University of Warwick), James Tripp (0000-0003-2471-3411, University of Warwick)
Year2024
Volume11
Issue2
Publication date2024-06-01
Peer ReviewedYes
Open AccessYes
TypeARTICLE
VenueBig Data & Society (JOURNAL)
Journal identifiersISSN: 2053-9517 • E-ISSN: 2053-9517
PublisherSAGE Publications Inc (PUBLISHER)
DOI10.1177/20539517241255103
OpenAlexW4399358739
LanguageEN
Citations received23
References cited30

Following the release of large language models in the late 2010s, the backers of this new type of artificial intelligence (AI) publicly affirmed that the technology is controversial and harmful to society. This situation sets contemporary AI apart from 20th-century controversies about technnoscience, such as nuclear power and genetically modified (GM) foods, and disrupts established assumptions concerning public controversies as occasions for technological democracy. In particular, it challenges the idea that such controversies enable inclusion and collective processes of problem definition ('problematisation') across societal domains. In this paper, we show how social research can contribute to addressing this challenge of AI controversies by adopting a distinctive methodology of controversy analysis: controversy elicitation. This approach actively selects, qualifies and evaluates controversies in terms of their capacity to problematise AI across the science and non-science binary. We describe our implementation of this approach in a participatory study of recent AI controversies, conducted through consultation with UK experts in AI and society. Combining an online questionnaire, social media analysis and a participatory workshop, our study suggests that civil society actors have developed distinctive strategies of problematisation that counter the strategic affirmation of AI's controversiality by its proponents and which centre on the public mobilisation of AI-related incidents: demonstrations of bias, accidents and walkouts. Crucially, this emphasis on 'AI frictions' does not result in the fragmentation of AI controversies, but rather enables the articulation of AI as a 'super-controversy': the explication of connections between technical propositions, situated troubles and structural problems in society (discrimination, inequalities and corporate power

Data science · Engineering ethics · Environmental ethics · Epistemology · Political science · Sociology · Computational and Text Analysis Methods · Computer Science · Engineering · Ethics and Social Impacts of AI · Law, AI, and Intellectual Property · Philosophy

Tools

Open DOIOpen Access