by Glen Berman, Kate Williams, and Eliel Cohen
Berman, G., Williams, K., & Cohen, E. (2025). The benefits of being between (many) fields: Mapping the high-dimensional space of AI research. Big Data & Society, 12(1). https://doi.org/10.1177/20539517241306355
Artificial Intelligence is being mobilized across the university sector to coordinate a wide range of research programs and activities. AI is ‘in’, and nobody wants to be left out. But this begs the question: what is it about AI—as a discursive concept—that makes it so adaptable? What is it about the notion of AI that enables such a heterogenous array of actors—computer scientists, climate scientists, roboticists, ethicists, lawyers, artists, and more—to enroll it in their efforts to attract resources, coordinate research projects, and communicate research impacts? And, across this heterogenous network, what, if anything, lends the field of actors associated with AI coherence and stability?
In our paper, we begin to answer these questions. Through semi-structured interviews (n = 90) with academics affiliated with AI research networks in the UK, US, and Australia, we explore how notions of the ‘AI researcher’ and the ‘AI research field’ are constructed. Through our empirical account, we describe an AI research field that is uncertain and unstable. The field emerges at the intersection of multiple overlapping boundaries between disciplines, between academia, industry and government, between national and international hierarchies. And, the field is marked by commitments to highly applied, interdisciplinary and intersectoral research activities. Within this context, notions of AI are strategically mobilized by AI researchers and AI research networks to position themselves as intermediators at these intersections. In this light, we argue, that the definitional messiness associated with AI is a strategic commitment of the field—the fact that there are no shared or bounded definitions of what constitutes AI helps enable AI researchers to move between disciplines and sectors.
Yet, not anyone can claim to be an AI researcher. To be legitimized in a high dimensional field requires cultivation of an expertise that can be readily translated into the evaluation structures of many different fields. AI researchers’ commitment to applied research can be interpreted as one strategy for achieving this. Applied research can meet the demands of government funders and industry partners, can translate into news media reporting, and can be published in parallel academic disciplines. And, in the absence of formal qualifications or training programs, legitimization relies on affiliation with organizations or institutions whose value readily translates across fields. For individual researchers and for university administrations, then, establishing new, explicitly intersectoral and interdisciplinary AI-branded research networks is a self-reinforcing response to the high dimensionality of the AI research field.
