Research in the global field of artificial intelligence is increasingly hybrid in orientation. Researchers are beholden to the requirements of multiple intersecting spheres, such as scholarly, public, and commercial, each with their own language and logic. Relatedly, collaboration across disciplinary, sector and national borders is increasingly expected, or required. Using a dataset of 93,482 artificial intelligence publications, this article operationalises scholarly, public, and commercial spheres through citations, news mentions, and patent mentions, respectively. High performing publications (99th percentile) for each metric were separated into eight categories of influence. These comprised four blended categories of influence (news, patents and citations; news and patents; news and citations; patents and citations) and three single categories of influence (citations; news; patents), in addition to the ‘Other’ category of non-high performing publications. The article develops and applies two components of a new hybridity lens: evaluative hybridity and generative hybridity. Using multinomial logistic regression, selected aspects of knowledge production – research context, focus, artefacts, and collaborative configurations – were examined. The results elucidate key characteristics of knowledge production in the artificial intelligence field and demonstrate the utility of the proposed lens.

The Artificial Intelligence research field sits at the intersection of several overlapping spheres (academia, industry, media), each with their own logics and commitments. The influence of research within these worlds is studied through a number of bibliometric methods, including citation metrics for measuring influence within academia, and counts of patents and news-media mentions for influence in industry and the media. Using a large-scale, publicly-available dataset of research outputs, we compare the topical content of outputs that are highly influential in each of these worlds. We identify significant differences between the content of influential research in these worlds, indicating that the academic, industry and media worlds value different aspects of the Artificial Intelligence field. These differences provide new insights on the evaluation of research produced within the Artificial Intelligence field.

<p>Complex systems incorporating interconnected social, ecological, and technological components are often the subject of analysis and intervention. Such systems frequently give rise to wicked problems [1-4] – problems that “prove to be highly resistant to resolution through any of the currently existing modes of problem-solving” [2-3]. Such problems require a transdisciplinary approach – one where multiple perspectives and realities can inform decisions for intervening in the system. This is well understood in many policy-relevant fields. Inquiry into climate change impacts or water policy, for example, can only proceed effectively with some understanding of the “partiality, plurality and provisionality of knowing” [5]. In working with these types of complex systems, transdisciplinary teams capable of effectively engaging with many worldviews and ways of creating knowledge [2] are increasingly seen as essential for carrying out impactful research-based work. Despite the increasing importance of transdisciplinary practice, educational programs designed to help students effectively carry out this work remain rare, and researchers engaging in this kind of practice often must navigate institutional structures designed to reinforce rather than permeate boundaries between disciplines.</p><p>The School of Cybernetics at the Australian National University is one of an increasing number of institutions where transdisciplinary practice is a norm rather than an exception. Staff have been recruited from diverse scholarly and professional backgrounds and career trajectories, and activities encourage engagement in transdisciplinary inquiry. This is in service of the central mission of the School: to identify and develop the knowledges and practices required to take AI-enabled cyber-physical systems safely, responsibly and sustainably to scale in the world.  Experimental transdisciplinary masters and PhD programs have been convened since 2019 to help achieve this mission.</p><p>In this presentation, we will draw from the authors’ collective experience as supervisors, instructors, and students in these and other programs to provide guidance on designing and delivering effective transdisciplinary educational programs for higher degree research students. We will address the following aspects of postgraduate education: the student selection process, in which careful cohort selection is essential for identifying students likely to effectively engage in transdisciplinary work; our experience using formal and informal hands-on training in a range of research and relationship-management skills to support transdisciplinary practice; institutional structures and scaffolding to support transdisciplinary cultures and incentives; and ways of supporting students and supervisors to thrive through the creation of diverse and respectful research communities of practice that support collective learning.</p><p>[1] C. Andersson and P. Törnberg, Futures, 95, 118–138 (2018).</p><p>[2] V.A. Brown, Collective Inquiry and Its Wicked Problems, in <em>Tackling Wicked Problems: Through the Transdisciplinary Imagination</em>, edited by J.A. Harris <em>et al</em>., (Earthscan, New York, Oxon, 2010) pp. 61-83.</p><p>[3] H. Rittel and M. Webber, Policy Sciences, 4(2), 155-169 (1973).</p><p>[4] J. Mingers, and J. Rosenhead, <em>Rational analysis for a problematic world revisited</em>, Vol. 1 (John Wiley and Sons Ltd. Chichester UK, 2001).</p><p>[5] J.Y. Russell, A Philosophical Framework for an Open and Critical Transdisciplinary Inquiry, in <em>Tackling Wicked Problems: Through the Transdisciplinary Imagination</em>, edited by J.A. Harris <em>et al</em>., (Earthscan, New York, Oxon, 2010) pp. 31-60.</p>

No abstract