Maneuvering uncertainty and urgency in working with artificial intelligence

Business Division

Vrije Universiteit Amsterdam

Dr. Mohammad Rezazade Mehrizi (m.rezazademehrizi@vu.nl)

Extended Abstract Among today’s professionals across different domains, there is a shared sentiment that artificial intelligence (AI) will fundamentally transform their work. However, ideas on what the future of ‘working with AI’ will look like, vary starkly even within a given community of professionals. One reason for this is that the claims on the transformative power of AI have been disproportionate to the current stage of technological development. As one result, the term AI has taken on various definitions, making it more confusing for professionals to cope with the possibilities of working with AI. Another reason is that many speculations have been unsubstantiated, detached from the very practitioners whose work is claimed to be disrupted. Whereas so much has been talked about the technology and all the work that will be affected such as telemarketing, journalism and radiology, so little effort has been made to listen to telemarketers, journalists and radiologists who actually know their work. Claims on AI that are detached from the state of development and the professionals have resulted in a confusing situation, which is overly imbued with both uncertainty and urgency. How do professionals maneuver such a confusing situation and envision the future of their work? Tapping into their envisioned future of work could help us understand how professionals renew their societal relevance in an early stage of a possible technological disruption. We take the case of radiologists, who have been identified by the “godfather of deep learning”, Geoffrey Hinton, as those whose work will be taken over in the near future by AI (more specifically, deep learning). This is an early-stage empirical paper of an ongoing study. Using in-depth interview data from over 80 active practitioners in the field of radiology across Europe, we lay out some coping mechanisms of radiologists vis-à-vis AI. To begin with, we see that radiologists cope with the confusing situation by reframing the uncertainty around AI and discounting the urgency to react to AI. By doing this, radiologists reinterpret the situation and make it appear more favorable. Firstly, radiologists reframe uncertainty to appear less threatening, largely in two ways. Radiologists claim (1) that technology-driven changes are nothing new in radiology, juxtaposing AI with earlier technological advancements in the field such as MR, CT and ultrasound; and (2) that AI itself is not new and has been around since as early as the 1950s. Secondly, radiologists discount the sense of urgency in popular discourse by qualifying the claims regarding the impact of AI on their work. This is done by arguing (1) that in an early stage of any technological shift, there is always too much hype; (2) that many of these claims are made by people that have little understanding of the work of radiologists; and (3) that change will not happen overnight but gradually, thus leaving room for radiologists to adapt. In addition to reinterpreting the situation to make it more favorable, radiologists reinterpret their professional role in society. In such a situation, radiologists envision their future of work in starkly different ways, on a rather normative and abstract level. Two distinctly different images of the future radiologist emerge and they are ‘the human radiologist’ and ‘the digitally augmented radiologist’. The human radiologist seeks their value and societal relevance in skills that they believe are characteristic of humans and will remain relatively intact from the computing power of AI. The human radiologist should therefore specialize in communicating and building relationships with patients and other medical doctors, or develop manual dexterity and go into interventional radiology. The relationship between the human radiologist and AI is competition. AI corrodes the jurisdiction of radiologists and radiologists must make a new jurisdictional claim in areas where they can secure comparative advantage over AI. At the opposite end lies the idea of the digitally augmented radiologist who would be able to work faster, cheaper, more accurately and with consistent quality. The digitally augmented radiologist brings values that align with the promises of AI such as efficiency and accuracy. The digitally augmented radiologist should work with AI as a tool, an assistant, or a colleague. The assistant AI, for example, emancipates the digitally augmented radiologist from repetitive, low-risk, boring tasks and allow them to focus on more complex and rewarding tasks. The relationship between the digitally augmented radiologist and AI is collaboration. The jurisdiction of radiology remains unshaken and radiologists get to further legitimize their profession by being more efficient, accurate and consistent. Thus, in the future of the digitally augmented radiologist, ‘what’ radiologists do remains the same and only ‘how’ they do what they do changes, for the better. Keywords: technological disruption, professional legitimacy, radiology