How does the radiology community discuss the benefits and limitations of artificial intelligence for their work? A systematic discourse analysis

Bomi Kim*, Isabel Koopmanschap, Mohammad H.Rezazade Mehrizi, Marleen Huysman, Erik Ranschaert

*Corresponding author for this work

Research output: Contribution to JournalArticleAcademicpeer-review

Abstract

Purpose: We aimed to systematically analyse how the radiology community discusses the concept of artificial intelligence (AI), perceives its benefits, and reflects on its limitations. Methods: We conducted a qualitative, systematic discourse analysis on 200 social-media posts collected over a period of five months (April–August 2020). Results: The discourse on AI is active, albeit often referring to AI as an umbrella term and lacking precision on the context (e.g. research, clinical) and the temporal focus (e.g. current AI, future AI). The discourse is also somewhat split between optimism and pessimism. The latter considers a wider range of social, ethical and legal factors than the former, which tends to focus on concrete technologies and their functionalities. Conclusions: Further precision in the discourse could lead to more constructive conversations around AI. The split between optimism and pessimism calls for a constant exchange and synthesis between the two perspectives. Practical conversations (e.g. business models) remain rare, but may be crucial for an effective implementation of AI in clinical practice.

Original languageEnglish
Article number109566
Pages (from-to)1-6
Number of pages6
JournalEuropean Journal of Radiology
Volume136
Early online date26 Jan 2021
DOIs
Publication statusPublished - Mar 2021

Funding

Not applicable.

Keywords

  • Artificial intelligence
  • Deep learning
  • Discourse
  • Machine learning
  • Perceived benefits and limitations
  • Social media

Fingerprint

Dive into the research topics of 'How does the radiology community discuss the benefits and limitations of artificial intelligence for their work? A systematic discourse analysis'. Together they form a unique fingerprint.

Cite this