Webbläsaren som du använder stöds inte av denna webbplats. Alla versioner av Internet Explorer stöds inte längre, av oss eller Microsoft (läs mer här: * https://www.microsoft.com/en-us/microsoft-365/windows/end-of-ie-support).

Var god och använd en modern webbläsare för att ta del av denna webbplats, som t.ex. nyaste versioner av Edge, Chrome, Firefox eller Safari osv.

Foto av Charlotte Högberg

Charlotte Högberg

Doktorand

Foto av Charlotte Högberg

Four Facets of AI Transparency

Författare

  • Stefan Larsson
  • Kashyap Haresamudram
  • Charlotte Högberg
  • Yucong Lao
  • Axel Nyström
  • Kasia Söderlund
  • Fredrik Heintz

Redaktör

  • Simon Lindgren

Summary, in English

Transparency in artificial intelligence (AI) can mean many things, but at the same time, it is currently a central focus for both scientific and regulatory attention. We seek to critically unpack this conceptual vagueness. This is particularly called for given recent focus on transparency in much of AI policy. To this end, we construct our analysis of AI Transparency into four facets. Firstly, (1) explainability (XAI) has become an expanding field in AI, which we argue needs to be complemented by more explicit focus on the (2) mediation of AI-systems functionality, as a communicated artefact. Furthermore, in the policy discourse on AI, the importance of (3) literacy is underscored. We draw from the rich literacy literature in order to show both promising and troubling consequences of this. Lastly, we unpack transparency as a form of governance, within a (4) legal framework encompassing a structure of trade-offs. By these four facets we aim to bring more clarity to the multifaceted concept of transparency in AI.

Avdelning/ar

  • Institutionen för teknik och samhälle
  • AI och samhälle
  • Fastighetsvetenskap
  • EPI@LUND
  • Avdelningen för arbets- och miljömedicin

Publiceringsår

2023-11-27

Språk

Engelska

Sidor

445-455

Publikation/Tidskrift/Serie

Handbook of Critical Studies of Artificial Intelligence

Dokumenttyp

Del av eller Kapitel i bok

Förlag

Edward Elgar Publishing

Ämne

  • Other Humanities not elsewhere specified
  • Social Sciences Interdisciplinary
  • Other Engineering and Technologies not elsewhere specified

Nyckelord

  • AI Transparency
  • explainability
  • xAI
  • AI
  • machine learning
  • literacy
  • AI literacy
  • explainable AI
  • law as tradeoff
  • mediation

Status

Published

Projekt

  • Perceptions of AI and robotics in Sweden and Japan
  • AI-förordningen - mellan fixering och flexibilitet
  • The Automated Administration: Governance of ADM in the public sector
  • AI Transparency and Consumer Trust
  • AI in the Name of the Common Good - Relations of data, AI and humans in health and public sector
  • Mammography Screening with Artificial Intelligence
  • AIR Lund - Artificially Intelligent use of Registers
  • Automated decision-making – Nordic perspectives

Forskningsgrupp

  • AI and Society
  • EPI@LUND

ISBN/ISSN/Övrigt

  • ISBN: 9781803928562
  • ISBN: 9781803928555