Accountability in Artificial Intelligence: What It Is and How It Works

Подотчётность в области искусственного интеллекта: что это такое и как она работает
Luciano Floridi, Mariarosaria Taddeo, Claudio Novelli
2022-01-01

AI accountabilityAI governanceaccountability architectureanswerabilitysociotechnical systems
Abstract Accountability is a cornerstone of the governance of artificial intelligence (AI). However, it is often defined too imprecisely because its multifaceted nature and the sociotechnical structure of AI systems imply a variety of values, practices, and measures to which accountability in AI can refer. We address this lack of clarity by defining accountability in terms of answerability, identifying three conditions of possibility (authority recognition, interrogation, and limitation of power), and an architecture of seven features (context, range, agent, forum, standards, process, and implications). We analyze this architecture through four accountability goals (compliance, report, oversight, and enforcement). We argue that these goals are often complementary and that policy-makers emphasize or prioritize some over others depending on the proactive or reactive use of accountability and the missions of AI governance.
1
It develops a seven-feature accountability architecture comprising context, range, agent, forum, standards, process, and implications.
2
It identifies three conditions enabling accountability: recognition of authority, interrogation, and limitation of power.
3
The framework analyzes accountability through four goals: compliance, reporting, oversight, and enforcement.
4
The paper defines AI accountability as answerability, addressing ambiguity caused by AI’s multifaceted values and sociotechnical structure.
5
These accountability goals are often complementary, but policymakers prioritize them differently according to proactive or reactive governance and specific AI-governance missions.

accountability in artificial intelligence governance

the conceptual architecture, conditions, features, and governance goals that define and operationalize accountability in AI

Publication Details
Publication Date
2022-01-01
Journal
Publisher
ISSN
Cited by
496
Access Type
Author Information
Authors
Luciano Floridi
Mariarosaria Taddeo
Claudio Novelli
Explore further
Open the scid.ai AI chat with a ready-made request: it will find papers on a similar topic and help build a literature review.
Find similar papers in the chat →
Make a presentation
100%