Responsible AI Starts with the Artifact: Challenging the Concept of Responsible AI in IS Research

Summary

An EJIS editorial arguing that the dominant principles-first approach to "responsible AI" — which begins with predefined ethical principles (fairness, transparency, inclusiveness) and asks systems to conform to them — is in tension with the actual characteristics of contemporary AI artifacts. The authors highlight that AI applications are characterised by inscrutability, learning, and autonomy, "aspects that are antithetical to some of the key principles of responsible AI," producing a persistent gap between abstract responsibility frameworks and the systems they purport to govern.

Contribution

The editorial reframes responsible AI as starting "with the artifact": IS research should ground responsibility claims in the material properties of the AI system (predictive, generative, or agentic) rather than in a top-down principles checklist. This positions IS as the discipline best placed to bridge artifact-level mechanics and governance-level norms.

Method

Conceptual editorial; reviews the responsible-AI literature, diagnoses the principles-first logic, and proposes an artifact-centred alternative.

Relevance to RISE

EJIS editorial arguing that principles-first responsible-AI frameworks conflict with the artifact's intrinsic characteristics (agentic, autonomous, adaptable, inscrutable). Strong methodological challenge to how RISE systems should be evaluated against governance principles.

Critique / open questions

As an editorial the piece does not provide empirical evidence or a worked example of artifact-first evaluation; the principles vs. artifact opposition risks being a false binary if governance must ultimately translate artifact properties into normative criteria anyway.

Key quotes

"A prevailing issue with the current discourse on responsible AI frameworks and associated research is that it builds on a principles-first approach."

"Contemporary AI applications are characterised by inscrutability, learning, and autonomy, aspects that are antithetical to some of the key principles of responsible AI."