Skip to content

R0058/2026-04-03/C001/SRC04/E01
Research R0058 — Candidate evidence test
Run 2026-04-03
Claim C001
Source SRC04
Evidence SRC04-E01
Type Reported

NeurIPS 2019 workshop identifies "epistemic fractures" between fairness, ethics, and safety research

URL: https://mindingthegap.github.io/

Extract

The workshop description states:

  • Researchers, practitioners, policymakers, and the public discuss AI impacts using "multiple conceptual frames that don't align easily"
  • "Questions of algorithmic fairness arising from concerns similar but distinct from those around AI safety, which in turn overlap with but are distinct from questions motivating AI ethics work"
  • The workshop addresses "epistemic fractures that undermine valuable synergy and productive collaboration"
  • "Misunderstandings arising from linguistic slippage between different frames" contribute to the divide
  • The workshop aims to "translate between these ongoing efforts and bring them into necessary conversation"

Relevance to Hypotheses

Hypothesis Relationship Strength
H1 Supports The workshop's existence and framing confirm that the safety-ethics divide was recognized by the research community, consistent with the claim's premise
H2 Supports Consistent with the divide being real; the workshop predates the quantitative measurement by Roytburg & Miller, showing the problem was recognized before being measured
H3 Contradicts The fact that a major venue hosted a workshop specifically about this gap makes it harder to argue the divide does not exist

Context

This workshop predates the Roytburg & Miller study by six years, demonstrating that the safety-ethics divide was recognized qualitatively before it was measured quantitatively. The workshop framing identifies the divide as spanning three communities (fairness, ethics, safety) rather than just two (safety, ethics), which adds nuance to the binary framing in the claim.

Notes

The workshop identifies "linguistic slippage" as a driver of the divide — different communities using different terminology for overlapping concepts. This aligns with the vocabulary exploration in Step 1 and suggests the divide is partly terminological rather than purely substantive.