Claims

Factual · AI policy and governance

“there's no clear guidelines around how to do that now, right.”

Daniela Amodei, Anthropic president · 2022 · Podcast Transcript: Daniela and Dario Amodei on Anthropic (LessWrong)

As of 2022 there are no clear guidelines for AI developers on how to avoid advancing race dynamics or do the right thing on safety.

Resolved true The record shows it happened or was so.

The second judge said mixed; the verdict shown is the first judge's.

What would count as true or false

Judged as of 2022. True: as of 2022 there was no established, widely accepted guidance or standard telling AI developers how to avoid fuelling competitive race dynamics or how to act responsibly on safety (only drafts or general principles existed, for example a NIST AI RMF still in draft). Mostly true: some general principles existed (for example OECD AI Principles, Partnership on AI materials) but they were vague or non-binding, consistent with 'no clear guidelines'. Mixed: there were some reasonably concrete guidelines for safety but none on race dynamics, or the reverse. Mostly false: clear and widely recognised guidelines on both safety practice and race dynamics existed, and the claim overstates their absence. False: specific, authoritative guidelines covering these matters were in place and widely adopted by 2022.

Evidence

1. Microsoft's framework for building AI systems responsibly - Microsoft On the Issues Microsoft · 21 June 2022

   “It provides specific, actionable guidance for our teams that goes beyond the high-level principles that have dominated the AI landscape to date.”

2. Microsoft's framework for building AI systems responsibly - Microsoft On the Issues Microsoft · 21 June 2022

   “and yet, our laws are lagging behind. They have not caught up with AI’s unique risks or society’s needs.”

3. NIST Seeks Comments on AI Risk Management Framework Guidance, Workshop Date Set | NIST National Institute of Standards and Technology (NIST) · 18 August 2022

   “NIST is seeking comments on a second draft of the NIST Artificial Intelligence Risk Management Framework (AI RMF). The AI RMF is intended for voluntary use in addressing risks in the design, development, use, and evaluation of AI products, services, and systems.”

4. [PDF] Safety Not Guaranteed: International Races for Risky Technologies Governance.AI (authors: Eoghan Stafford, Robert F. Trager, Allan Dafoe) · November 2022

   “What has not been adequately studied are the conditions under which these dynamics are more and less severe.”

Reasoning

E3 shows that as of August 2022 the NIST AI Risk Management Framework was still a second draft open for comment, and was meant for voluntary use. E1 describes the AI landscape up to mid-2022 as "dominated" by high-level principles. E2 says laws had not caught up with AI's risks. Microsoft's own standard (E1) was guidance for its internal teams, not a widely accepted standard. E4 notes that the conditions that make race dynamics more or less severe had not been adequately studied. This matches the "true" criterion: apart from one company's guidance for its own teams [E1], only drafts or general principles appear in the evidence for 2022, and none of it shows established, widely accepted guidance [E3].

Cite this claim

Daniela Amodei, “As of 2022 there are no clear guidelines for AI developers on how to avoid advancing race dynamics or do the right thing on safety.” (said 2022). Verdict: resolved true. Reliability API, claim c_bc978704239a, method 0.1.0+28da144b. https://epistacklabs.com/claims/c_bc978704239a