Ai Safety And The Us China Arms Race

2024-10-28 · Source: Our Work | Center for AI Policy (CAIP) · Field: Government & Public Sector — Public Policy & Governance, International Relations & Diplomacy, Artificial Intelligence & Machine Learning · Depth: Intermediate, quick

Summary

An analysis challenges the notion that AI safety requirements undermine US technological competition with China, concluding that safety measures will not hinder the US in an "AI race." The report finds that inexpensive AI safety costs, specifically mandatory pre-deployment evaluations for powerful models, are unlikely to impact innovation, costing less than 0.5% of upfront training costs, even as training costs approach \$1 billion by 2027. Furthermore, safer AI can foster public trust and drive adoption, crucial for economic growth, and these safety requirements on public models will not impede military capabilities, as the government can partner to access models pre-release. The US can maintain its lead through other strategies like immigration reform and investment in upskilling.

Key takeaway

For policymakers weighing AI regulation against national competitiveness, this analysis suggests that introducing AI safety measures, such as pre-deployment evaluations, does not create a trade-off. You should pursue these safety requirements to build public trust and drive economic growth, while simultaneously implementing strategies like targeted immigration reform and investment in AI upskilling to maintain an innovation lead over competitors. A "wait and see" approach endangers public safety without improving national competitiveness.

Key insights

AI safety measures are compatible with US national competitiveness and do not hinder its lead in innovation.

Principles

• Low AI safety compliance costs do not impede innovation.
• Increased AI trust drives adoption and economic growth.
• Public model safety requirements do not affect military AI.

Method

Implement mandatory pre-deployment evaluations for developers of the most powerful AI models to assess risk before public release.

In practice

• Supplement hardware industrial policy with targeted immigration reform.
• Invest in upskilling business users of AI.
• Support military technology innovators with government contracting.

Topics

• AI Safety
• US-China Competition
• AI Regulation
• Pre-deployment Evaluation
• National Security
• Economic Growth

Best for: CTO, VP of Engineering/Data, Director of AI/ML, Policy Maker, Executive, Consultant

Related on AIssential

• Trump abruptly cancels EO signing event after top AI firm CEOs declined to go — The US faces internal and external pressures balancing AI innovation with government-mandated safety testing, impacting its global regulatory standing.
• Competition may push AI firms to favor speed over safety, new study finds — Competition in AI development risks prioritizing speed over safety, challenging policymakers.
• US restricts AI models so China builds better ones and everyone switches — US AI restrictions may inadvertently cede global AI leadership and monitoring capabilities to China.
• A new plan emerges for AI apocalypse avoidance — AI 2040: Plan A proposes a US-China pause on frontier AI research to build safety, asserting safety and growth can coexist.
• Ai Safety Is Becoming Ai Security — The UK's AI Safety Institute rebranding to AI Security Institute signals a policy shift towards adversarial risks, potentially neglecting inherent system failures.
• Washington weighs new bipartisan AI proposal — Rapid AI advancement necessitates urgent, multi-faceted regulatory and safety measures, alongside addressing infrastructure challenges.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Our Work | Center for AI Policy (CAIP).