NIST’s Budgetary Constraints: NIST, tasked with creating AI safety standards by July 26, 2024, is struggling with budget limitations. The agency’s 2023 budget was $1.6 billion, with a proposed 29% increase for 2024, yet this increase isn’t directly allocated for AI-related initiatives. NIST’s current resources are considered insufficient for independent AI safety testing, particularly compared to the vast budgets of AI industry leaders like OpenAI, Google, and Meta.
Congressional Concerns: Concerns are mounting among some members of Congress over NIST potentially relying on private AI companies for expertise, given their vested interests. On December 16, a bipartisan group of lawmakers expressed apprehensions regarding the transparency of NIST’s collaborations with external organizations.
State of AI Safety Research: The letter from Congress also highlights the nascent stage of AI safety research and the existing disagreement among experts on defining and measuring AI safety issues. This scenario complicates NIST’s task of establishing standards.
Moves Towards Transparency: In response to these concerns, NIST issued a request for information on December 19 to gather external expert opinions on AI model evaluation standards. This step might be a move towards increasing transparency in the process.
Expert Opinions: AI experts emphasize NIST’s crucial nonpartisan role in AI risk assessment but stress the need for more resources. They point out the disparity in resources between NIST and big tech companies. Additionally, the growing secrecy around commercial AI models presents further challenges in measurement and standardization.
Global Context and Comparisons: The executive order tasks NIST with several key objectives, including establishing an Artificial Intelligence Safety Institute and aligning international standards on AI development. For context, a UK taskforce on AI safety was announced in April with $126 million in seed funding, showcasing global initiatives in this domain.
Involvement of Tech Companies: While NIST’s current engagement with major tech firms remains unclear, its earlier discussions on AI risk management involved significant players in the industry, such as Microsoft, Anthropic, and the Partnership on AI.
In summary, while the Biden administration’s ambitious plan for AI safety is a step forward, its successful implementation hinges on addressing the financial and collaborative challenges faced by NIST. Ensuring transparency and independence in the development of AI standards remains a key concern amid these challenges.
Primary Sources: Ars Technica Rating, Wired Rating
Your Feedback Matters!
How engaging was the article?
Discover more from News Facts Network
Subscribe to get the latest posts sent to your email.
