Yesterday, the House Science, Space, and Technology Committee leaders sent a letter to the National Institute of Standards and Technology (NIST) Director Laurie Locascio, emphasizing the significance of careful consideration in funding research on artificial intelligence (AI) safety. The letter specifically addressed the establishment of the Artificial Intelligence Safety Institute (AISI) at NIST, which was mandated by President Biden's October 30 Artificial Intelligence Executive Order.
In the letter, the committee members highlighted the importance of NIST awarding research funds transparently and promoting scientific and methodological quality in collaboration with its AI safety research partners. They also requested a briefing from NIST to discuss the AISI process and the utilization of funds.
The committee members expressed concerns regarding the current state of the AI safety research field, stating that there are challenges as NIST takes on a leadership role in addressing the risks associated with AI. They noted that the findings within the community often lack the quality that comes from revision in response to critiques by subject matter experts. Additionally, there is significant disagreement within the AI safety field regarding scope, taxonomies, and definitions.
The letter emphasized the committee's long-standing support for NIST's efforts in advancing trustworthiness in AI systems. They highlighted previous legislation, such as the National Artificial Intelligence Initiative Act of 2020 and the CHIPS and Science legislation, which demonstrate their commitment to this cause. However, they stressed the need for thoroughness and cautioned against rushing the process at the expense of accuracy.
The committee members acknowledged that developing evaluation suites and appropriate metrics for AI trustworthiness could take years, considering the deployment of AI systems across various sectors and use cases. They emphasized that scientific merit and transparency should be paramount considerations as NIST prepares to fund extramural research on AI safety. They expect NIST to enforce rigorous guidelines of scientific and methodological quality for recipients of federal research funding in the AI safety research field.
The letter from the committee leaders can be found [here](link to the letter).
In summary, the House Science, Space, and Technology Committee leaders have stressed the importance of diligence in NIST's AI safety research funding. They have called for transparency, scientific rigor, and methodological quality in the allocation of research funds and the collaboration with AI safety research partners. The committee members have also expressed concerns regarding the current state of the AI safety research field, highlighting the need for revision and consensus in the community. As NIST moves forward with the AISI, they expect strict adherence to scientific guidelines and a commitment to promoting trustworthiness in AI systems.
For additional details, please follow this link: https://science.house.gov/press-releases?ID=0E928EB3-9452-47D1-84C1-B912CCC0125B
