The director of the White House Office of Science and Technology Policy (OSTP) said last week that the National Institute of Standards and Technology (NIST) must resume working on “going back to basics” and creating AI standards.
“I think we need to go back to the basics at NIST and go back to the basics centered around what NIST exists. That's about doing the best advertising on class standards and doing important metrics or measurement science centered around AI models.”
Kratsios claimed that the Biden administration “hijacked” the NIST and “turned it into a safety model assessment agency,” and referred to the AI Safety Institute, which was established in the agency to assess risk assessments that existed.
Instead of focusing on assessments, Kratsios proposed that NIST, responsible for developing standards, measurements and technologies for security and innovation in the US, would become the “gold standard” agency, and later standardize the assessment.
“We need to be able to be in a position where you speak the same language about how you rate. What does a good rating look like?” Kratsios said. He added that sector-specific assessments are possible “if they can do a much better job of defining standards regarding EVALS.”
The OSTP Director explained that his vision for NIST focused on defining a good standard of methodology, allowing each industry to create sector-friendly metrics and lumics.
“The role of government is not to create something similar to a similar standard for the financial industry,” Kratzios said as an example.
“But at the end of the day when NIST can play core values, you can naturally understand what the ratings look like,” he continued.
Last month, the Ministry of Commerce Renamed the American Institute of Artificial Intelligence Safety (AISI) Remove safety features from its name to the AI Standards and Innovation Center (CAISI). The department said that brands' brands do not mean they will reduce safety concerns, but will focus on AI innovation.
The list of goals to be achieved under NIST's new CAISI is committed to developing guidelines and best practices to measure and improve AI security and to develop voluntary standards.
