Person
1 episodes · 12 quoted claims
Navrina Singh says Credo AI created and coined the term 'AI governance' when it was founded in 2020, when no market for it existed.
“there was no market in AI governance, you know, so it was we actually at Credo AI created and coined the term AI governance”
Navrina Singh says she sits on President Biden's National AI Advisory body, which gave her a chance to bring AI governance into government policy.
“I currently sit on President Biden's National AI Advisory Commission, and so really getting that opportunity to bring AI governance to policy and government ecosystem”
Navrina Singh says she has served for about two years on the National AI Advisory Committee, which is part of the US Department of Commerce.
“it's just been such an honor past two years to be on the National AI Advisory Committee that is part of, you know, Department of Commerce.”
Navrina Singh says that when AI took off around 2018 and 2019, many analysts believed governance would not be important.
“when artificial intelligence started to take off back in 2018, 2019, there was a thesis that, oh, governance is not gonna be important.”
Navrina Singh says companies can align their AI systems to standards such as the NIST Risk Management Framework and ISO standards.
“you can align to standards like NIST Risk Management Framework. You can align to standards like ISO.”
Navrina Singh says the European Commission began work on the EU AI Act almost five years before the interview.
“European Commission started working on this almost like five years ago.”
Navrina Singh says the EU AI Act sets risk categories and different requirements depending on whether a company develops or provides an AI system.
“EU AI Act has a certain uh risk categorizations, uh, depending upon whether you are a developer of artificial intelligence system or you're a provider of artificial intelligence systems.”
Navrina Singh says there is no common definition of fairness for facial recognition systems, and that about twenty-one definitions exist.
“There is no common definition. I mean, there are twenty one definitions.”
Navrina Singh says the US AI Safety Institute was recently established and, with NIST, is working on AI evaluations and benchmarks.
“I am thrilled for the work that National Institute of Science and Technology and the US AI Safety Institute, which was recently established, is doing in that space.”
Navrina Singh describes watermarking as embedding information in AI-generated content so it can be identified as machine-made, and says it is only one piece of the solution to misuse.
“We've been discussing the concept of watermarking, where essentially you are embedding some information into AI generated content that you can decipher that this is AI generated and it's not human created.”
Navrina Singh argues that voluntary commitments from technology companies don't work and that big tech must be held accountable to guardrails.
“Because as you and I have seen, and and you've been in the space for a long time, is that voluntary commitments don't work.”
Navrina Singh says she is testing 'adaptive policymaking' at the state and local level, in which policies are tried in a sandbox and revised before being finalized.
“But a concept of this adaptive policymaking is something that I've been socializing and testing out at state and local level to really understand when we let's put something out there within a sandbox, see how it gets implemented.”