Google questions OpenAI’s requires a authorities AI tsar
Kent Walker speaks at a Grow with Google launch event in Cleveland.
Google and OpenAI, two US leaders in artificial intelligence, have different ideas about how the technology should be regulated by the government, a new filing shows.
Google commented Monday in response to a request from the National Telecommunications and Information Administration to consider AI accountability in an era of rapidly advancing technology, the Washington Post first reported. Google is one of the leading developers of generative AI alongside its chatbot Bard Microsoft-supported OpenAI with its ChatGPT bot.
While OpenAI CEO Sam Altman touted the idea of a new government agency focused on AI to manage its complexity and license the technology, Google said in its filing that it would take a “layered, multi-stakeholder approach to AI.” governance” prefer.
“At the national level, we support a hub-and-spoke approach — with a central authority like the National Institute of Standards and Technology (NIST) informing the sectoral regulators that oversee AI implementation — rather than a ‘department for AI’,” Google wrote in its filing. “AI will pose unique problems in financial services, healthcare and other regulated industries, as well as problem areas that will benefit from the expertise of regulators with experience in those sectors – which works better than a new regulator making and enforcing upstream rules that don’t do.” adaptable to the different contexts in which AI is used.”
Others in the AI field, including researchers, have expressed similar opinions, saying government regulation of AI may be a better way to protect marginalized communities — despite OpenAI’s argument that the technology is advancing too quickly for such an approach.
“The problem I see with the ‘FDA for AI’ regulatory model is that it assumes that AI needs to be regulated separately from other things,” posted Emily M. Bender, professor and director of the University of Computational Linguistics Laboratory Washington Twitter. “I totally agree that so-called ‘AI’ systems should not be deployed without a certification process first. However, this process should depend on what the system is intended for. … Existing regulators should retain their jurisdiction. And enforce it.”
This contrasts with OpenAI’s and Microsoft’s preference for a more centralized regulatory model. Microsoft President Brad Smith said he supports a new government agency to regulate AI, and OpenAI founders Altman, Greg Brockman and Ilya Sutskever have publicly expressed their vision of regulating AI in a similar way to nuclear power, within the framework of a global AI regulator similar to the International Atomic Energy Agency.
OpenAI executives wrote in a blog post that “any effort above a certain skill (or resource, such as computing power) threshold must be subject to an international authority that can inspect systems, require audits, and verify compliance with security standards.” [and] Set constraints on deployment levels and security levels.
In an interview with the Post, Kent Walker, Google’s president of global affairs, said he was “not opposed” to the idea of a new regulator to oversee the licensing of large language models, but said the government should take a “more holistic” look at the technology. And NIST is already well positioned to take the lead, he said.
Google’s and Microsoft’s seemingly opposing stances on regulation point to a growing debate in the AI space that goes well beyond how much the technology should be regulated and how organizational logistics should work.
“It begs the question of whether or not there should be a new agency specifically for AI?” Helen Toner, director at the Center for Security and Emerging Technology in Georgetown, told CNBC, adding, “You should do this with the existing regulators rules that operate in specific sectors, or should there be something centralized for all types of AI?”
Microsoft declined comment and OpenAI did not immediately respond to CNBC’s request for comment.
WATCH: Microsoft is releasing another wave of AI features as the race with Google intensifies