Artificial intelligence (AI) technologies holds great promise to improve the lives of millions of Americans yet also pose risks that include biased application and potential misuse by foreign adversaries, Senate Homeland Security and Governmental Affairs Committee Chairman Gary Peters, D-Mich., said on March 8.
Peters opened a committee hearing on “Artificial Intelligence: Risks and Opportunities” by offering an overview of what he sees as the positive and negatives of a technology that has been an increasing focus of public debate in recent years.
“The adoption of AI in government and industry and civil society has led to rapid growth of advanced technologies in virtually every sector, transforming millions of American lives all across he country,” said Peters, who co-sponsored a law signed by President Biden last October that creates a training program to help Federal employees who purchase AI technologies understand the benefits and risks.
As examples of AI’s vast benefits, Peters cited its role in creating life-saving drugs and self-driving vehicles, along with helping government and businesses better serve the public.
At the same time, Peters said AI’s lack of transparency and accountability risks leading to outputs that discriminate on the basis of race or sex, while the technology emboldens adversaries such as China in ways that could threaten U.S. national security.
“The right safeguards need to be in place,” said Peters, who indicated “future legislative activities and oversight” from the committee on AI, but did not provide specifics.
The committee examined various aspects of AI and heard from three outside experts at a time when AI is at the center of debate over its benefits and risks, in part because of ChatGPT – the chatbot technology launched late last year to much fanfare for its ability to create highly realistic narratives.
The discussion is resonating inside the Federal government, where efforts are underway to train employees and the future workforce and identify staff with AI skills. Recent MeriTalk research, in partnership with Future Tech, shows that 95 percent of Federal IT decision makers say their organization is investing in-house AI skills development.
Few Republican committee members attended the hearing because of a dispute with Democrats over witness testimony, and the ranking member, Sen. Rand Paul, R-Ky., was not present.
One witness who testified, RAND Corp. President and CEO Jason Matheny, emphasized that threats from AI development pose national security challenges, and he called on intelligence agencies to expand collection and analysis of information on adversary states.
“By most measures, the U.S. is currently the global leader in AI,” Matheny told the committee. “However, this may change as China becomes the world’s primary AI innovation center by 2030.”
Sen. Maggie Hassan, D-N.H., asked Matheny how AI impacts the cybersecurity threat landscape.
In reply, he expressed concern that the application of large language models could allow even non-state actors to obtain offensive cyber capabilities, potentially threatening critical infrastructure. But he added that the same tools could be used to counter those threats.
“I think this will be a cat and mouse race to figure out if the applications on the defensive side are keeping up with the applications on the offensive side,” Matheny said.
Hassan then expressed concern that AI development could “change the nature of work in fundamental ways,” potentially displacing workers. She asked a second witness, Suresh Venkatasubramanian, to comment. He is a professor of computer science and data science at Brown University.
He responded that the Federal government should address that by investing “effort and research into helping workers train for our STEM-enabled workforce.”