A fear that AI may be as dangerous as nuclear warfare is spurring demands for federal oversight, which could include establishing a distinct federal body tasked with AI regulation. But some experts contend creating an AI regulator could be a mistake.
The idea of establishing a federal AI regulator was discussed during OpenAI CEO Sam Altman's testimony in front of a Senate committee in May. Following Altman's testimony, Sen. Michael Bennet, D-Colo., introduced a bill to create a new agency to oversee AI and digital platforms.
The Biden administration is exploring AI regulation. It has asked for public input on what standards and regulations are needed for the technology, but it has yet to move forward on measures like creating a separate regulatory body for AI. That is something experts argued against during the Center for Data Innovation's recent web panel.
"I'm not in favor of the idea of an AI regulator that regulates models or technologies or algorithms," said Ben Schneiderman, a professor in the department of computer science at the University of Maryland.
Regulating AI with existing laws, agencies
Schneiderman said he's not against regulation, but the goal should be to regulate the outcomes, not the technology itself. In terms of outcomes, that could mean ensuring AI algorithms don't discriminate in employment, housing and finance.
"What we want to do is make AI that people feel in control of," he said during the panel.
To do that, instead of creating one overarching AI regulator, Schneiderman said existing agencies -- the U.S. Securities and Exchange Commission, the Federal Trade Commission, and others -- are well-placed to act against companies using AI that violate existing consumer protection laws. Multiple enforcement agencies, including the FTC, recently warned businesses that AI is not immune to consumer protection laws.
The federal government has 2.1 million civilian employees, 50 independent regulatory bodies and 430 different federal departments. "It is crazy to think that nobody there is paying any attention to AI," said Adam Thierer, a senior fellow at R Street Institute, during the panel. Agencies are already pursuing AI cases, such as the Equal Employment Opportunity Commission's open proceeding on employment hiring algorithms, he said.
"Almost every agency today is looking at how AI, machine learning and robotics affects their agency and their authority," he said.
Gaps in existing law need to be addressed
Existing laws do have gaps and don't deal with some of the bigger issues with AI, said Lee Tiedrich, a faculty fellow in ethical technology at Duke University, during the panel discussion.
Ben SchneidermanProfessor, department of computer science, University of Maryland
The laws, for instance, don't explain what it means for AI to be fair and transparent -- a request seen in guidance like the White House's Blueprint for an AI Bill of Rights.
"We need to be really smart and strategic about it and make sure that we figure out where are the gaps, what are the best ways to fill the gaps and be very thoughtful about putting regulation in place," she said.
Tiedrich said that before creating a specific AI regulator, federal agencies can review their resources and assess what additional capacity they may need to handle AI regulation.
"That would give the government and the rest of the community a really good landscape on where we exist today," she said. "Then we could go back on making smart decisions on how to fill the gaps."
Makenzie Holland is a news writer covering big tech and federal regulation. Prior to joining TechTarget, she was a general reporter for the Wilmington StarNews and a crime and education reporter at the Wabash Plain Dealer.