‘Wonder and Worry’: How Biden Views Artificial Intelligence

WASHINGTON—President Biden has watched as aides pulled up fake images of him generated with artificial intelligence. Together, they have asked ChatGPT to explain an obscure Supreme Court case and used AI to make artistic renderings of Biden’s dog, Commander.

The president observed “with wonder and worry,” according to a senior White House adviser.

These instances were among many closed-door discussions Biden has held on AI as he and his administration prepare to impose a broad new set of guardrails around the technology. Addressing his cabinet in June, the president asked his appointees to report back by their next meeting in the fall on what actions to include in a comprehensive government approach toward the rapidly evolving technology.

Biden’s effort faces a steep climb. Previous administrations and U.S. lawmakers largely failed in regulating social-media companies before they became too powerful to rein in, even though there was bipartisan support to do so. There is little appetite in a gridlocked Congress to pass major bills in the run-up to a presidential election, and Biden’s aides acknowledge the president is limited in his powers to act unilaterally.

The urgency to act reflects how so-called generative AI tools such as ChatGPT, which can instantly create high-quality digital content, have captured Washington’s attention at the highest levels. The issue of AI has taken on a deeper meaning for Biden, who, in conversations with top officials in his administration, has suggested that decisions made around the technology could impact generations and shape part of his legacy.

“He views this as an existential challenge,” Bruce Reed, Biden’s deputy chief of staff, said in an interview. “This is not just another issue that came along. This is a technology that has the potential for phenomenal good or ill.”

There are already concerns that AI will displace low-wage workers from jobs, threaten data privacy and fuel online misinformation. Its algorithms are vulnerable to ethical bias and discrimination. 

Some conservatives and industry executives are worried that a broad government effort to manage AI could thwart the technology’s benefits.

“You need to give innovation a wide berth and not try to micromanage it,” said Adam Thierer, a senior research fellow in the R Street Institute, a conservative think tank. “If you spend all your time worrying about hypothetical worst-case scenarios and you base public policy on worst-case scenarios, then a very many best-case scenarios don’t come about.”

Others say that what the Biden administration has rolled out thus far, such as recent voluntary commitments from seven big AI companies to watermark AI-generated content and test their systems for security risks, doesn’t go far enough. Without any penalties for falling short, the commitments don’t challenge the AI companies involved to go beyond the safety practices they have already implemented or promised, said Sarah Myers West, a researcher at the AI Now Institute, a New York University research group.

“A closed-door deliberation resulting in voluntary safeguards isn’t a substitute for enforceable guardrails,” she said. “These companies certainly know the difference between voluntary commitments and regulation that carries the force of law.”

In previous AI guidelines issued by the White House, steps were also nonbinding and didn’t include enforcement measures. White House officials noted that these initial actions aren’t a substitute for legislation or government regulation, and that the companies agreeing to external audits was a first step toward accountability.

The White House is in the process of developing an executive order on AI that an official said will aim to protect workers and consumers and address cybersecurity risks. Previous guidelines issued under Biden have been aimed at safeguarding personal data from misuse in AI algorithms, such as those that drive hiring, lending and other business decisions, and were described by the White House as a blueprint for an AI bill of rights.

Aides say the push to do more has been driven by Biden himself, who has voiced particular concern over the technology’s implications for foreign policy and national security. The president has cited the potential danger of a deep fake featuring the U.S. president being mistaken as authentic overseas, according to people who have spoken with Biden about AI. The president has also questioned the technology’s potential effects on democracy at a time when online misinformation is already looming over elections.

Biden’s own experimentation with ChatGPT has left the president amazed by the speed with which the tool can generate material. At an Oval Office meeting in April, aides prompted ChatGPT to create content related to a lawsuit, known as New Jersey v. Delaware, that Biden had joked about at a recent event honoring Bruce Springsteen.

The president and his aides saw how ChatGPT would explain the lawsuit to a first-grader, how it would write a brief on the case for an associate fresh out of law school and even how it would write a song about the lawsuit in the style of Springsteen. 

“His directive was, we’ve got to move fast here,” White House chief of staff Jeff Zients said in an interview. “Too often, technology moves faster than we do as a federal government.”

Recent scenarios involving fake images have demonstrated the potential consequences. In May, the stock market briefly dipped after an AI-generated image of a fake explosion at the Pentagon went viral. 

Tech executives and artificial-intelligence scientists issued a joint statement in May warning that AI poses an extinction risk as great as pandemics and nuclear war. Signatories included Sam Altman, chief executive of OpenAI, the company that developed ChatGPT, and executives from Microsoft and Google. 

These companies are seeking to shape any regulations, and Altman has spent weeks meeting with lawmakers and other leaders globally to discuss AI’s risks and his company’s idea of safe regulation. Altman has called for the U.S. government to create a new agency that would license large-scale AI systems to ensure the technology was developed safely.

Biden administration officials say there is also much to seize on with respect to AI’s capabilities, including improvements in cancer research and detection and mitigating climate change. The White House points to how AI tools can both increase productivity and create better and more skilled jobs for some workers.

Still, some view Biden’s attitude toward Big Tech as more aggressive than his predecessors.

When Facebook struggled to contain conspiracy theories about Covid-19 vaccines in 2021, the president accused the social-media giant of “killing people”—comments he later walked back, but not without calling on the company to do more to combat misinformation. In his State of the Union address last year, Biden pledged to hold social-media platforms accountable “for the national experiment they’re conducting on our children for profit.”

That skepticism was on display when Biden convened civic leaders at the Fairmont in San Francisco for a discussion on AI last month. When attendee Tristan Harris introduced himself as the CEO of the Center for Humane Technology, the president joked, “Is that an oxymoron?”

—Ryan Tracy and Deepa Seetharaman contributed to this article.

Write to Sabrina Siddiqui at [email protected]