Reps. Ted Lieu, D-Calif., left, and Pete Aguilar, D-Calif., on Jan. 25 in Washington. Ricky Carioti/The Washington Post

Rep. Jay Obernolte has said it many times: The biggest risk posed by artificial intelligence is not “an army of evil robots with red laser eyes rising to take over the world.”

It’s the “less obvious” and more mundane issues such as data privacy, antitrust issues, and AI’s potential to influence human behavior. All these take precedence over the hypothetical notion of AI ending humanity, the California Republican says.

Obernolte would know: He’s one of a handful of lawmakers with a computer science degree – including graduate research on AI in some of its earliest stages. With the rise of generative AI applications like ChatGPT – what some observers have dubbed a “big bang” moment – Obernolte has emerged as a leading expert in Congress on how the technology works and what lawmakers should worry about.

An array of AI boosters and doomsayers have already converged on Washington, seeking lawmakers’ ears as they gear up for a drawn-out policy debate that is likely to grip Congress. As an AI expert and a leader in the House AI Caucus, Obernolte wields power in steering that debate and explaining what is and is not possible, his colleagues say. Yet it remains unclear whether Obernolte and his very small club of tech-savvy policymakers will have a meaningful effect in Congress, which has failed time and again to pass substantive tech regulation.

“If we don’t pay attention to this and act on it more quickly and more effectively than we have on social media, we could see even more harmful consequences for our society, our culture, our economy moving forward,” Sen. Christopher A. Coons, D-Del., told The Washington Post on Tuesday ahead of a hearing with OpenAI CEO Sam Altman, whose company created ChatGPT.

With some in Congress saying the rise of AI has given lawmakers a new sense of urgency, Obernolte and his colleagues are optimistic there will be bipartisan momentum.

Rep. Don Beyer, D-Va., who is pursuing a master’s degree in machine learning, is also eager to identify areas where Congress can act.

“My team and I have been trying to figure out, if there were a legislative effort, what would it look like?” Beyer said in an interview. “When we even get to a first semi-draft, Jay will have to be one of the first people I talk to.”

Obernolte takes pride in his ability to explain complicated technical concepts in simple terms. And there’s no question that a good number of his colleagues need that. Members of Congress often take heat for lacking knowledge about basic concepts while questioning tech sector leaders, an affliction most recently on display in March during a House panel’s grilling of TikTok’s CEO.

But Obernolte stood out for the opposite reason. He asked Shou Zi Chew pointed questions about the company’s plan to protect U.S. consumer data from the Chinese government, saying he was skeptical that the plan as it stands guaranteed the app would remain free from foreign influence.

More broadly, he is blunt about the paucity of tech-savvy lawmakers. “We need more computer science professionals in Congress given the complexity of some of the technical issues we grapple with,” he told The Post.

Obernolte latched onto computers at the tender age of 8 when his father gave him an Apple II computer and a book on how to code in BASIC, an early computer language.

“I enjoyed it,” Obernolte said. “It catalyzed a lifelong interest in computer science for me.”

He took that passion to a math- and science-focused high school, where he showed such a knack for computer science that he was placed in an independent study program writing educational software.

Obernolte received a degree in engineering and applied sciences, with a focus on computer science, from the California Institute of Technology in 1992. He then continued to program video game software, scoring a gig with a company that worked with Nintendo, maker of the most popular home-gaming systems at the time.

He recalls that when he told his mother the news, she responded: “That sounds great, honey, but wouldn’t you rather do something that might have a future for you?'”

That early success ended up defining his future. Obernolte enrolled at the University of California at Los Angeles for a graduate degree in artificial intelligence. He studied “natural language processing” – the process of translating written text into something a machine can understand – as well as “computer vision,” the technology that self-driving cars now use, he said. Both areas, Obernolte noted, “are still a subject of intense research.”

Obernolte continued to write video games on the side, and he says he was on his way to earning a doctorate until a game he programmed – “NFL ’95” for the Sega Genesis – became a big hit. The game’s success meant Obernolte had a decision to make.

“It was very clear that either I could finish my doctorate or I could start a company,” he said.

Obernolte decided to focus on his company, FarSight Studios, many of whose games are sports and family-friendly games, including realistic pinball games and an adventure-themed puzzle game called “Orbals.”

He also has other passions. He enjoys water skiing and snowboarding and holds a black belt in an American style of karate called puma.

But it was Obernolte’s flying hobby that led him into politics in the early 2000s when he sought a seat on the panel overseeing the airport in Big Bear, where he was a tenant. He eventually ran for the Big Bear City Council and served as mayor. He then spent six years as a state legislator before winning his congressional seat in 2020 with the backing of former president Donald Trump.

Obernolte showed his loyalty to Trump just three days after he was sworn into Congress when he voted against certifying the 2020 presidential election results in Arizona and Pennsylvania – claiming that the votes were not valid because of “unilateral” changes to election laws in those states – even though Trump’s legal challenges in those states had already failed. That same day, a violent mob of Trump supporters stormed the Capitol. Days later, Obernolte voted against impeaching Trump for his role in inciting the attack, saying such an action “will only divide our nation further.” Obernolte added that the “people who ransacked our nation’s Capitol must be found, arrested, prosecuted, and convicted.”

Through his office, he declined to comment on those votes and whether he’ll support Trump in the 2024 presidential election.

On other matters, Obernolte sits closer to the center of the political spectrum, with a voting record that suggests he’s more liberal than 87% of House Republicans, according to an analysis by Voteview. And Obernolte’s Republican affiliation has not stopped some Democrats from wanting to work with him on AI policy, and vice versa.

Gregory Allen, a senior fellow in the strategic technologies program at the Center for Strategic and International Studies, said Congress frequently legislates around issues it doesn’t understand – and its members often look to colleagues whom they trust as experts to guide them. When NASA Administrator Bill Nelson was a U.S. senator, for example, his colleagues sought the former astronaut’s guidance on space issues, Allen said.

“Congressman Obernolte is trying to develop the trust of his colleagues on this issue and to be their sort of go-to resource for advice,” Allen added while declining to comment on whether Obernolte has yet achieved that role.

Sitting in his Capitol Hill office recently, Obernolte insisted that programs such as ChatGPT are not self-aware – and may not be for a while: “There is no there, there. There’s no deeper understanding.”

Aparna Pappu, vice president and general manager of Google Workspace, talks about generative AI this month at the Google I/O conference in Mountain View, Calif. Melina Mara./The Washington Post

Instead, Obernolte compared large language models, the technology that powers ChatGPT and rival chatbots, to parrots – as others in the field have.

“You could be amazed at the sentences that a parrot is saying because it’s repeating what it has heard – what it’s been trained on,” he explained. “But no one pretends that the parrot understands what it’s saying.”

But the launch of ChatGPT in November touched off an arms race among tech’s most powerful players. Google, Microsoft, and other companies are moving at such breakneck speed to develop the technology that some industry groups have called for a pause, citing risks to society and humanity.

Obernolte acknowledges that the explicit mission of some companies, such as OpenAI, is to develop programs that are smarter than humans. This development will prompt humans to challenge preexisting notions of “intelligence” and “sentience” and ask how they could interact with that technology in the future, he argues.

“It’s something that, frankly, we’re going to struggle with,” he said.

Obernolte does not think companies should pause development and wrote in a recent op-ed in the Hill that AI’s “benefits to society will likely far outweigh the costs,” even if that requires shifts in law, education, and workforce.

But he told The Post that he also has concerns about the technology: the potential for Big Tech companies to hoard AI training data and AI’s ability to aid in cybercrime. He also points to AI’s “uncanny ability” to break past digital data privacy and create accurate predictions of human behavior – an issue that long predates generative AI going mainstream.

One response, he said, is possibly new data privacy legislation being hammered out in the Energy and Commerce Committee, a key panel Obernolte sits on. In December, at the very end of the 117th Congress, the committee passed a data privacy bill with backing from both sides of the aisle – but time ran out on the session before a full House vote could be held and the measure flamed out, joining every other major data privacy bill that Congress has failed to pass.

But Obernolte said he was encouraged by the bipartisan support. “I am cautiously optimistic that, in the next year or so, we will be successful in getting that federal data privacy regulation enacted,” he said.

Another measure in play is a bipartisan bill, which he introduced in March, that would clear the Defense Department to more easily procure AI-powered cybersecurity systems.

As far as specifically regulating the latest wave of AI tools, however, Obernolte has yet to act. As he sees it, Congress should wait until it develops a regulatory framework that clearly defines the technology’s biggest risks. “If you don’t understand what you’re trying to protect against, you can’t create regulations that enact that protection,” he said. Through his office, Obernolte declined to say whether he’ll support a bill that Rep. Ted Lieu, D-Calif., said he plans to introduce establishing a bipartisan commission to better understand AI.

Obernolte and Lieu each said regulating AI is a bipartisan issue. But they differ on whether AI regulation needs an entirely new agency.

In February, Lieu told The Post that having a “general agency do regulations” would be more effective than regulating “artificial intelligence in every discrete instance in which it is used.”

In contrast, Obernolte told The Post that punting the “entire issue to bureaucracy would be the wrong approach” and said regulations should instead move through Congress and be continually updated by the legislature. Federal agencies, he said, should “fill in the details.”

Despite their differences, Obernolte and Lieu both said they’re working together on the issue.

“I’m not sure there’s a disagreement on the result,” Lieu said in a recent interview. “A number of Democrats and Republicans do believe that some AI should be regulated. The question is how we go about doing that.”

He added: “There is nothing inherently partisan about AI – and it’s great to have experts like Congressman Obernolte on this issue.”