Senators studied AI for a year. Critics call the result ‘pathetic.’

“This road map leads to a dead end,” Evan Greer, the director of Fight for the Future, an advocacy group, said in a statement that called the plan “pathetic” — a criticism echoed by others.

The immediate divisions over the plan bring to the fore the challenges of regulating the swiftly evolving technology in a deeply polarized Congress during an election year. Rather than pursuing a single comprehensive bill, the AI Gang has decided to delegate the legislative work to Senate committees, which are at drastically different stages in their efforts to understand the promise and peril of AI.

Schumer (D-N.Y.) expects that some AI bills could pass the Senate and potentially the House by the end of the year, but he noted that much of this work will extend into the next Congress. But the plan faces an uncertain future next year, as key lawmakers working on tech issues are scheduled to retire and the 2024 elections could reshape the leadership of Congress and the White House. The Senate Rules Committee on Wednesday advanced a trio of bipartisan bills addressing the role of AI in elections, which Schumer noted was one of the most urgent issues facing Congress amid the 2024 elections.

“We’re not going to wait on legislation that addresses every aspect of AI in society,” Schumer said. “If some areas are ready earlier than others, they should go forward.”

Other congressional committees are just beginning their work on artificial intelligence, as major tech companies are plowing forward with ever more advanced systems intended to further entrench the technology in consumers’ lives. On Monday, OpenAI announced a handful of upgrades that will make it easier for people to talk to ChatGPT, drawing comparisons with the 2013 film “Her,” which depicts a human falling in love with a digital voice assistant. On Tuesday, Google announced that it would roll out this week AI-generated answers to the top of everyone’s search results in the United States, transforming the way people access information online.

Lawmakers have repeatedly promised that they will regulate AI with greater urgency than they did with social media. For the last half decade, lawmakers have held numerous hearings and introduced a flurry of bills to address the ways social media allegedly harms children, undermines elections and imperils users’ privacy. But to date, the main social media legislation that Congress has passed is a law that could force a sale or ban of TikTok. With the new roadmap, critics say lawmakers risk repeating the same mistakes.

Tech industry groups were largely supportive of the road map. TechNet, whose members include OpenAI, Google and Meta, said in a statement that the directive “will strengthen America’s global competitiveness in AI and emerging technologies” through providing $32 billion for AI research and development, which will be distributed to the Energy Department, Commerce Department, National Science Foundation, and National Institute of Standards and Technology.

Tony Samp, the head of AI policy at the DLA Piper law firm and former founding director of the Senate AI working group, said the Senate’s process helped raise awareness about AI among lawmakers and helped prompt ideas for how to legislate. DLA Piper, including Samp, has lobbied on behalf of clients including OpenAI, according to federal disclosures.

“There are some who think you can wave a magic wand and Congress could pass comprehensive AI legislation, but that thinking ignores the political realities in the United States, as well as the real objectives of the AI Insight Forums and the road map,” Samp told The Washington Post.

Reggie Babin, a senior counsel at the law firm Akin and a former chief counsel to Schumer, told The Post that the working group succeeded in its goals. Babin has lobbied on behalf of Akin clients, including Adobe, according to federal disclosures.

“The goal of the process was to figure out how to make sure that 80 percent of stakeholders in the middle of this conversation are satisfied while preserving space for continued engagement on all fronts,” he said. “I think the working group hit that mark.”

The lawmakers gathered input for the road map in private sessions dubbed Insight Forums. Over the nine sessions, lawmakers met with executives including OpenAI CEO Sam Altman and Tesla CEO Elon Musk, as well as consumer advocates and civil rights leaders, such as Maya Wiley, the president and CEO of the Leadership Conference on Civil and Human Rights, a Washington-based group that represents a number of civil rights organizations. Liz Shuler, the president of the AFL-CIO who participated in the Insight Forums, called the road map an “important start” to passing AI legislation.

But civil society leaders were frustrated that the road map only made a cursory mention of AI bias, amid widespread concerns that the technology can replicate and exacerbate harmful stereotypes. Nik Marda, a technical lead on AI governance at Mozilla, noted on X that “bias” was mentioned in the road map as many times as “space debris.”

Rashad Robinson, the president of the civil rights group Color of Change, said the report shows Schumer “is not taking AI seriously.” Robinson called for lawmakers to move swiftly to respond to the bias that AI can pose.

Schumer told The Post in a statement that he shares the goals of the advocates.

“Leader Schumer agrees with their goals and we’re going to continue to work closely with them as legislation is written,” said Schumer spokeswoman Allison Biasotti.

Greer said the report reads like industry had outsize influence over the process, and it was written by Altman and other tech lobbyists.

“They heard from experts about the urgency of addressing AI harms and then paid lip service to that while giving industry most of what they want: money and ‘light touch’ regulatory proposals,” Greer told The Post.

The United States’ efforts to regulate AI lag far behind those of the European Union, which last year advanced a wide-ranging AI Act that sets limits on AI based on how risky regulators deem an application to be. The E.U. AI Act, for instance, bans social scoring systems similar to those used in China, and it places transparency requirements on high-risk applications of AI in medical devices or employment settings.

Many observers hoped that the road map would provide clarity on a path forward for Congress to address some of the thorniest issues in AI governance that have divided the tech industry — including the future of copyright law and the growing debate over the regulation of AI models that are open source, or freely available to the public, without the guardrails that OpenAI, Google and Microsoft place on their models. But Schumer’s report doesn’t mention open source, and it largely punts issues of intellectual property rights to government agencies, directing lawmakers to review existing and forthcoming reports from the U.S. Copyright Office and the U.S. Patent and Trademark Office on AI.

In a briefing with reporters Tuesday night, Sen. Todd Young (R-Ind.), a member of the AI Gang, said the report had “a high level of specificity” for a document of its size and scope. But senators acknowledged the report leaves key questions unanswered. The report calls for a $32 billion investment in nondefense AI research and development, but it does not specify how much funding should be directed to the military. The report also leaves key questions about how Congress should regulate consumer privacy in the age of AI to the committees.

“Where vagueness was required to come to an agreement, we embrace vagueness,” Young said.

But that strategy doesn’t work, said Suresh Venkatasubramanian, a Brown University professor who co-authored the White House’s AI Bill of Rights. He participated in the forums and felt as if lawmakers ignored the problems raised during the sessions.

“Embracing vagueness at this point is basically saying the status quo is just fine," he said.

The report “repackages” many issues around AI that have been debated on Capitol Hill for years, and its delivery may prompt some legislators to “yearn for more than just polished reiterations,” said Divyansh Kaushik, a vice president at Beacon Global Strategies, which advises companies on national security issues.

“After almost a year of hearings, briefings and forums, I think members are hungry for tangible, actionable steps and crisp legislative blueprints, something beyond the theoretical … a call for concrete, actionable strategies,” he said. “The real challenge begins now and it’s one for congressional committees: ensuring that this report does more than stir the pot, but rather sparks a sustained drive toward innovative and decisive policymaking to ensure American competitiveness on these critical technologies."