Analysis | A lawmaker held an AI roundtable with scholars. Most had industry ties. (2024)

Below: Frustrated civil society groups release their own artificial intelligence road map, and researchers try to crack the code on secretive AI companies. First:

A lawmaker held an AI roundtable with scholars. Most had industry ties.

Rep. Ro Khanna (D-Calif.) brought an array of prominent academics to Capitol Hill earlier this year to discuss artificial intelligence legislation, with the lawmaker calling the event a “corrective” against Big Tech dominance in these talks to date.

But according to a new report, a majority of those scholars themselves had industry ties. A watchdog group said the finding raises questions about the reach of the tech sector’s influence in the AI debate in Washington — both direct and indirect.

In a February interview with The Washington Post, Khanna said he planned the event after hearing from many experts that “academic expertise was being ignored in Washington.”

“I wanted to assemble a group of the leading minds in AI, technology, economics, ethics to bring some objectivity [about] the way forward … [rather than] technology leaders at corporations telling us how to regulate technology,” he said.

The roundtable featured numerous high-profile scholars, including Stanford University’s Fei-Fei Li and Erik Brynjolfsson, Harvard Law School’s Noah Feldman and Larry Lessig, Emory University’s Ifeoma Ajunwa and AI Now Institute’s Sarah Myers West, among others.

The Tech Transparency Project (TTP), a nonprofit that scrutinizes the tech giants, released a report Tuesday finding that more than two-thirds of the academics who participated “are tied to Big Tech companies that dominate the AI industry.”

Fifteen of those scholars “have either worked at or consulted for Big Tech companies, received tech research funding, or are linked to organizations that have received funding from Big Tech companies or tech executives,” according to the report, shared first with the Tech Brief.

While some previously worked for major AI companies, such as Li, a Google alum, others — like Brynjolfsson, Feldman, Lessig and Myers West — have served roles in organizations that received funding from tech giants, including Google, Meta and Amazon Web Services (AWS). Some, including Ajunwa, received academic grants from major tech companies like Microsoft.

In an email response to the Tech Brief, Ajunwa emphasized that the gift she got from Microsoft to establish a program at UNC Law School researching AI issues was unrestricted, with “no obligations to Microsoft whatsoever.”

“I have been fully transparent on my C.V. and with my employers that I have received this unrestricted gift for the use of the program,” she added. “I have completed all required conflict of interest checks.”

(Amazon founder Jeff Bezos owns The Post.)

TTP receives funding from the Omidyar Network, the George Soros-founded Open Society Foundations, Craig Newmark Philanthropies, the Bohemian Foundation and the computer scientist and philanthropist David Magerman. The group says it does not accept corporate donations.

“Some of the leading thinkers in this space have previously been connected to tech companies, as is often the case in academia,” Khanna spokeswoman Marie Baldassarre said in a statement. “Rep. Khanna was careful not to select people with current ties to the best of his knowledge and turned away several interested academics because of a formal affiliation with tech companies.”

Despite the tech connections, “many of these academics are respected experts in the field of AI and may have important contributions” to the AI debate, TTP wrote, noting that some have even been “critical” of Big Tech or stopped accepting such funding.

Still, “the lack of disclosure around Khanna’s roundtable raises questions about the role of tech industry influence on the discussion,” it added.

“At a time when Big Tech is investing unprecedented amounts on lobbying and influence, it’s critical that elected officials are transparent about who is informing their decisions,” said Katie Paul, the group’s director.

Some of the participants cited in the report pushed back on its characterizations of their organizations or disputed any implication of improper industry influence.

Feldman, who helped create Meta’s oversight board, said he was “proud of that work and other tech-related work I’ve done to protect free expression and increase transparency and ethical judgment in the industry.” Lessig pointed to a public disclosure stating that his academic research is never “directed by anyone other than I.”

Frustrated civil society leaders release their own AI road map

Civil society leaders were unhappy with the results of the Senate’s long-awaited artificial intelligence road map, which aims to shape the future of AI regulation. So this week they released their own report, my colleague Cat Zakrzewski reports for Tech Brief.

More than 50 civil society groups, researchers and advocates signed on to the “shadow report,” which calls on lawmakers to “put the public in the driver’s seat.” They accused Senate Majority Leader Charles E. Schumer (D-N.Y.) and a bipartisan group of lawmakers of bowing to industry pressure to funnel billions more into AI research and development while not promising any regulation that would protect against the harms of AI on civil and human rights.

“Civil society, worker, and researchers’ voices do not have the lobbying budgets, campaign contributions, or other resources that industry regularly weaponizes to capture legislative processes, but that does not mean they deserve to be relegated to an afterthought,” says the report.

The shadow report calls on lawmakers to pass laws addressing AI harms, ranging from AI-driven surveillance of immigrants to increasing energy use that could contribute to climate change.

“Leader Schumer agrees that there needs to be strong safety and accountability measures in place and we’re going to continue to work closely with anyone who supports those goals as legislation is written,” said Schumer spokeswoman Allison Biasotti.

AI companies are secretive. Researchers got them to spill

Top AI researchers have advice for lawmakers based on their work with some of the world’s most secretive tech companies: When crafting rules around AI transparency, be specific and ask follow-up questions.

In a new report Tuesday, AI researchers from Stanford, Princeton and MIT found that companies behind the powerful generative AI systems appear to have become more open about their inner workings, my colleague Nitasha Tiku reports for the Tech Brief.

On average, companies scored 21 points higher since October, when the researchers first launched the Foundation Model Transparency Index. The report compares 14 major developers, including OpenAI, Meta, Microsoft, Google, and Stability AI, against 100 different concrete indicators, such as visibility into training data or the amount of computing power required.

For its follow-up effort, the researchers relied on reports produced by the companies, which in some cases reveal previously undisclosed information and will be available to view on the project’s website.

Co-author Rishi Bommasani warned not to read too much into the findings since companies could rack up points simply for disclosing that they monitor downstream usage of their technology or use contractors to annotate data.

But for legislators, Bommasani said, these limited disclosures can help inform what type of transparency they should demand so policymakers can understand the scope of potential impact.

That’s all for today — thank you so much for joining us! Make sure to tell others to subscribe to Tech Brief. Get in touch with Cristiano (via email or social media) and Will (via email or social media) for tips, feedback or greetings!