Google wants governments to form a ‘global AI corps’

In a new white paper unveiled Tuesday, the tech giant calls on governments to beef up their AI training and skilling initiatives to help develop a “global AI corps.” 

The paper urges governments to take concrete steps to achieve that goal — such as expanding certificate and skilling programs and building out local research centers — while calling on policymakers to focus more on channeling AI’s potential benefits.

“Responsibility and opportunity are two sides of the same coin. We need to get both right,” Kent Walker, Google’s president of global affairs, said in an interview Monday.

The paper, penned by Walker, calls on governments to “scale up AI training programs” to “catch workers that are impacted by AI and reskill them so they can quickly bounce back into new and better jobs.” It also urges officials to create “flexible immigration pathways for AI experts.”

“We want to learn the lessons of globalization, we want to minimize disruption and [ask] how do we do broad-based skilling for our workers so that they can thrive,” Walker said.

The agenda is likely to guide how the tech giant, one of Silicon Valley’s most powerful companies, will look to shape regulatory talks as they gain steam in Washington.

In the paper, Google throws its support behind an idea floated by Senate Commerce Committee Chair Maria Cantwell (D-Wash.) to create a new “GI Bill” for AI — a reference to the World War II-era measure that still provides education and skilling benefits for veterans.

“Instead of a GI Bill, we need an AI education bill,” Cantwell said at an August forum. She added that she wants to see “at least 1 million people retrained and skilled, particularly in apprentice programs.”

The company earlier this year released a separate set of recommendations that focused more on tackling potential harms caused by AI, an area that’s more likely to bring Google and other tech giants into tension with policymakers as they look to set new guardrails for the tools.

The white paper arrives as Senate lawmakers dial up early discussions about how to boost the development of AI tools while setting guardrails around their use. 

And it comes on the heels of President Biden’s sweeping AI executive order, which looks to guide the federal government’s deployment of the tools while setting some new checks on their use in the private sector. 

For Google’s Walker, the executive order has some “very promising elements,” such as its calls for more research to be done through the National AI Research Resource and easier pathways for immigrants with AI expertise to work and study in the United States. 

But there are still key questions, he said, about how federal agencies will write and implement their own guidelines around their use of the technology, as tasked under the executive order. 

“Part of this white paper is to encourage the agencies to think about the fact that the biggest responsibility they may have is to get AI right, making sure they’re not missing the boat in terms of using it in really constructive ways that would make them more efficient,” Walker said.

Google pays Apple 36 percent of the revenue it makes from search advertising transacted through Apple’s Safari browser, Bloomberg News’s Leah Nylen reports, citing Monday testimony from the main economics expert Google hired to defend itself.

University of Chicago professor Kevin Murphy disclosed the number in testimony at the Google antitrust trial in Washington. The Justice Department has alleged that the tech giant illegally maintains dominance in the online search market, hurting competitors and affecting consumer choice as it secured deals to make its popular search engine the default on iPhones and other Apple devices.

The companies have had a deal in place since 2002 that makes Google the default search offering on the Safari browser. Google declined to comment to Bloomberg News, while Apple did not immediately return a request for comment.

“John Schmidtlein, Google’s main litigator, visibly cringed when Murphy said the number, which was supposed to remain confidential,” Nylen writes. Google said in a court filing last week that publicly disclosing too much information about the deal “would unreasonably undermine Google’s competitive standing in relation to both competitors and other counterparties.”

The White House unveiled its “National Spectrum Strategy” on Monday, ordering more spectrum — or radio wave bandwidth — to be reserved in coming years for emerging technologies such as drones, autonomous cars, automated factories and satellites, our colleague Eva Dou reports for The Technology 202.

“Spectrum is a limited resource, and these new technologies are rapidly increasing demands upon it,” Lael Brainard, director of the National Economic Council, said in a press briefing.

The private sector has called for such a move for years, citing the massive amounts of data that need to be transferred for these next-generation consumer technologies. The issue of reassigning spectrum bandwidth has been contentious because all of the spectrum bands are already occupied, so to make room for newer functions, some older ones will have to be squeezed.

The administration said it is also working on a related R&D plan to help maintain U.S. technological leadership. “As a Nation, we must deepen our collective understanding of the electromagnetic (EM) spectrum,” the strategy says.

Tech nonprofit Responsible Innovation Labs (RIL) released a set of guidelines for artificial-intelligence start-ups today that were developed in consultation with the Commerce Department, as well as academics, other public-sector stakeholders, and leading AI companies including OpenAI and Anthropic, our colleague Nitasha Tiku reports for The Technology 202.

The tools aim to help smaller, early-stage companies that might use “frontier models” figure out how to interpret broad mandates like President Biden’s recent AI executive order, said Hemant Taneja, co-founder and chairman of Responsible Innovation Labs and managing partner of General Catalyst.

RIL previously announced voluntary AI commitments, but the fresh guidance provides off-the-shelf tools and details that start-ups can directly implement, including red-teaming models to test for flaws, model cards to document audits, risk frameworks and more. Thirty-five top venture firms, including General Catalyst, Felicis Ventures, Lux Capital and Insight Partners, have already signed on to the protocol.

“It’s going to take awhile before we understand how these technologies really ought to be regulated,” Taneja said. “But I can tell you, if you reflect back to when ChatGPT hit the zeitgeist, every CEO in every boardroom, of every company, in every industry, in every country was talking about how they use AI. So this stuff is going to happen fast.”

YouTube said Tuesday that it will allow people to upload AI-generated videos that realistically show an event that didn’t happen, or depict a person saying something they didn’t say, as long as the content is labeled and follows the platform’s other rules around not misleading users when it comes to sensitive topics like elections and wars, our colleague Gerrit De Vynck reports for The Technology 202.

The policy update comes as YouTube begins making its own generative AI tools available, including one to let people translate their voices into other languages and another to create backgrounds for videos using an AI image-generator. Tech companies are struggling to make rules to contain and control the AI images and video created by technology they themselves developed. 

That’s all for today — thank you so much for joining us! Make sure to tell others to subscribe to The Technology 202 here. Get in touch with tips, feedback or greetings on Twitter or email.