Good morning. The U.S. presidential election is nearly upon us and campaign ads have reached a fever pitch. (That is, unless you’re in one of the seven swing states. In which case, sorry.)
There’s an awful lot of tech money fueling those ads: crypto money, VC money, Musk money. But in tech, we love cold, hard data. So: Do political ads actually work?
According to 2021 research from Northwestern Kellogg, they do. Positive ads stimulate voter turnout generally, but negative ads—the kinds that typically run closest to election day—boost voter turnout for the favored candidate, even as they suppress overall turnout.
The catch? Those voters might turn out in places where the race isn’t close enough to matter. But where it does, well… —Andrew Nusca
Before we go to more news…request your invitation to Fortune’s premier event, the Fortune Global Forum, convening in New York City Nov. 11-12. Fortune Global 500 executives and international policy leaders will discuss managing workforces in an AI-powered economy, climate and energy issues, global trade, and the future of democracy.
Want to send thoughts or suggestions to Data Sheet? Drop a line here.
White House issues national security memo on AI
The Biden administration on Thursday issued a comprehensive artificial intelligence memo—an unprecedented call to arms on “era-defining technology” in the context of national security as well as global competitiveness.
The memo called for necessary guardrails and warned that AI misuse could “bolster authoritarianism worldwide, undermine democratic institutions and processes, facilitate human rights abuses, and weaken the rules-based international order.”
Though the current era of AI development relies on the expertise and computational might of the private sector, the directive reads, the U.S. government “must urgently consider” how to join the fray on national security grounds. If the government doesn’t act quickly enough—and partner with academia, civil society, and industry—it risks losing ground to competitors, frustrating its foreign policy objectives, and harming democratic norms as we know them.
The memo is organized around two principal needs: talent and computational power. On the first, “advancing the lawful ability of noncitizens highly skilled in AI and related fields to enter and work in the United States” is now a priority, and the departments of Defense and Homeland Security are tasked with expediting the arrival of AI semiconductor and software experts to U.S. soil.
On the second, the departments of Defense and Energy have been tasked with considering “the applicability of large-scale AI” to projects involving computational facilities, plus facilitating approvals and incentives for related infrastructure. Meanwhile the National Science Foundation is ordered to distribute “resources, data, and other critical assets for AI development” to the universities, nonprofits, and researchers who otherwise wouldn’t have access.
The document is certainly comprehensive. Will it spur enough action? According to White House officials, the lack of policy and legal clarity around AI has held back experimentation and adoption; this new framework tries to fix that. —Jenn Brice, Andrew Nusca
Surge pricing for…eggs?
Grocery giants Kroger and Walmart deny adding digital price tags to their store shelves as a way to rapidly hike food prices, according to the New York Times.
The comments followed inquiries from Senators Elizabeth Warren (D-Mass.) and Bob Casey (D-PA), as well as Rep. Rashida Tlaib (D-Mich.), about the new technology, which some feared would set the stage for Uber-like “surge pricing.”
Kroger dismissed the idea, saying digital price tags, in fact, save employees time and are routinely used during sales, or when products near expiration dates, rather than for gouging customers.
In addition to digital price tags, store officials denied using facial-recognition technologies to identify shoppers and adjust prices based on who they are.
Coming off of several years of rapid inflation, consumer fears about surge pricing are understandable. But if grocers ever did try to follow in Uber’s footsteps, you can bet there would be plenty of blowback. —Jason Del Rey
Intel wins $1.2 billion appeal over antitrust fine
Floundering Intel can catch a break after all.
The EU’s highest court, the Court of Justice, yesterday backed the overturning of a $1.15 billion fine that antitrust enforcers issued all the way back in 2009.
The EU’s General Court ruled in 2022 that the Commission had mostly blown its analysis of the case, which revolved around the secretive rebates that Intel gave PC makers like Dell for shunning AMD’s processors in their products.
The money has been sitting in escrow for the last 15 years and Intel will now get most of it back. But not all of it, as the General Court did confirm that it had imposed some illegal sales restrictions, and the Commission then issued a narrower $406 million fine on that basis.
It will take years for that separate appeal to play out. In the meantime, anything that can help Intel keep its head above water will surely be most welcome. —David Meyer
Another OpenAI safety exec departs
OpenAI has been hit with a reputational double-whammy.
Miles Brundage, the guy who’s been in charge of “artificial general intelligence” safety, has quit to pursue independent safety research. His team is being redistributed throughout the company.
Brundage didn’t exactly slam OpenAI on the way out, but he did strongly suggest that it’s been cutting safety corners in a race to commercialize its tech. He’s the latest in a long line of AI safety folks to head for the exits at OpenAI.
Meanwhile, former OpenAI researcher Suchir Balaji has gone public with his claims that the company broke copyright law when training its models. This is ammunition for the many news publishers and authors that have sued OpenAI for copyright infringement.
OpenAI insists that it hasn’t broken the law—and says it’s “excited” to follow the “impact” of Brundage’s future research. —DM
Intel + Nvidia, the merger that might have changed history
From the department of "What ifs," the New York Times reported Thursday that Intel once considered acquiring Nvidia for what would have been a fraction of the AI chipmaker's current $3.4 trillion valuation.
The idea for a potential deal was hatched in 2005 by Intel's then CEO Paul Otellini. At the time, AI wasn't on anyone's mind. Nvidia's bread-and-butter was graphics chips for video gaming, and the company was just beginning to explore using GPUs for scientific research.
Otellini went to Intel's board and proposed acquiring Nvidia for $20 billion, but the board rebuffed him and nothing more ever came of it, according to the Times.
Would Intel be the king of AI today, rather than struggling to stay alive, had Otellini gotten the greenlight? Impossible to say, but there's an argument to be made that it would not. For one thing, Intel had its own in-house graphics effort at the time, which it killed a few years later.
Rather than building Nvidia’s GPUs into the computing champs they are today, Intel may well have stifled their development in order to maintain the primacy of its PC microprocessor business. —Alexei Oreskovic
More data
—Bluesky now has more than 13 million users. To celebrate, it raised $15 million.
—Are M4 Macs coming on Monday? The Apple rumor mill turns.
—TSMC production yields in Arizona exceed those in Taiwan. Not too shabby for the oft-delayed chipmaking plant.
—Baby Boomer politicians use TikTok to reach Gen Z. The algorithm’s allergy to echo chambers is an asset.
—The price of chatbots is too damn high. Hugging Face believes it has a solution.
Endstop triggered
