Technology

➚ MOVE FAST: The Chosen. Investors who want to be part of OpenAI’s $6.5 billion fundraising round are finding out today if their funds have been accepted. The bid to raise money is oversubscribed, so Sam Altman’s firm gets to be picky about who gets to back the venture, which will be valued at $150 billion.

➘ BREAK THINGS: The black sheep. Brazil’s Supreme Court has imposed a $920,000 daily fine on Elon Musk’s X for allegedly trying to skirt a ban on the platform. It’s the latest twist in a weekslong drama that has also dragged in one of Musk’s other companies, SpaceX and its Starlink service.

No need to poll humans, bots will do

THE SCOOP

In a closely watched New York Democratic primary in June, centrist George Latimer ousted incumbent Jamaal Bowman by a wide margin of 58.7% to 41.3%.

Ahead of the vote, two 19-year-old college dropouts in Manhattan conducted a poll that accurately predicted the results within 371 votes. Their secret? They didn’t survey a single person. Instead, they asked thousands of AI chatbots which candidate they preferred.

Welcome to the future of polling, according to Cam Fink and Ned Koh, co-founders of a seven-person company called Aaru. They say they’ve cracked the code for predicting accurate election results, which have come under increasing fire since most public polls failed to predict Donald Trump’s victory in 2016. The answer is ignoring the humans whose behavior they are trying to capture.

For election results, Aaru uses census data to replicate voter districts, creating AI agents essentially programmed to think like the voters they are copying. Each agent is given hundreds of personality traits, from their aspirations to their family relationships. The agents are constantly surfing the internet and gathering information meant to mimic the media diets of the humans they’re replicating, which sometimes causes them to change their voting preferences.

For instance, when Donald Trump was shot during an attempted assassination, a large chunk of Aaru’s agents immediately switched voting preferences to support the former president. But as more information came out about the shooter in the hours after the attack, many of them switched back.

The polls usually draw on responses from around 5,000 AI respondents, and it takes anywhere from 30 seconds to 1.5 minutes to conduct. Aaru charges less than 1/10th the cost of a survey of humans.

“No traditional poll will exist by the time the next general election occurs,” Fink said in an interview with Semafor. “There are massive issues when you’re using real people. You never know if someone is telling the truth.”

He said the company has been hired to conduct polls for Fortune 500 companies, political campaigns, think tanks and super political action committees. One campaign in California is relying mainly on Aaru for its polling, he said.

In one survey, the company noticed that one of the AI agents said it was going to vote for Mickey Mouse in the upcoming presidential election. Fearing one of their bots had gone off the rails, the Aaru team investigated. It turned out the bot had an explanation. “The agent’s response was ‘I hate Kamala and I hate Trump. I’m writing in and voting for Mickey Mouse,’” Fink said.

The teaming up of Microsoft, BlackRock and Abu Dhabi’s MGX to form a $30 billion AI infrastructure fund got me thinking that this is all going to end up in the hands of Uncle Sam.

Once the scale gets to a certain size, the US government will have to bring it all under one roof (physically, maybe two or three roofs), controlled by the Defense Department, given the critical nature of the infrastructure.

In this scenario, multiple tech companies would work together on one model, which will be top secret. Once they’ve reached the holy grail of artificial general intelligence, the firms will be allowed to use distilled, less powerful “aligned” versions of the model for commercial purposes and it will be like a regulated utility.

That’s because the US isn’t going to allow AGI to be built in Saudi Arabia, the UAE, or other countries. But those nations can get in on these deals in the US, which would essentially be a concession to keep them from embracing China. The oil money would help finance the massive infrastructure projects and the US government would make them profitable.

So to me, the Microsoft-BlackRock-MGX partnership is akin to government contracting. And taxpayers are the ultimate customers for these mega infrastructure projects. The alternative is for the US government to leave all this up to the private sector to sort out. The national security stakes are too high for that.

Nvidia and the UAE’s tech giant G42 are getting cozier. The two companies announced Thursday that they are building a new Climate Tech Lab in Abu Dhabi and collaborating on new weather prediction models that can get as precise as a square kilometer. Weather predictions using deep learning instead of physical simulations are kind of a hot area. Check out our story on a startup that wants to put tens of thousands of weather balloons into the sky to make the most accurate weather prediction system in the world.

But the announcement Thursday is also significant in the context of US chip exports to the Gulf region. As we scooped last week, G42 got a sizable shipment of Nvidia H100s and had to jump through a bunch of national security hoops to convince the Commerce Department that China wouldn’t benefit.

I heard about this new directory created by Stanford adjunct professor Steve Blank that lists points of contact within the department of defense. Creating phone books isn’t usually what Stanford professors do, so I asked him why.

• AI startup aimed at demystifying the world of public policy secures new funding.
• The late GOP push to deny Kamala Harris a Nebraska electoral vote.
• CERN to expel hundreds of Russian scientists.