Have your friends signed up?
Okay, let's start with ...
Illustration: Sarah Grillo/Axios
In New York, official scrutiny is triggering doubts about the future of a much-publicized $3 billion concession to Amazon to build a new headquarters employing tens of thousands. And in Wisconsin, Foxconn has created uncertainty about whether it will fulfill promises to build a huge new factory there employing 13,000 people, Erica reports.
The study — an examination of 164 deals in Texas since 2003, conducted by Nate Jensen, a professor at the University of Texas at Austin, and Calvin Thrall, a graduate student — found that about a quarter of the companies walked back on promised jobs after signing.
"It's only when you have these rare public auctions that the mask gets ripped off. It's just too easy. There's too much money sloshing around, and it's too easy to qualify for it."— Greg LeRoy, head of Good Jobs First, a nonprofit that tracks incentives
The Amazon HQ2 search threw the curtains open. Critics say too much was granted in a deal that could create havoc in and around Long Island City, including the displacement of longtime residents by skyrocketing rents.
The examples of companies failing to deliver appear to be rife.
Milwaukee has the highest-profile case: Foxconn got one of the biggest economic development grants in memory: $4.5 billion for 13,000 jobs, the rough equivalent of Wisconsin paying 30% of Foxconn's payroll for 15 years, says Tim Bartik, a senior economist at the Upjohn Institute. But over the last week, Foxconn has thrown the project into a whirlwind of uncertainty.
Evergreen Solar, a solar panel manufacturer, received a $58 million grant commitment to build a plant in Massachusetts, but it eventually scrapped construction plans. The state recovered just $3 million of $21 million Evergreen had already received.
MIT researchers are using a wooden Jenga tower in a new effort to accomplish one of the hardest challenges in robotics — to build a bot that can grab, pack and assemble things with the dexterity of a human hand.
Kaveh writes: The MIT robot arm brings bots closer to assembling or packing finicky objects in a factory — jobs that, for now, can only be done by people.
The robot can gingerly poke out block after block from the tower, relying on feedback from a camera and — in a novel twist — its own sense of touch.
Jenga is "a little like chess or Go but for manipulation," says Ken Goldberg, a Berkeley roboticist who was not involved with the MIT research. "It really requires a pretty sophisticated level of skill."
Details: When the MIT robot played its first game, it attacked blocks at random. But it learned as it played, and 300 fallen towers later, it developed a system to predict how blocks in different parts of the tower would behave.
The bot is pretty good at the game, says Fazeli — but not world class. "If you have a little skill, you can definitely beat it."
What's next: A yet-unsolved next step with enormous commercial potential, says Goldberg, is putting things back into narrow gaps.
Photo: Jaap Arriens/NurPhoto/Getty
As reported above, robots are perfecting skills meant for factories and warehouses — but they’re targeting desk jobs, too, Erica writes.
About 1/3 of the stories published on Bloomberg are written with some degree of AI help, reports NYT.
In addition to markets news, another big area of bot reporting is sports. Per NYT, the AP uses AI for minor league baseball stories and the Washington Post for high school football recaps.
Here are examples of machine-generated articles from the AP:
Illustration: Aïda Amer/Axios
Pick your poison. Photo: Maria Alejandra Cardona/Los Angeles Times/Getty
A small cannabis company in Vancouver called Weekend Unlimited has gotten its hands on the best stock ticker symbol for a business of its kind: POT.
Erica writes: The symbol used to belong to Potash Corp., which, as it sounds, produced potash, Quartz reports. When Potash merged with another company, the symbol was suddenly up for grabs.