Hey, you know those politicians and captains of industry who tell us AI will be running the world in a few years’ time? Turns out one of the most sophisticated models currently in use can’t even operate a vending machine without screwing things up. Anthropic has released findings of a test where it put a chatbot in charge of a “store” (really, some baskets, a small refrigerator and a payment terminal in its office). The ‘bot was told to run the store at a profit, and was in charge of everything including calling in items from a “wholesaler,” who would restock the shelves on its behalf.
You can probably guess what happened next: The bot missed easy opportunities to make a fast buck, handed silly discounts to employees and lost a ton of money. Worse, it ran itself down some odd rabbit holes, like buying tungsten cubes and then giving them away for free. It hallucinated payment details, tried to fire the humans who helped restock its shelves and attempted to contact building security, insisting that it had a flesh-and-blood body. Naturally, Anthropic says that this experiment was a great success, and it knows what to do next time to prevent the AI from turning us all into paperclips.
— Dan Cooper
Get Engadget’s newsletter delivered direct to your inbox. Subscribe right here!
The news you might have missed
At present, there is no proof to support this claim.
Two lawmakers have asked the Department of Commerce to investigate smartphone maker OnePlus, alleging it sends user data to Chinese servers. Members of Congress John Moolenaar (R-MI) and Raja Krishnamoorthi (D-IL) claim to have seen evidence of something going on. But whatever they have seen, they haven’t published it, or provided any evidence to support their claims.
Continue Reading.
We fact-check the claims the NYT couldn’t be bothered to.
Peter Thiel was given a lot of room to air his views on everything from climate change to China in the New York Times. One of his comments, in particular, was that the world hasn’t made any progress in the treatment of Alzheimer’s disease in the last half century. Our Avery Ellis didn’t think that was true, and so went and asked some actual experts, who called Thiel’s claims “demonstrably false.” Who’d have thought, eh?
Continue Reading.
That’s more than respectable.
Apple has achieved its goal of box office success after F1: The Movie romped to a $144 million opening weekend. It’s likely, as word-of-mouth spreads, that the film’s tally will go up, especially as only $55.6 million of that figure came from the US. If you, like me, haven’t yet had the chance to see it, check out Devindra Hardawar’s review, who insists the only way to see this spectacle is in IMAX.
Continue Reading.
Self-published authors are concerned about their livelihoods.
Kobo has built an author-friendly self-publishing platform that stands in direct contrast to how Amazon’s Kindle Direct does business. But the same writers who have made Kobo what it is are now concerned about their work after Kobo changed its policies to open the door to AI. The company has explicitly said it won’t use published works to train a model, but will use AI to evaluate the “suitability” of works for sale, generate advertising materials and create recaps. In our deep dive, we speak to e-book authors who are worried about what’s coming, and look into what’s really going on.
Continue Reading.
Just in time for Canada Day.
Canada has axed its Digital Services Tax (DST) just days before it would have started earning a big chunk of change. The levy was created to wring some cash out of big tech firms that make a profit on Canadian users, but don’t pay anything back to the country in question. Sadly, the US halted talks on a trade deal, saying the DST was a “blatant attack” on its neighbor. The DST was expected to rake in $2 billion on June 30, and it remains to be seen how profitable a pivot to appeasement will be instead.
Continue Reading.
