If the anthropic succeeds, a country can be born of Benevolent AI genius


When Dario Amodei Glad about AI – which is almost always – he moves. Cofounder and CEO sprout from a chair in a conference room and darts to a whiteboard. She scrawl charts with swooping hockey-stick curves showing how to bend intelligence intelligence towards eternity. His hand was rising in his curly hair mop, as if he were caressing his neurons to focus on a system crash. You almost feel his bones vibrating as he explains how his company, anthropo, is Unlike other AI model builders. He tries to create an artificial general intelligence – or as he calls it, “powerful AI” – it will never be a rogue. It will be a good person, a Usher of Utopia. And while Amodei is ancient to anthropic, he comes to second in the company Most It is important to contribute. Like other extraordinary creatures (Beyoncé, Cher, Pelé), the latter passes through a single name, in this case a pedestrian, reflecting its pliancy and its office. Oh, and it's an AI model. Hello, Claude!

Amodei just returned from Davos, where he entertained fires in fireside chats by declaring two or more years Claude and his peers would exceed the people in each of the work -free work. It's hard to recover from the trip, he and Claude are now talking to an unexpected crisis. A Chinese company called Deptseek has just released a state-of-the-art large language model purportedly built for a portion of what companies such as Google, Openai, and Anthropic have spent. The current AI cutting paradigm, consisting of multibillion-dollar expenditures to hardware and energy, suddenly seems trembling.

Amodei is probably the person most relates to the approach of these maximalist companies. Back when he worked in Openai, Amodei wrote an internal paper on something he wanted for many years: a hypothesis called the Big Blob of Compute. AI architects know, of course, that the more data you have, the stronger your models are. Amodei suggested that that information could be more raw than they assumed; If they feed the megatons of things in their models, they can facilitate the arrival of strong AI. Today's theory is usually practice, and this is why top models are so expensive to build. Only a few of the deeply pocketed companies can compete.

Now a newcomer, Deepseek– From a country subject to export controls to the strongest chips – is waltzed to no big Blob. If the powerful AI may originated everywherePerhaps the anthropical and its peers are computational emperors that no moats. But Amodei made it clear that Deepseek did not keep her at night. He rejects the idea that better models will allow competitors to the low budget to jump in front of the line. “It's just the opposite!” he said. “The amount of what you do is climb up. If you get more intelligence every dollar, you might want to spend more dollars in intelligence!” More important than saving money, he argued, reached the AGI end line. That is why, even after Deepseek, companies such as Openai and Microsoft have announced plans to spend the road -billions of dollars more than data centers and power plants.

What Amodei is doing is how people can reach AGI safely. It was a question so furry that forced him and six other anthropic founders to leave Openai in the first place, as they thought it would not be solved with CEO Sam Altman at the helmet. In the anthropic, they in a sprint To set global standards for all future AI models, so that they can really help people instead of, one way or another, explode. The team hopes to prove that it can build an agi so it's safe, so ethical, and its competitors can effectively see the wisdom in keeping the suit. Amodei calls it the breed at the top.

That is where Claude enters. Stay around the anthropical office and soon you will observe that the mission is impossible without it. You never run to Claude at the café, sitting in the conference room, or ride in the elevator on one of the company's 10 floors. But Claude was everywhere and since the early days, when anthropical engineers first trained it, it raised it, and then used it to produce better claudes. If Amodei's dream came true, Claude would be both our wing model and fairy godmodel as we entered an abundance. But here's a trippy question, proposed by the company's own research: Can Claude be trusted to play Nice?

One in Amodei's Anthropic cofounders are none other than her sister. In the 1970s, their parents, Elena Engel and Riccardo Amodei, moved from Italy to San Francisco. Darius was born in 1983 and Daniela four years later. Riccardo, a leather manufacturer from a small town near the island of Elba, became ill when children were small and died when they were young. Their mother, an American American born in Chicago, worked as a project manager for libraries.

Leave a Reply

Your email address will not be published. Required fields are marked *