When Dario Amodei Excited by artificial intelligence – which is almost always – movement. Co -founder and general director from sitting in the conference room and throws himself on the board. Charts with hockey -falling curves run, which show how machine intelligence bends towards infinity. His hand rises to curly hair mop, as if caressing his neurons to prevent a system disaster. You can almost feel how his bones vibrate when he explains how his company, Anthropic, is different than other builders of AI models. He tries to create artificial general intelligence – or as he calls, “powerful AI” – which will never become dishonest. He will be a good guy, it will decide utopia. And although Amodei is necessary for Anthropic, it ranks second Very An crucial colleague. Like other unusual beings (Beyoncé, Cher, Pelé), the latter has one name, in this case for pedestrians, reflecting her maiden and comic. Oh, and this is the AI model. Hi, Claude!
Amodei has just returned from Davos, where he spread the flames on Fireside chats, declaring that in two years Claude and his peers would exceed people in each cognitive task. He hardly recovered after the journey, he and Claude are now dealing with an unexpected crisis. The Chinese company called Deepseek has just released the most newfangled enormous language model, which allegedly built for a fraction of what companies such as Google, Opeli and Anthropic. The current paradigm of the most newfangled artificial intelligence, which consists of expenses worth many billion dollars on equipment and energy, suddenly seemed trembling.
Amodei is probably a person most associated with the maximalistic approach of these companies. When he worked at Openai, Amodei wrote an internal article about something that he wondered over the years: the hypothesis entitled The Large Blob of Compute. AI architects, of course, knew that the more data you had, the more powerful your models could be. Amodei suggested that this information may be more harsh than they assumed; If they fed megatons of things to their models, they could accelerate the arrival of powerful AI. Theory is now a standard practice and this is the reason why leading models are so costly to build. Only a few companies from a deep pocket could compete.
Now the novice Deepseek – from a country subject to the export of control of the most powerful tokens – fought without a enormous stain. If powerful artificial intelligence can come from anywhere, perhaps anthropic, and its peers were computing emperors without moat. But Amodei explains that Deepseek doesn’t keep him at night. He rejects the idea that more competent models will allow low budget competitors to jump to the front of the line. “That’s the opposite!” says. “The value of what you create is growing. If you get more intelligence for a dollar, you can spend even more dollars on intelligence!” He claims that it is much more crucial than saving money to reach the Aga finish line. That is why even after Deepeek, companies such as OpenAI and Microsoft have announced plans to spend hundreds of billion dollars more on data centers and power plants.
What Amodei is obsessed is the way people can safely reach Aga. This is such a hairy question that he forced him to six other founders of Anthropica to leave OpenAi, because they felt that he could not be solved at the director of Sam Altman at the forefront. In Anthropic they are in a sprint to determine global standards for all future AI models, so that they actually assist people instead of in one way or another, blowing them up. The team is hoping to prove that they can build agi so protected, so ethical and so effective that his competitors see wisdom in following. Amodei calls this race to the top.
This is where Claude appears. Go around around the anthropic office, and soon you will notice that the mission would be impossible without it. You never fall into Claude in a cafe sitting in the conference room or ride on the elevator to one of the 10 floors of the company. But Claude is everywhere and exists from the early days when anthropic engineers first trained it, raised it and then used better Claudes for production. If Amodei’s dream comes true, Claude will be both our wing and fairy -tale model when we enter the era of abundance. But here is the Trippy question, suggested by the company’s own research: can Claude be trusted that it will have fun?
One of Amodea Anthropic co -founders are nobody other than his sister. In the 70s, their parents, Elena Engel and Riccardo Amodei, moved from Italy to San Francisco. Dario was born in 1983, and Daniel four years later. Riccardo, a leather craftsman from a tiny city near the island of Elba, got ill when the children were tiny and deceased when there were newborn adults. Their mother, a Jewish American born in Chicago, worked as a project manager for libraries.