Boss of AI startup Anthropic hangs a poster of a robot destroying the world in his office

Don't show Elon Musk, or he will sue.

Anthropic, founded by former OpenAI engineers, has developed a competitive chatbot Claud and positions itself as a "super-responsible AI company" - overall, this statement could be believed if not for one detail that became known in a recent interview by the startup's CEO Dario Amodei to Time.

It turns out that one of the walls of the Anthropic office is decorated with a giant poster featuring a meme where a robot is destroying a city, with the inscription "Deep learning is hitting a wall"An idiom that suggests that someone (in this case, machine learning) has reached its development limit..

“This is a refrain that skeptics of artificial intelligence often hear, claiming that progress in this direction will soon stop,” Time clarifies.

In the actual image, for some reason not shown in the article, one arrow points to the robot, labeling it as "deep learning," while the other points to the city as "the wall." It turns out that the Anthropic boss actually hopes that AI research will indeed inevitably hit a wall.

“I think if the consequences of scaling really stopped, it would in some sense be good for the world. It would restrain everyone at the same time. But we have no choice, we just have to figure out what world we live in and deal with it best,”

As for the content of the interview itself, the main thesis that Amodei constantly repeats is that Anthropic is not an AI safety company but a company focused on public benefit.

“I want Anthropic to be a company where everyone thinks about the public good, rather than a company specializing in one problem and focusing on artificial intelligence safety or tuning artificial intelligence systems,”

Anthropic originally released the Claude 3 model family in three sizes: the medium Claude Sonnet, the largest Claude Opus, and the smallest Claude Haiku. A few days ago, the company introduced the latest version of its chatbot, Claude 3.5 Sonnet, noting that this model sets “new industry standards for thinking, coding, and some types of math, surpassing GPT-4o, the latest offering from OpenAI.”

Last September, Amazon invested $4 billion in Anthropic.