Anthropic suggests a global reduction in AI development

Anthropic proposes a global slowdown in AI development

The company says it is concerned about AI's ability to create its own successors.

Anthropic says AI is developing so quickly that the trend is for systems to become capable of developing their own successor. We're not there yet, but she believes it "could happen sooner than most institutions are prepared for." In a blog post, Anthropic explains that an AI capable of building itself could “bring enormous good to the world” in science and health, among others. However, it could also “increase the risks of humans losing control of AI systems.” So how can humanity stop Skynet from becoming a reality? Anthropic suggests that the answer lies in globally slowing or temporarily pausing AI development “to allow societal structures and the search for alignment to keep pace with the technology's progress.”

The company is now one of the leading names in the AI ​​race and, unlike its competitors who have yet to make money, is reportedly on track to have its first profitable quarter. It recently filed documents with the SEC to make them public, likely before the end of the year. As the Wall Street Journal notes, critics suggest that Anthropic's warnings about its own technology are a marketing ploy, perhaps intended to make it seem the least egregious among all AI companies or to make its products appear to be the best on the market.

Critics specifically cite the limited release of its Mythos cybersecurity AI model as an example. Anthropic said it was making Mythos available to a select group of partners because of the potential damage its ability to quickly identify vulnerabilities could cause in the wrong hands. But people think it's just a ploy to promote the product or to cover up the fact that Anthropic only wants to sell it to bigger companies.

It's worth noting, however, that the company's suggestion is based on findings from the Anthropic Institute, a research division established in March. Anthropic said at the time that the institute's role was to "tell the world" what it was learning about the challenges that arise as AI companies develop more advanced AI systems. The institute, in collaboration with its collaborators, will conduct research on what is needed to “build the systems that a credible slowdown or pause would require.”

If AI companies accept a slowdown, mechanisms must be in place to verify that they have all actually stopped or slowed down AI development. Otherwise, some could get ahead of others after developing their technology in secret. “A significant slowdown or pause would require multiple well-resourced laboratories at or near the border, in multiple countries, to agree to halt under the same conditions,” Anthropic writes. “Everyone should also be able to check that the others have stopped.”

His suggestion would only come true if all AI companies around the world could come together and commit to stopping AI development for a period of time. Anthropic says it's not impossible, citing nuclear weapons treaties as an example, while admitting that these agreements have been decades in the making. We don't have much time, given that AI development is progressing at a rapid pace. The company plans to speak with policymakers, researchers, and other AI companies about this issue over the coming months and will publish the results of these conversations.