AI researchers share videos warning about superintelligent AI risks

In short: A nonprofit-backed website published interviews with AI researchers who warn that very advanced AI could pose serious risks, including human extinction.

What happened

A group called Palisade Research has released a series of video interviews with AI researchers on a website called frominside.ai. Palisade Research says it is a nonprofit that studies what AI systems can do and what they might try to do.

The interviewees include current and former employees from OpenAI, Google DeepMind, and Anthropic. In the videos, several researchers argue that future AI could become so capable that people may not be able to control it.

Geoffrey Irving, a former OpenAI and Google DeepMind employee, said he views the chance of human extinction as “about a coin flip.” Neel Nanda, a Google DeepMind research scientist, said there is “at least a 10 percent chance” AI could cause human extinction, and called that “ridiculously high.”

Former OpenAI researcher Daniel Kokotajilo described “superintelligent” AI as “god-like powerful” and said, “we don’t know how to control them at all.” The site also groups together short clips of different people answering similar questions, including whether the messaging is hype or marketing.

Some interviewees acknowledge conflicts of interest. Google researcher Mary Phuong says viewers should be “suspicious” because she is paid by an AI lab. Nanda says he believes his work reduces these risks, and he argues the companies will keep building such systems even if individual researchers quit.

Why it matters

These interviews show that concern about AI safety is not only coming from outsiders. Some of the warnings are coming from people who helped build the technology. For the public, it is a reminder that debates about AI are not just about chatbots and convenience, but also about how to set limits before systems get more powerful (like building brakes while the car is still being designed).

Source: The Verge AI