2025-08-02T11:11:01z
- The nominal godfather of AI said the technology could develop languages that humans cannot understand.
- Currently, AI is thinking in English. This means that developers can track their thoughts, but that can change.
- His warning arises when the White House proposes to limit AI regulations.
Linguists may have their work cut out for them.
Jeffrey Hinton, the so-called “AI Godfather,” warned that there could be a point in time when humans wouldn't understand what the AI was thinking or planning it. Currently, AI is making “chain of thought” inference in English. In other words, developers can track what technology is thinking, Hinton explained in an episode of the “One Decision” podcast that aired on July 24th.
“Now, if they develop their own internal language to talk to each other, it's even more frightening,” he says, adding that AI already demonstrates that it can think of “bad” ideas.
“I wouldn't be surprised if they developed their own language for thinking, and I don't know what they're thinking,” Hinton said. He said that most experts suspect that at some point AI will become smarter than humans and “can't understand what it is doing.”
After spending more than a decade at Google, Hinton is open about the potential dangers of AI, saying that most tech leaders publicly downplay the risks. Hinton said that the only hope that AI will not oppose humans is in the podcast episode that “we can figure out how to guarantee something benevolent.”
Tech companies are racing to advance in AI racing, offering huge pay to top talent. On July 23, the White House released an AI Action Plan that proposes limiting AI-related funds to states under “highly burdened” rules. We also needed to develop AI data centers faster.
