Warns of own robot soldiers: "Regret"

Artificial Intelligence  

Godfather of AI technology quits: "Regrets my life's work"  

Of:  

Joachim Kerpner  

Published: Very recently  

NEWS 

Geoffrey Hinton is usually called the godfather of AI technology. 

Without him, today's text robots would not have existed. But now he drops out: 

- Part of me regrets my life's work, says Hinton.  

When over 1,000 AI researchers and tech leaders in an open letter recently called for a six-month halt to the development of text bots like ChatGPT, Geoffrey Hinton, 75, was not one of the signatories.  

On Monday, he quit Google's AI laboratory in Toronto, Canada, to publicly criticize the technology:  

- I don't think the companies should develop this further until they know if they can control the technology, says Geoffrey Hinton to The New York Times newspaper. 

A decade ago, the search giant Google paid 44 million dollars, about 280 million kroner, for the company that Hinton started with two students.  

The company built on technology that Hinton and his students created in 2012, which enabled computers to identify objects in photographs, such as flowers and cars. Using self-learning algorithms that try to mimic what happens in the human brain - a neural network - computers were trained to recognize patterns.

Geoffrey Hinton. Photo: Twitter  

"Perhaps better than what goes on in our brains"  

Hinton's new technology allowed OpenAI—which hired one of Hinton's students—to create the ChatGPT text bot, while Google, which hired Hinton, could develop Bard. He is concerned by the explosive development of text bots: 

- What goes on inside these systems is perhaps much better than what goes on inside our brains, he says, noting that the language handling within OpenAI's and Google's text robots has become incredibly much better in the past year: 

- Look at how it looked five years ago and how it looks now. Note that difference and think about what might happen going forward. It feels scary.  

In the near term, he worries that the internet will be flooded with fake, AI-generated photos, videos and texts, leaving the public no longer aware of what's real. 

In the 1980s, Geoffrey Hinton was a professor of computer science at an American university that received grants from the Pentagon, the US Department of Defense.

Concerned about future use of 'robot soldiers'  

 

Then he moved to Canada, because he did not want to contribute to the US's future use of "robot soldiers" on the battlefield.  

 

He still fears that AI soldiers could become a threat to humanity, if the AI systems become smarter than humans. And it can happen sooner than expected: 

 

- Most people thought that day was far away. I thought machines would be smarter than humans in 30 or 50 years, or even more. I don't think so anymore, says Hinton to The New York Times.

 

He believes that his old employer Google acted cautiously until recently. But when OpenAI and Microsoft (via Bing) released their vastly improved text bots recently, Google was forced to pick up the pace. A global race has begun that may become unstoppable, as no one can stop the companies. Even if restrictive global rules are introduced, no one can control whether AI technology develops quietly, he 

believes:  

 

- It is difficult to see how to prevent dangerous actors from using the technology for bad purposes, says Geoffrey Hinton.