San Francisco (AP) – Jeffrey Hintonan award-winning computer scientist known as the “Godfather of Artificial Intelligence,” has some serious second thoughts about the fruits of his efforts..
Hinton helped pioneer the AI technologies necessary for a new generation of highly capable chatbots such as ChatGPT. But in recent interviews, he has said that he recently quit a high-profile job at Google specifically to share his concerns that unchecked AI development could be dangerous to humanity.
“I suddenly changed my views on whether these things are going to be smarter than us,” he said in an interview with MIT Technology Review.. “I think they are very close to it now and will be smarter than us in the future…. How do we survive that?”
Hinton is not alone in his concerns. Shortly after Microsoft-backed startup OpenAI released its latest AI paradigm called GPT-4 in March, more than 1,000 researchers and technologists signed a letter calling for a six-month halt to AI development. Because, they said, they pose “grave dangers to society and humanity.”
Here’s a look at Hinton’s biggest fears about the future of AI…and humanity.
It’s all about neural networks
Our human brains can solve calculus, drive cars, and track characters in “cascades” thanks to their original talent for organizing and storing information and deducing solutions to thorny problems. The roughly 86 billion neurons packed into our skulls – and more importantly, the 100 trillion connections those neurons make between themselves – make this possible.
By contrast, the technology behind ChatGPT features between 500 billion and 1 trillion connections, Hinton said in the interview. While this seems to put it at a disadvantage to us, Hinton notes that GPT-4, OpenAI’s latest AI model, knows “hundreds of times more” than any single human. Perhaps, he suggests, it has a “much better learning algorithm” than we do, making it more efficient at cognitive tasks.
Artificial intelligence may indeed be smarter than us
Researchers have long noticed that artificial neural networks take much longer to absorb and apply new knowledge than it does for people, because training them requires huge amounts of energy and data. That’s not the case anymore, Hinton says, pointing out that systems like GPT-4 can learn new things very quickly once researchers are properly trained. This is not unlike the way a trained professional physicist can wrap her brain around new experimental results much more quickly than the average high school science student can.
This leads Hinton to the conclusion that AI systems may actually be outpacing us. Not only can AI systems learn things faster, he noted, but they can also share copies of their knowledge with each other almost instantly.
“It’s a completely different form of intelligence,” he told the paper. “A new and better form of intelligence.”
Wars and rumors of wars
What can AI systems smarter than humans do? One worrying possibility is that malign individuals, groups or nation-states may simply co-opt them for their own ends. Hinton is particularly concerned that these tools could be trained to influence elections and even wage wars.
For example, election disinformation spread via AI chatbots could be the future version of election disinformation spread through Facebook and other social media platforms.
And this may just be the beginning. “Don’t think for a moment that Putin won’t build super-intelligent robots to kill Ukrainians,” Hinton said in the article. “He won’t hesitate.”
lack of solutions
What is not clear is how anyone could prevent a power like Russia from using AI technology to control its neighbors or its own citizens. Hinton suggests that a global agreement similar to the 1997 Chemical Weapons Convention might be a good first step toward establishing international rules against weaponized AI.
Although it is also worth noting that the chemical weapons deal did not stop what investigators found likely were Syrian attacks using chlorine and the nerve agent sarin against civilians in 2017 and 2018 during the country’s bloody civil war.