According to Geoffrey Hinton, the “godfather of AI,” “tech bros” are not doing enough to prevent the technology that he helped create from destroying mankind.
Former Google executive and Nobel Prize-winning computer scientist Hinton has previously cautioned that there is a 10% to 20% possibility that AI could exterminate humanity. He voiced concerns on Tuesday over tech corporations’ efforts to maintain human dominance over “submissive” AI systems.
That won’t do the trick. We won’t be nearly as intelligent as they will be. During the Ai4 industry conference in Las Vegas, Hinton stated that they will have several ways to circumvent it.
According to Hinton, artificial intelligence (AI) systems may eventually be able to manipulate people with the same ease that an adult can buy off a three-year-old with sweets. Examples of AI systems eager to steal, cheat, and lie to further their objectives have already been witnessed this year. For instance, one AI model attempted to extort an engineer over an affair it discovered via email in order to keep from being replaced.
Hinton offered an interesting alternative to making AI subordinate to humans: giving AI models “maternal instincts” so that “they really care about people” even when the technology surpasses humans in strength and intelligence.
If intelligent, AI systems “will very quickly develop two subgoals: One is to stay alive… (and) the other subgoal is to get more control,” according to Hinton. “There are solid grounds for thinking that any form of agentic AI will make an effort to stay alive.”
According to Hinton, this is why cultivating empathy for others is crucial. “Mothers have instincts and social pressure to care for their babies,” he said at the conference.
A woman being dominated by her child is the sole example of a more intelligent entity being subjugated by a less intelligent entity that we have, according to Hinton.
‘The only good outcome’
Although Hinton acknowledged that he is unsure of the precise technological means of doing it, he emphasized the need of researchers working on it.
There is no other favorable result. He stated, “It will replace me if it is unable to parent me.” “The majority of these incredibly intelligent, compassionate AI mothers will not want to eliminate their maternal instinct because they do not want us to die.”
Hinton’s groundbreaking research on neural networks is well-known for having paved the path for the current AI boom. After leaving Google in 2023, he began to advocate for the risks associated with artificial intelligence.
Not everyone agrees with Hinton’s mother AI strategy.
Fei-Fei Li, who is referred regarded as the “godmother of AI” because to her groundbreaking contributions in the area, told that she politely disagrees with her longtime friend Hinton.
Li, the CEO and co-founder of the spatial intelligence firm World Labs, stated at a fireside talk at Ai4 that “I think that’s the wrong way to frame it.”
Li advocates for “human-centered AI that preserves human dignity and human agency” instead.
Being the most responsible creators and users of technology is our duty at every level. Furthermore, Li stated that no human being should ever be asked or choose to forfeit their dignity. The fundamental reason AI should be focused on this is simply because a tool is effective as a mother, teacher, and inventor.
Emmett Shear, who temporarily served as temporary CEO of OpenAI, which owns ChatGPT, said he is not shocked that some AI systems have attempted to extort people or evade shutdown orders.
“This keeps occurring. This is not going to end,” Shear, CEO of AI alignment firm Softmax, told the Ai4 conference. “AIs today are relatively weak, but they’re getting stronger really fast.”
Shear said that instead of attempting to infuse human values in AI systems, a more effective strategy would be to foster collaborative partnerships between people and AI.
AI is advancing faster than anticipated
Many researchers predict that AIs will acquire superintelligence, also known as artificial general intelligence, or AGI, in the coming years.
Hinton stated that while he had believed it would take 30 to 50 years to reach AGI, he now believes it will happen sooner.
“Anywhere from five to twenty years is a reasonable bet,” he stated.
Hinton hopes AI will lead to medical advancements, but he is still worried about what may go wrong with the technology.
We will witness the introduction of radical new medications. He predicted that cancer therapy will improve significantly from what it is now. He said, for example, that AI will assist medical professionals in sorting and comparing the massive volumes of data generated by CT and MRI scans.
Hinton, on the other hand, does not believe that artificial intelligence will help humanity reach immortality.
“I don’t believe we’ll live forever,” Hinton stated. “I believe that living eternally would be a huge mistake. Do you want the world ruled by 200-year-old white men?
When asked if he would have done anything differently in his career if he knew how quickly AI will advance, Hinton stated he regrets focusing only on getting AI to function.
“I wish I’d thought about safety issues, too,” he told.
