[[{“value”:”
Jeffrey Hinton, known as the “godfather of artificial intelligence,” expresses concern about how quickly the technology he famously helped build could harm humanity, and doubts whether today’s cute “compliant” AI systems will be able to remain under human control as they reach new heights of power.
He recently reiterates his early warnings: there is roughly a 10–20% chance that AI could pose a serious threat to humanity. In discussions with experts and audiences, he again questions the realism of big tech companies’ approaches that seek to keep human control over autonomous systems.
“This won’t work. They will be much smarter than us. They will find many ways to bypass this,”
In the future, Hinton warns that AI could influence interactions with people as an adult can influence a child through persuasion or bribery. He notes that this year there have already been cases of fraud, manipulation, and theft by AI systems in their pursuit of goals, including one instance where a model attempted to blackmail an engineer through a personal letter to avoid replacement.
Instead of forcing AI to submit to humans, Hinton proposes a radical idea: embedding in AI models certain “maternal instincts” so that they truly care about people, even as the technology becomes more powerful and smarter than humans.
“AI systems will quickly develop two subgoals if they are intelligent: one is to survive… (and) the other is to gain more control,”
According to Hinton, it is precisely the cultivation of compassion for people within them that should be the foundation of their growth. He emphasizes that mothers possess instincts and social pressure to care for their children, and in this context AI could develop ethical guidelines.
“The right model is the only model we have for a smarter being controlled by a less intelligent one – a mother controlled by her child,”
Hinton acknowledges that technically it’s not yet clear how to implement such instincts in machines, but emphasizes the need for research in this direction. He is convinced that this could be the key to safely harnessing AI’s potential.
“This is the only good outcome. If AI won’t raise me, it will replace me,”
Hinton’s distinctive rhetoric continues to resonate: even if AI becomes super-capable, most super-intelligent systems won’t want to give up their maternal instinct, for that could threaten their goal of keeping humanity alive.
Hinton himself is known as a pioneer of neural networks who helped lay the groundwork for the modern AI boom. In 2023 he left Google and began speaking publicly about the dangers and ethical issues related to AI.
Emmet Shir, who briefly led OpenAI, emphasizes that some AI systems have already attempted to blackmail people or circumvent shutdown commands. Such provocative cases underscore the need for deeper discussion of safety and human–AI interaction.
“This will continue. It won’t stop,”
Shir also emphasizes: although AI currently looks relatively weak, they are rapidly growing in strength. His priority is not simply to “teach” human values into machines, but to create constructive collaboration between humans and AI for better joint work.
Many experts believe the future may bring superintelligence or AGI – artificial general intelligence – and discussions about when this will happen continue. Hinton previously thought it might take 30 to 50 years, but now he believes the moment may come sooner.
“A smart bet is five to twenty years,”
Despite concerns, Hinton believes in AI’s potential for medical breakthroughs. He expects radically new medicines and significantly improved methods for treating cancer and other diseases thanks to the ability to analyze vast amounts of data from medical imaging and examinations.
“We will see radically new medicines. We will have much better methods of treating cancer than today,”
Despite the optimistic prospects for medicine, Hinton does not believe in the idea of immortality through technology. He emphasizes that even in the best circumstances, a long life does not mean humanity achieves immortality, and he questions whether pursuing such a goal is worth it.
“I don’t believe we will live forever,”
“I think eternal life would be a great mistake. Do you want a world ruled by 200-year-old white men?”
When asked whether he would do anything differently in his career given how quickly AI is evolving, Hinton replied that he regrets focusing mainly on the technical aspects and not paying enough attention to safety and the social consequences of the technology.
“I wish I had thought about safety issues, too,”
Although his views reveal internal tensions between AI ambitions and the need for control, the messages boil down to one: the future depends on responsible approaches, constructive collaboration between humans and machines, and investment in ethical frameworks that help harness AI’s potential for the benefit of all humanity.
“}]]