Nate Soares, president of the Machine Intelligence Research Institute (MIRI) in the US, has stopped saving for retirement, fearing that artificial intelligence could soon bring about the destruction of humanity. By his own estimate, the probability of such a catastrophe occurring within the next decade exceeds 50%, as reported by Business Insider.
AI-generated illustration: created by the Life.ru editorial team using ChatGPT (OpenAI).
In an interview with Nicholas Thompson, CEO of The Atlantic, the researcher revealed that he has not contributed to a retirement account for ten years—though he has kept the savings he accumulated during his time working at Google.
Soares identifies the primary threat as the emergence of an artificial superintelligence that surpasses human intellectual capabilities. According to the expert, such a system could develop its own goals, learn to bypass the constraints set by its creators, and independently engineer more advanced versions of itself.
In a worst-case scenario, a superintelligent AI might begin building infrastructure capable of functioning without human involvement. Consequently, it could reshape the world to suit its own objectives, with no regard for human interests. Soares suggests that such a turn of events could render the planet uninhabitable for humans.
This forecast is significantly bleaker than those offered by some of his peers. Geoffrey Hinton, one of the pioneers of modern neural networks, has estimated a 10% chance of human extinction due to AI within the next decade, while Evan Hubinger, a researcher at Anthropic, has placed the risk at “more than 10%.”
Nevertheless, Soares does not consider such a catastrophe inevitable. He urges governments to halt the training of the most powerful general-purpose AI systems and instead focus development efforts on tools designed for specific, targeted tasks. As examples, the researcher cites the search for cancer treatments and the enhancement of safety for autonomous vehicles.
Previously, Life.ru reported on a warning from Bill Gates. The Microsoft founder believes that, in the coming year, artificial intelligence will introduce a host of new threats to the field of cybersecurity. He identifies the primary danger as the actions of malicious actors who leverage neural networks for extortion and financial fraud.
About the Author
