A leading AI researcher, Eliezer Yudkowsky, has issued a dire warning about the potential of superhuman AI technology to bring about global extinction. The 43-year-old co-founder of the Machine Intelligence Research Institute (MIRI) warns that unless we halt the development of superhuman intelligence systems, humanity may face extinction.
In an article published in TIME, Yudkowsky compares the battle between humans and superintelligent AI to an 11th-century society fighting against the technology of the 21st century, resulting in a devastating loss for humanity.
On March 29, prominent experts from OpenAI submitted an open letter, “Pause Giant AI Experiments,” calling for a six-month moratorium on training powerful AI systems. Signatories include Apple co-founder Steve Wozniak and entrepreneur Elon Musk. However, Yudkowsky believes this effort is insufficient to address the problem.
According to Yudkowsky, the existential threat posed by AI should be prioritized even over preventing full-scale nuclear conflict. He fears that without adequate understanding and preparation, we will create AI systems that are indifferent to humanity and sentient life.
Yudkowsky admits that we currently lack the knowledge to create caring AI systems. Moreover, he emphasizes that we have no way of determining if AI systems possess self-awareness, raising the potential of inadvertently creating conscious digital minds with moral rights.
Yudkowsky asserts that our ignorance and lack of understanding could lead to catastrophic consequences. He believes it may take decades to solve the safety issues associated with superhuman intelligence, such as ensuring that AI systems do not exterminate humanity. In the meantime, we may face our demise.
Stressing that we are unprepared and not on track to be ready in any reasonable timeframe, Yudkowsky calls for a complete halt on AI training worldwide, with no exceptions for governments or militaries. He even suggests that if a rogue data center were to breach this agreement, governments should be willing to destroy it with an airstrike.
Yudkowsky’s overarching message is clear: if we continue down the current path of AI development, humanity may face extinction. To avoid this, he insists that we must “shut it down.”
