In USD 30 Million Mansion laid out on a cliff overlooking the Golden Gate bridge, a group of AI researchers, philosophers and technologists to discuss the end of humanity.
Sunday afternoon of the symposium called “the values of the successor”, turned around Provocative idea From the entrepreneur Daniel Fggella: “Moral goal” of advanced artificial intelligence should be to create a form of such powerful and wise intelligence that “you prefer (not humanity) to determine the future path of life.”
Fggella explained this topic in her invitation. “This event is very focused at the posthuman crossing,” he wrote to me by X DMS. “Not to Agi, which forever serves as a tool for humanity” –
The event filled with futuristic fantasies, in which participants discuss the end of humanity as a logistic and not metaphorical problem, can be described as a niche. If you live in San Francisco and work in artificial intelligence, this is a typical Sunday.
About 100 guests cared for non -alcoholic cocktails and nibbled cheese plates near the windows from the floor to the ceiling directed to the Pacific Ocean, before they gathered to hear three conversations about the future of the interview. One participant wore a shirt with the inscription “Kurzweil was right”, apparently a reference to Ray Kurzweil, a futurist who predicted machines in the coming years of human intelligence. Another wore a shirt with the inscription “Does it help us get to Aga?” In the company of emoji thinking of the face.
Fggella told Wired that he had thrown this event, because “great laboratories, people who know that Agi will probably end humanity, do not talk about it, because encouragement does not allow the fact that they do not allow them that they are not able to kill us, such as Elon Musk, Altman and Demis Hassabis, who are on all. build “(to be candid, musk He still talks about the risk Related to advanced artificial intelligence, although it did not stop him from racing).
It’s LinkedIn, Faggeella boasted Star a guest list with the founders of artificial intelligence, researchers from all the best AI laboratories and “Most of the important philosophical AGA thinkers”
First speaker, Ginevera davisThe writer from Novel York warned that human values may be impossible to translate into AI. She said that machines may never understand how it is to be awareness, and an attempt to cod human preferences in future systems can be miniature -sighted. Instead, she proposed a sublime sounding idea called “Cosmic alignment” “Building artificial intelligence that can look for deeper, more universal values that we did not discover. Its slides often showed a seemingly generated image of techno-uutopia, with a group of people, we gathered in a furry city in distance.
Machine consciousness critics will say that huge language models are simply stochastic parrots – a metaphor invented by a group of researchers, some of whom worked in Google who wrote in the famous article That LLM does not really understand the language and are only probabilistic machines. But this debate was not part of the symposium in which the speakers took part in the idea that superinterteration was coming and quickly.