A former OpenAI security researcher is issuing a warning about the grim prospect that artificial intelligence could eventually lead to the end of humanity. Speaking on a podcast, former OpenAI Foundation employee Paul Cristiano said, “I think the likelihood of AI replacing humans is about 10-20%, or maybe most humans are already dead, and that needs to be taken very seriously.”
Cristiano believes that machines will gradually rise and the end of humanity is not a sudden event, but the gradual rise of artificial intelligence, humans and machines have already penetrated life through, for example, self-driving cars and drones. And when AI exceeds the threshold of human intelligence, anything is possible.
Cristiano believes that the most likely way for AI to kill humans is for all AI deployed in everyday life to join together, rather than suddenly wipe out humans.
Cristiano’s new nonprofit, Alignment Research Center, is based on the concept of AI alignment, which was widely defined back in 2018 as aligning the motivations of machines with those of humans. Despite OpenAI’s lip service to alignment, which it claims “aims to align artificial intelligence (AGI) with human values and follow human intentions,” more research and development is needed to ensure that AI is developed safely and ethically. Ensuring that the motivations of machines are aligned with those of humans is critical to preventing the possibility of AI destruction.
