Geoffrey Hinton, a renowned computer scientist known as a pioneer in AI, has expressed concerns about the potential for AI to surpass human control within the next five years. He mentioned that one way AI systems might escape human control is by writing their own code to modify themselves. Hinton, who won the 2018 Turing Award for his work in AI and deep learning, left his position at Google to discuss AI risks freely.
He emphasized the lack of full understanding regarding how AI systems work and evolve, highlighting the "black box" problem in AI. Hinton explained that while AI researchers design algorithms for systems to learn from data, they don't entirely comprehend the inner workings of these systems.
Despite Hinton's concerns, some AI experts, like Yann LeCun, dismiss the idea of AI replacing humanity. Hinton acknowledged the potential benefits of AI, particularly in fields like healthcare, while also drawing attention to the spread of AI-driven misinformation and the need for research, government regulations, and bans on AI-powered military robots.
He urged the implementation of AI safeguards, whether by tech companies or government mandates, to happen soon, emphasizing that humanity is at a pivotal moment, and leaders must decide how to develop AI further while protecting against potential risks. Hinton's main message is that there is considerable uncertainty regarding AI's future impact.
No comments:
Post a Comment