Home > News > Techscience

In the 2024 Global Developers Pioneer Conference held in Shanghai, experts in the field of artificial intelligence (AI) discussed how to address the impact of AI from various perspectives including technology, industry, and humanity. The conference provid

ZhangYi Tue, Mar 26 2024 06:14 AM EST
Dr. Shen Xiangyang, former Executive Vice President of Microsoft and responsible for AI business, described AI as having the potential to dominate all industries. However, the development of large-scale models currently relies heavily on computing power. At the top level, there are general-purpose models like ChatGPT with training parameter scales in the trillions and training on thousands of cards. Below that, there are industry-specific models with parameter scales in the billions and enterprise-specific models with parameter scales in the hundreds of millions. Dr. Shen emphasized the issue of computational power and quoted the saying, "poverty will limit imagination." Dr. Qiao Yu, the leading scientist at the Shanghai Artificial Intelligence Lab, raised the question of when the demand for computing power and resources in developing large models will reach its limit. He believes that in the next few years, the consumption of computing power and parameters by large models may increase by one to two orders of magnitude. However, he also questioned whether this increase is sufficient to lead humanity into the era of general artificial intelligence (AGI). He thinks this is a question that every developer should contemplate. Recently, DeepMind, a subsidiary of Google, introduced the mathematical model "Alpha Geometry" with only 150 million parameters. It was able to solve 25 out of 30 international mathematical Olympiad problems, reaching the average level of gold medalists. Dr. Qiao suggested that instead of relying solely on large mode... ——Will this lead to a general decline in human intelligence? In the era of AGI, with AI replacing humans in various tasks and solving various problems, is it really possible for humans' reliance on AI to lead to this outcome? In order to avoid this trap, Kass believes that when designing AI, developers should not just let it think for humans, but should inspire human inspiration. ——Will it bring about an identity crisis? With AI replacing a large amount of work, even if people can "get things without effort" and obtain food and wealth, can we really adapt to a life that completely separates personal identity from occupation? Can we separate our sense of self from our profession? ——Can it truly "align" with human values? Can those beautiful things that parents pass on to their children today, such as courage, kindness, intelligence, and empathy, continue to be embraced by machines? Kass says that in the era of AGI, the more machines are capable of, the faster intelligence loses value. When people are no longer bound by work, he suggests that ordinary people should carefully consider how to live, especially how to make good use of the suddenly gained leisure time, "We can communicate more with family and loved ones, and share our happiness." According to Kass, the most important thing is to find the essential human qualities outside the realm dominated by AI, such as trust, cooperation, and care. He predicts that in the future, a person's value in the world will be defined by their ability to collaborate with others. As for the idea of "AI resurrection of the deceased" that is currently popular, Kass bluntly says that he is very worried that people really think they can replace humans with machines, but machines have no souls and do not generate ideas. "We should draw a clear line between the virtual and the real, and the most important thing in the future is the int...