On October 11, the forum of "Talking about Science and Technology Imagination, Meeting the Future of Science Fiction" was held on the first floor of Zhongguancun Science Fiction Industry Innovation Center. As one of the key activities of the 26th Beijing Science and Technology Exchange Academic Month, the forum is hosted by China Science Fiction Research Center, Beijing Yuanyu Science Fiction Future Technology Research Institute, and Beijing Shijingshan District Future Science Fiction Industry Development Center.
Group photo of the forum (Photo courtesy of the organizer)
Chen Weicheng, member of the party group and vice chairman of Beijing Science and Technology Association, and Zhang Pengxiang, deputy director of Beijing Shijingshan District Science and Technology Committee, attended the event and delivered welcome speeches.
Han Liqun, academician of the Academy of Engineering and Technology for the Developing World, the first batch of members of the Chinese Engineering Intelligence Society, and the executive director of the China Society for Education Development Strategy, delivered a keynote report on "Brain-like AI Aspects" at the scene.
Han Liqun spoke. (Photo courtesy of the organizer)
Han Liqun, who has long been engaged in cutting-edge academic research in the fields of intelligent systems, pattern recognition, and artificial neural networks, mentioned that the current understanding of advanced functions such as thinking, cognition, decision-making, and learning of the human brain in brain science is far from enough, and through brain-like intelligence, it cannot become an effective way to general artificial intelligence, "The limitation is that the current human cognition of the brain has not reached a very clear level, and the ceiling is still very low." ”
She said that over the past decade, national governments, international organizations and famous companies around the world have proposed brain research programs, which integrate brain science and artificial intelligence projects. These include both soft brain research based on models and algorithms, as well as hard brain research based on physical implementation, but no breakthrough results have been announced.
At present, the research of hard brain mainly includes three levels: artificial synapses, brain-like chips and brain-like networks. Artificial synapses are mainly devices that use physical effects, such as memristors and synaptic transistors, to simulate synapses in the human brain. Synapses are characterized by the integration of storage and computing, that is, the basic unit of information processing can store data and perform calculations. This is also the biggest advantage of the human brain compared to computers in terms of information processing mechanism. Han Liqun specifically mentioned the latest memristor memory and computing integrated chip recently developed by the Tsinghua University team. At present, the relevant international research still stays in the memristor array level learning function demonstration, the whole system integrated memristor on-chip learning chip has not yet been realized, and Tsinghua's new chip can support efficient on-chip learning, that is, machine learning can be directly completed on the hardware side, and the energy consumption is only 3% of the application-specific integrated circuit system under the advanced process. "At this level, it is indeed the first breakthrough in the world, and it is of great significance." She said on the spot.
But Han Liqun also said frankly at the scene: "In the case of little understanding of the biological mechanism of human brain intelligence, we want to imitate it through brain-like, which must be a blind person touching the elephant to gradually approach, and sometimes it is inevitable to fall into the situation where only the trees do not see the forest." This is not something that the parts can be combined to arrive at the correct answer after breaking through one by one, nor is it that countless relative truths together will approach absolute truth. Therefore, Han Liqun believes that in the case that brain science has not been able to interpret the advanced functions of the human brain in the near future, the realization of general artificial intelligence does not need to blindly pursue brain-like technical routes.
At the forum, Ye Yurui, Co-Secretary-General of the Metaverse Industry Committee of China Mobile Communications Association, Yu Fei, Science and Technology Project Manager of Beijing Municipal Science and Technology Commission and Zhongguancun Administrative Committee, Liu Yong, Ambassador of China Aerospace Science and researcher of the National Space Science Center of the Chinese Academy of Sciences, and other experts also shared their research progress and latest views from the metaverse, Internet 3.0 industry status, aerospace science popularization and other fields.
The forum aims to provide a communication platform for researchers, the business community and the public to discuss the construction of the science fiction metaverse and empower the industry as the main goal in the context of the exponential development of the technology industry, cultivate new technologies, new products and new models, and jointly promote the high-quality development of the metaverse industry.
Written by: Nandu reporter Lv Hong