Zhou Bowen, a prominent artificial intelligence (AI) scientist previously associated with IBM and JD.com, has been appointed as the new director of the Shanghai AI Laboratory, one of China's premier AI research institutions.
His appointment follows the death of the lab's founding director, Tang Xiao'ou, a leading facial recognition expert and co-founder of SenseTime Group.
Zhou, 47, who made his public debut as director at the World Artificial Intelligence Conference (WAIC) in Shanghai, takes over an institution pivotal to China's AI ambitions (via South China Morning Post).
Zhou Bowen, former IBM and JD.com AI scientist, appointed director of Shanghai AI Laboratory. He aims to develop "balanced AI" focusing on safety and performance, integrating generalization and specialization capabilities.Image via Shanghai AI Lab
China's Top AI Scientists
Zhou's journey to this prestigious role is marked by significant achievements. He graduated from the University of Science and Technology of China and the University of Colorado Boulder.
Zhou worked at IBM from 2003 to 2017, serving as the chief scientist of the Watson Group and director of the AI Foundations Labs, where he contributed to the development of IBM's renowned supercomputer.
Post-IBM, Zhou joined JD.com, a leading Chinese e-commerce platform, where he held roles as cloud and AI president and director of AI research.
In addition to his corporate achievements, Zhou is the founder of the AI start-up Frontis and has been a chair professor at Tsinghua University since May 2022. He also serves as a non-executive independent director at China Citic Bank International and holds US citizenship.
In his address at WAIC (translated by Google), Zhou highlighted the current imbalance in global AI development, where developers prioritize performance over security.
He emphasized the need for a technical system that places safety at the forefront while pursuing long-term advancements in AI performance.
Under Zhou's leadership, the Shanghai AI Lab aims to build on its existing successes, such as the development of a text-to-video model used in China's first generative AI-assisted cartoon series aired by China Media Group in February.
Zhou also discussed the need for AI systems that integrate both generalization and specialization capabilities, which he believes are crucial for achieving significant productivity enhancements.
His vision includes a three-layered technical system encompassing a basic model layer focusing on general capabilities like mathematics and reasoning, a fusion and collaboration layer combining generalization and specialization, and an autonomous evolution and interaction layer for AI's self-exploration.
Additionally, Zhou Bowen has been contemplating the concept of artificial general intelligence (AGI) since 2015-2016, inspired by the success of AlphaGo and Deep Blue.
He distinguishes AGI from narrow (ANI) and broad (ABI) artificial intelligence, defining ABI as self-supervised, end-to-end, and generative.
According to Zhou, the emergence of ChatGPT in 2022 marked the beginning of the ABI era. He advocates for a two-dimensional approach to AGI, integrating both generalization and specialization capabilities to significantly enhance productivity.
He also pointed to AI's potential in scientific research, breaking disciplinary information cocoons and accelerating innovation.
Examples include developing open-source models in chemistry and breeding, creating algorithms for faster analysis of experimental results, training meteorological models for superior weather forecasts, and using AI to propose and verify new biomedical hypotheses.
Trustworthy AI
Concluding his address, Zhou expressed a firm commitment to achieving trustworthy AI by integrating general and specialized knowledge.
Drawing inspiration from historical scientific advancements, he celebrated the contributions of young scientists and envisioned AI as a valuable intern and helper for humans.
