Key Takeaways from AI Leaders at the 2024 Inclusion·Bund Conference

The 2024 Inclusion·Bund conference showcased a dazzling array of AI technologies and discussions, featuring over 10,000 m² of exhibition space and 44 insight forums where industry leaders debated AI limits, deployment challenges, and global compute gaps.

01 Richard Sutton – The Experience Era of AI

Richard Sutton, 2024 Turing Award winner and reinforcement‑learning pioneer, warned that the human data dividend is nearing its limit and that AI is entering an “experience era” driven by continual learning and meta‑learning.

Human data dividends are approaching their limit; artificial intelligence is moving into a continuous‑learning‑centric “experience era” with potential far beyond previous expectations.

He emphasized that current machine‑learning models mainly transfer static knowledge without autonomous learning, and that new data sources generated through direct agent‑world interaction—observation, action, and reward—are needed. Sutton predicts four outcomes: no consensus on world operation, eventual understanding and creation of intelligence, super‑intelligent AI surpassing humans, and power concentrating in the smartest agents.

02 Wang Jian – Open‑Source as a Strategic Variable

Wang Jian, founder of Alibaba Cloud, argued that the choice between open‑source and closed‑source has become a decisive factor in AI competition.

Open‑source is not a strategic mistake; it is a historical choice that shapes AI’s future.

He highlighted the U.S. export‑control move targeting closed‑source model weights, the evolution from code‑level openness to resource‑level openness (data, compute, model weights), and the launch of an 8‑billion‑parameter model on twelve satellites as part of the “Three‑Body Constellation” project, demonstrating true space‑borne AI.

03 Ma Yi – AI Is Still in Its Early Life Stage

Ma Yi, dean of HKU’s School of Computing and Data Science, compared current AI to the earliest stage of biological life, arguing that true intelligence requires moving beyond large‑model dependence to “white‑box” models grounded in mathematics and closed‑loop feedback.

Artificial intelligence today lacks a scientific understanding of its essence; it is still in the primordial “species‑level” stage.

He suggested that AI should learn from nature’s efficient learning mechanisms, emphasizing observation‑action‑reward loops as the core of intelligence.

04 Wang Xingxing – The Dawn of Embodied Intelligence

Wang Xingxing, CEO of Yushu Technology, described the emergence of an embodied‑intelligence industry where AI and robotics combine to give machines autonomous perception, planning, and action.

AI‑driven robots can now perform tasks autonomously, but data quality, multimodal fusion, and model‑control alignment remain major challenges.

He noted that scaling hardware brings management challenges and that AI‑enabled small teams will drive future innovation.

05 Yuval Harari – Measuring Progress by Cooperation, Not Speed

Historian and author Yuval Harari warned that AI is not merely an automation tool but an autonomous decision‑making entity, and that progress should be measured by collaborative effort and empathy rather than raw technological velocity.

Progress is not about speed; it is about the depth of cooperation and shared empathy.

He cautioned against deploying AI systems before establishing global governance and verification mechanisms.

In addition to these five speakers, dozens of other experts—including investors, scientists, and entrepreneurs—shared insights on topics such as infrastructure scaling, nuclear fusion, healthcare AI, and more, reflecting the conference’s breadth and depth.