In late March 2026, spring was just beginning to spread across the willow tips along both sides of the canal in Changzhou. Amidst this ordinary spring day in the southern Jiangsu region, the 2026 Wu Wenjun Artificial Intelligence Innovation Conference and the 15th Wu Wenjun Artificial Intelligence Science and Technology Awards Ceremony, hosted by the Chinese Association for Artificial Intelligence, came to a close here.

At this moment, the global artificial intelligence competition is entering a new phase. Across the ocean, the United States has launched the "Genesis Initiative," attempting to accelerate scientific research with AI; the United Kingdom has introduced the "AI-Empowered Science Strategy" to vie for a leading position in the global AI-driven scientific revolution. Domestically, the proposals for the 15th Five-Year Plan are pushing the "AI Plus" initiative to new depths. While the hype around large models has yet to subside, new tracks such as embodied AI and scientific AI have quietly emerged.

As top-level designs are implemented layer by layer, a more practical question comes to the fore: With policies in place and goals set, who will bring the technology into workshops, production lines, and hospitals?

The awarding of the Wu Wenjun Award this time precisely responds to this question.

At this conference, 116 projects and individuals took the stage to receive awards, with the award categories expanding from traditional fields like computer vision and natural language processing to emerging areas such as embodied AI and scientific AI. The awards ceremony also ventured beyond the usual locations of Beijing, Shanghai, Guangzhou, and Shenzhen, settling in Changzhou. This city, with an industrial scale exceeding 2.2 trillion yuan, boasts a complete range of manufacturing sectors and a solid new energy industry cluster.

Placing the Wu Wenjun Award, known as the "highest honor in China's intelligent science field," in Changzhou signifies that academia is stepping out of the laboratory and turning toward industrial hubs. And thus, the story of AI implementation begins.

 Behind the Record-Breaking Numbers: The Rising Concentration of Industry-Academia-Research

What makes this year's Wu Wenjun Award different?

Let's start with the numbers. In 2025, there was a significant increase in submissions, with 413 valid applications. Based on this, a total of 116 projects and individuals received awards, more than double the 53 winners from the previous year, setting a new historical record. At the same time, the award categories expanded from traditional fields like computer vision and natural language processing to emerging areas such as embodied AI.

The increase in award winners and the expansion of field boundaries point to the same conclusion: The development of artificial intelligence in China is entering a more active and diverse phase.

However, what deserves even more attention than these numbers is the rising concentration of "industry-academia-research integration" in this year's winning achievements. In other words, these awards are no longer purely academic honors but increasingly resemble industrial implementation report cards.

Let's examine a few specific award-winning projects.

First, consider the recipient of the Technology Contribution Award. This year's Wu Wenjun Award Technology Contribution Award went to Professor Sun Fuchun of Tsinghua University and Song Yongduan, a foreign academician of the Chinese Academy of Engineering. One scholar delves deeply into embodied AI and robotic manipulation, while the other specializes in adaptive control of neural networks.

Professor Sun Fuchun's team proposed the "Cognition-Action Entity" framework for embodied AI, attempting to solve a fundamental problem: How can intelligent agents achieve closed-loop perception, cognition, and behavior in the physical world? The team developed China's first multi-degree-of-freedom tactile dexterous hand, filling a domestic gap, and has promoted its application in over a hundred institutions. The technology has not only been implemented in national major project scenarios such as spaceflight teleoperation, unmanned aerial vehicle flight control, and dexterous operations in the 3C industry but has also achieved large-scale applications in companies like Xiaomi Technology, BYD Electronics, and Haier Group.

Professor Song Yongduan's team approached the problem from another angle: How to make neural network control safer and more reliable. They proposed a time-varying ideal weight neural network structure and barrier Lyapunov function technology to address the stability challenges of control systems in complex environments. The team led the establishment of the Ministry of Education's "International Joint Laboratory for Safety and Control of Autonomous Unmanned Systems," which focuses on the safe operation of unmanned systems in open environments.

The two scholars have different research directions but share a commonality: They both target real technological bottlenecks in industry. Awarding the Technology Contribution Award to individuals like them demonstrates that the Wu Wenjun Award values "who solves real problems." In the field of artificial intelligence, the ability to address practical industrial problems has become an important criterion for measuring value.

Now, let's look at a benchmark for industry-academia-research integration. The project "Key Technologies and Applications of Deep Natural Language Understanding and Generation," jointly completed by Harbin Institute of Technology (Shenzhen) and the Alibaba team, won the Special Prize for Scientific and Technological Progress. This project has been applied in dozens of companies, including Alibaba and Kingsoft Office, generating over 16 billion yuan in revenue in the past three years, with annual usage exceeding 6 trillion times.

There are also hardcore technologies that solve real problems. Beyond the Special Prize, a group of first-prize projects that have already been implemented on a large scale are also worth examining in detail.

First, in the field of intelligent driving, the project "Spatiotemporal Cognition Key Technologies and Industrial Applications from Visual Understanding to World Models," jointly completed by NIO and the University of Science and Technology of China, won the First Prize for Scientific and Technological Progress. Its core technology, the NIO World Model, has been fully rolled out to over 600,000 users. This is China's first intelligent driving assistance system developed based on the "world model + closed-loop reinforcement learning" paradigm. With 600,000 vehicles on the road, each one is continuously iterating and evolving using this technology.

The project "Intelligent Sparse Representation Coding for Multimedia Large Models" by Xiong Hongkai's team from Shanghai Jiao Tong University won the First Prize for Technological Invention. Its core technology has been used to build cutting-edge models such as the Baishitong 4K ultra-high-definition video generation base model and Huawei Yinwang's autonomous driving large model training. From video generation to autonomous driving, two seemingly unrelated scenarios rely on the same underlying technology. A single breakthrough can give rise to multiple applications, which is precisely the value of foundational research.

If the first two projects lean toward exploring technological frontiers, then the First Prize for Scientific and Technological Progress, "Key Technologies and Applications of Power Inspection Robots," directly addresses safety needs in high-risk scenarios. Crawler-type nuclear power plant robots, substation intelligent operation robots, and distribution network live-working robots are all essential in high-risk environments. This technology has been deployed in nine nuclear power plants across the country and distribution stations in over 20 provinces and municipalities, as well as exported to South Korea and the United Arab Emirates. Power inspection robots have replaced manual operations in high-risk scenarios, not only improving inspection efficiency but also reducing risks associated with high-risk work.

Three projects, three paths. NIO follows a closed loop of "automaker poses the problem, university provides the solution, users validate the results." Xiong Hongkai's team takes the route of "breakthrough in foundational technology, cross-industry migration." Southeast University follows the traditional path of "national demand-driven, engineering implementation validation."

These projects come from university-enterprise collaborations, research institutes, and universities as single entities, respectively. While the paths differ, the endpoint is the same: Technology has left the laboratory and created value in real-world scenarios. This shows that the Wu Award is becoming an award that "connects academia and industry."

By holding the awards ceremony in Changzhou, the intention is equally clear: to bring these industry-academia-research integration achievements one step closer to implementation and to bring more resources to the city. And thus, the story of how AI moves from the laboratory into workshops, production lines, and hospitals begins.

 From Awards to Collaboration: AI Implementation in Progress

If the awards ceremony is about showcasing the latest achievements in AI technology, then true technological implementation is reflected in the advancement of specific collaborations.

This time, the conference specifically organized the "Shaping the Future with Intelligence: Wu Award Achievements Implementation Tour in Jiangsu" event, bringing together over a dozen award-winning team experts and more than forty local entrepreneurs at the same table. The awards ceremony itself became a centralized matching platform, placing top-tier award-winning achievements and local industrial needs in the same space. The conference simultaneously accomplished both awarding and matching.

The collaboration between Changzhou Weiyi Intelligent Manufacturing Technology Co., Ltd. and the Institute of Automation, Chinese Academy of Sciences, quickly moved from intention to substance. In the intelligent upgrading of manufacturing, how to enable machines to quickly learn to identify defects in new product categories with minimal samples has long been an industry challenge. The Institute of Automation, Chinese Academy of Sciences, has deep expertise in rapid model convergence and migration in cross-category, small-sample scenarios. Addressing real industrial pain points, the two sides launched joint research and development focused on technological breakthroughs in cross-category, small-sample scenarios, aiming to enable machines to quickly learn to identify defects in new product categories with minimal samples, allowing machines to learn by analogy.

On the other side, Dai Yukun, the ecological strategy manager of Changzhou Health Medical Big Data Operation Co., Ltd., set his sights on multimodal AI applications in healthcare. He hoped to secure support from Fan Junchao's team at Chongqing University of Posts and Telecommunications and Yang Xiaoshan's research team at the Institute of Automation, Chinese Academy of Sciences, to collaborate in areas such as medical imaging, electronic health records, and health data fusion. The issues of data silos in healthcare and the technical challenges of multimodal fusion have long been persistent problems in the industry, and the award-winning achievements have made breakthroughs precisely in these directions. The supply and demand sides (supply and demand sides) found a convergence point in Changzhou.

It is worth noting that implementation does not only involve the introduction of external technologies but also includes the scaling up of local achievements.

The project "Key Technologies and Applications of a Sensor-Computing-Control Integrated Universal Controller for Robots," led by Dr. Zhang Tianlei, chairman of Mainline Technology, a Zhongjingkai enterprise in Changzhou, won the First Prize for Technological Invention. This autonomous driving company, which has been established in Changzhou for just over a year, has already achieved large-scale operations in dozens of leading logistics hubs, including Tianjin Port and Ningbo Zhoushan Port.

Reviewing these collaboration cases, a common feature emerges: The needs of the enterprise side are extremely specific, and the expert teams provide precisely the technologically accumulated expertise that has already been validated in these directions. This precise matching of pain points and accumulated expertise is precisely the core value of the Wu Wenjun Award extending from award evaluation to matching.

The conference "delivered" award-winning experts to the doorstep of industry and "presented" enterprise needs to the experts. The awards ceremony compressed the matching cycle between academia and industry from months to days. From inviting experts in to keeping them, from experts to partners, a new type of industry-academia-research relationship is quietly taking root in Changzhou.

Matching has been established, but the true value will only be fully unlocked through countless subsequent communications, debugging, and iterations. This is precisely the direction that requires sustained effort in the next stage.

 The Long Run of AI: From Placement to Rooting

The matching results are exciting, but from a longer-term perspective, the current implementation is only the initial placement.

In Go, placing a stone is just the beginning; the true outcome depends on how the stones connect to create real value. The encounter between Changzhou and the Wu Wenjun Award is no different.

The first challenge is the establishment of a Normalized docking mechanism (regular matching mechanism).

While a face-to-face exchange is highly efficient, the transformation of technology from the laboratory to the production line often requires months or even years of continuous refinement. A matching event may establish connections, but actual industrial implementation requires long-term technological adaptation. This involves secondary development of technology, scenario adaptation and debugging, and repeated validation of business models. Changzhou has already recognized this issue and explicitly stated that it will systematically organize expert resources and enterprise needs to establish a regular matching mechanism. This means that future efforts cannot rely solely on the "one-time matching" of a single conference but must create a continuously operating matching channel, allowing enterprises to find the right experts at any time and enabling experts to promptly understand the real needs of enterprises.

The second challenge is upgrading from point-to-point collaboration to ecological support.

In his speech, Academician Dai Qionghai proposed a more systematic vision: The Chinese Association for Artificial Intelligence will serve as a "terminal" linking national scientific and technological innovation achievements with Changzhou. A terminal is not a place for passengers to stay but a hub for resource transit, distribution, and arrival at destinations. Corresponding to the industrial level, this means that merely having "an expert help a company solve a specific problem" is insufficient. The matching between industry and academia needs to evolve from point-to-point collaboration to ecological construction, enabling more cutting-edge achievements and industrial innovations to take root in Changzhou, forming innovation hubs and creating a group of trillion-yuan industrial chains. This cycle of "enterprises pose problems, universities provide solutions, production lines validate results" requires supporting conditions such as computing infrastructure, data supply, and scenario openness. Without these, even the best technology will struggle to take root in industrial practice.

The third challenge is connecting activated needs with latent needs.

The collaborations that have already begun are promising, but a large number of latent needs remain to be uncovered. At the matching event, science and technology investment promotion leaders from Changzhou Economic Development Zone and Tianning Economic Development Zone respectively proposed needs such as lightweight solutions for quadruped and humanoid robots and low-cost intelligent equipment for small and medium-sized enterprises. The project "Key Technologies of Multimodal Digital Experts," a collaboration between Unisound and the University of Science and Technology of China, is expected to find new implementation scenarios in Changzhou. Behind these needs lies a broader industrial landscape: As AI technology permeates from leading enterprises to small and medium-sized enterprises, and from flagship projects to universal solutions, what is needed is not just breakthroughs in individual technologies but a systemic match between technology supply and industrial demand.

As the expert team remarked at the matching event: "Changzhou possesses vast industrial scenarios, a complete data infrastructure, and strong demand for AI, making it the best soil for technology transfer (achievement transformation)." But on this fertile soil, continuous cultivation is still needed.

The conclusion of the Wu Wenjun Award ceremony heralds a new beginning. When the award-winning achievements truly enter the workshops, hospitals, and ports of Changzhou, and when the award-winning experts truly become advisors to Changzhou's enterprises, this two-way journey between academia and industry will have truly reached its destination.

Spring in the Jiangnan region always begins with the germination of a single seed. In the spring of 2026, the Wu Wenjun Award sowed seeds of technological and industrial convergence in Changzhou.

And Changzhou, this industrial powerhouse striving to become a national demonstration city for "intelligent agents + scenario applications," stands at the starting point of this transformation, awaiting bloom.