China’s technology sector showcased a significant push toward generative artificial intelligence at the 2024 World Artificial Intelligence Conference (WAIC) in Shanghai, held from July 4 to July 7. Major firms including Alibaba, Baidu, and SenseTime unveiled new large language models and humanoid robotics, signaling an intensified effort to compete with Western AI advancements despite ongoing international export restrictions.
Advancements in Chinese Generative AI Models
At this year’s WAIC, Chinese tech giants focused heavily on the commercial application of large language models (LLMs). According to Reuters, Alibaba Cloud introduced its Qwen 2-72B model, emphasizing its open-source nature to encourage domestic developer adoption. Baidu showcased updates to its Ernie Bot, which the company claims has reached hundreds of millions of users since its public release.
These developments occur against the backdrop of U.S. government restrictions on the export of high-end AI chips to China. Industry analysts note that Chinese firms are increasingly optimizing their software to perform effectively on domestic hardware and older-generation chips to mitigate the impact of these supply chain limitations.
The Rise of Humanoid Robotics in Shanghai
Beyond software, hardware integration was a primary theme of the conference. Companies like Fourier Intelligence and various startups displayed humanoid robots designed for industrial and service-oriented tasks.
As reported by Xinhua, the exhibition featured robots capable of performing fine-motor tasks, such as handling delicate objects and navigating complex environments. The focus on robotics reflects China’s broader industrial policy, which aims to integrate AI into manufacturing to offset labor shortages and improve production efficiency.
Comparison of AI Development Trajectories
The following table highlights the distinct focus areas of major Chinese firms presented during the 2024 WAIC:
| Company | Primary Focus | Key Technology |
|---|---|---|
| Alibaba | Open-source ecosystem | Qwen 2-72B LLM |
| Baidu | Consumer-facing AI | Ernie Bot / AI Cloud |
| SenseTime | Industrial & Vision AI | SenseNova 5.5 model |
While U.S. companies like OpenAI and Google prioritize multimodal capabilities and large-scale reasoning models, Chinese developers at the WAIC emphasized "practicality" and "industry-specific" solutions. This divergence is partly driven by the need to adapt to hardware constraints and a distinct regulatory environment that mandates strict data security and content moderation compliance for all generative AI products.
Regulatory Environment and Future Outlook
The growth of China’s AI sector remains tightly linked to state oversight. All companies exhibiting at the WAIC must adhere to guidelines set by the Cyberspace Administration of China, which requires AI models to reflect "core socialist values" and undergo rigorous security assessments before public deployment.
Looking ahead, the industry is shifting toward "agentic" AI, where models are designed to execute complex multi-step tasks rather than simply generating text. The success of these initiatives will depend on the ability of Chinese firms to scale their compute infrastructure domestically while navigating a fragmented global technology market.
Keep reading