Why 2024 Is the Year of AI Agents: Insights from Zhou Hongyi

National Committee member Zhou Hongyi published a signed article in the People’s Political Consultative Daily, discussing the latest progress in artificial intelligence (AI) and emphasizing that the rapid advancement of intelligent‑agent technology marks a key breakthrough for large‑scale commercial deployment, becoming a powerful driver of economic and social development.

AI is identified as a major force behind the current technological revolution and industrial transformation. A State Council executive meeting on July 31 called for vigorous AI+ action, leveraging China’s complete industry system, vast market size, and abundant application scenarios to accelerate AI’s integration across all sectors, declaring 2024 the “year of agents.”

The article points out two core issues with large‑model applications: insufficient reasoning ability for tasks like knowledge Q&A, and a lack of tool usage. While solutions like DeepSeek improve reasoning, they still lack the capability to act. Intelligent agents combine a model’s reasoning (the “brain”) with external tools (the “hands and feet”), enabling autonomous task understanding, planning, and execution.

To build professional agents, enterprises must convert internal expertise into knowledge bases and integrate agents with IT tools, allowing agents to perform specialized, personalized tasks. The stronger the tools, the more capable the agents become.

The article proposes a five‑level hierarchy of agents: L1 – chat assistants (e.g., ChatGPT, GPTs); L2 – workflow agents for fixed‑process tasks; L3 – reasoning agents that handle creative, open‑ended tasks; L4 – multi‑agent swarms that collaborate organically; L5 – self‑evolving AI agents that require no human definition.

Using 360’s swarm‑based scheduling center as an example, the system can continuously execute 1,000 steps with a 95.4% success rate, such as generating a 10‑minute video from script to storyboard without manual editing. 360’s “Intelligent Agent Factory” provides compute‑network scheduling, base models, storage, orchestration, and operating environments for easy enterprise customization.

While AI represents the pinnacle of digitalization, security remains the foundational layer. The shortage of security‑operation experts hampers many companies; intelligent agents could serve as “bodyguards” for security operations, but the rise of malicious agent hackers and AI‑driven cyber warfare also demands vigilance.

Looking ahead, 2025 is projected to be both the security year of agents and the birth year of security agents. Achieving breakthroughs in security capabilities will be essential to forging a new productive force for the AI industry.