AI needs realistic expectations, not excessive hype, Gartner says

As the rapid rise of AI agents reflects the evolving realities of China's rapidly advancing AI ecosystem, the technology should be met with realistic expectations rather than excessive hype to avoid a market bubble, according to a senior analyst at global research and advisory firm Gartner.

According to Gartner's new 2025 China Hype Cycle for AI technologies, AI agents have become one of the biggest movers on the curve and take the top spot at the "peak of inflated expectations".

"Powered by high-level automation, AI agents have the huge potential to transform industries ranging from consumer services and manufacturing to data analytics, content creation, and logistics," said Ben Yan, director analyst at Gartner.

AI agents are autonomous or semi-autonomous software entities that use artificial intelligence to perceive their surroundings, make decisions, take actions, and accomplish specific goals in digital or physical environments.

"One thing special thing is that they can make intelligent decisions and interact dynamically with their environments, so they have been attached to huge market expectations," Yan noted, adding that most current applications are, however, still far from being "true" agents.

"While AI agents built on large language models have shown strong perception and execution capabilities, their decision-making — especially in complex, context-driven enterprise settings — remains relatively weak," he said.

A similar trend can be observed with LLMs themselves. From the early excitement around ChatGPT to the "hundred-model competition", the sector has now passed the hype cycle's peak with the market's expectations cooling down, the senior analyst added.

"This cooling-off reflects a technical ceiling," he said. "The performance improvement of LLMs has become marginal. Despite different rankings, mainstream models perform similarly in practice, and the difference is barely noticeable for businesses or individual users."

Against this backdrop, Yan said that for China's data, analytics, and AI leaders, the priority now is to look past the hype and identify the real value of AI, which can help lay a solid ground for the next phase of enterprise-level AI competition.

lijiaying@chinadaily.com.cn