Ant Group CEO Han Xinyi: AI will not replace doctors

```

On September 11, at the opening ceremony of the 2025 Inclusion·Bund Conference, Ant Group CEO Han Xinyi, Xiaomi Group Vice President of Mobile Zhang Lei, GSR Ventures Managing Partner Zhu Xiaohu and others discussed issues related to large model applications.

Han Xinyi shared Ant's thinking and practices in the AI medical and health field. He pointed out that professional capability is the core competitiveness of vertical models, and general models are unlikely to be replaced in the short term. Ant will focus on solving key issues such as data, hallucinations, and ethics, to make AI a doctor’s assistant.

When talking about the reasons for Ant's exploration into the AI medical and health field, Han Xinyi emphasized the dual characteristics of “rigid demand + medium to high frequency.” He pointed out that although medical treatment is a low-frequency activity, health management is a high-frequency demand; the combination of both provides fertile ground for deep AI service.

In response to the question, “Which is more suitable for the medical and health field, general AI or specialized AI?” Han Xinyi made it clear that the particularity of healthcare determines the irreplaceability of specialized AI. “If AI medical and health reaches the extreme, users won’t leave; they can solve problems here.”

Han Xinyi further explained that the ultimate goal of AI in healthcare is to be able to provide personalized, precise, and credible advice like professional doctors, including understanding the user's physical condition, reasonably recommending medications, and ongoing health management.

To achieve this, for quite a long time, general large models will be difficult to replace specialized vertical models. “Strong professional capability is itself a moat; the more it is used, the better it understands users. This deep understanding is exactly what many basic general models find hard to achieve.”

“Will AI replace doctors?” When faced with this industry focal question, Han Xinyi’s answer was clear and firm: “For a fairly long time, AI cannot replace doctors, and should instead be a doctor’s assistant.”

He stated that Ant is committed to providing important support for doctors through AI: helping specialist doctors expand the boundaries of their capabilities, moving closer to ‘general practitioners’; setting up a medical health laboratory, and conducting AI-empowered MDT (multidisciplinary team) consultations and other cutting-edge explorations.

“The only way out for AI in healthcare is human-machine integration,” Han Xinyi said. “We hope renowned doctors can have digital clones, allowing them to focus more on research and treatment of difficult or complicated illnesses, while large numbers of grassroots doctors also have great assistants.”

Han Xinyi also admitted that AI healthcare still faces three core problems: high-quality data, suppressing hallucinations, and medical ethics.

“High-quality data is fundamental, and its investment far exceeds imagination.” He revealed that the threshold for labeling and training medical data is extremely high. “The cost of a single piece of data may not be a few dollars, but hundreds or even more, and it also requires the participation of deputy chief or chief physician-level medical experts to ensure the quality of the training.”

Suppressing hallucinations is another challenge. Han Xinyi pointed out that the key is to “suppress hallucinations without reducing capability”— “The hard part isn’t just suppressing hallucinations, but reducing errors while not lessening the model’s service ability. This requires repeated polishing and balance.”

The most complex is medical ethics. To address this, Ant has specially established a medical ethics advisory committee, inviting top experts in the medical field to jointly explore norms. “No one has encountered this issue before, we must explore it together.”

Although healthcare is already a 10-trillion-yuan market, Han Xinyi stated that in the next few years Ant is in no rush to push for commercialization, but will focus more on solving issues such as accumulation of specialized data, suppressing model hallucinations, and building medical ethics. He believes that as long as the service is professional and trustworthy enough, users will naturally stay.

It is understood that since 2023, Ant Group has accelerated exploration in AI medical care, and in June this year launched the AI health manager AQ. At present, the total number of users served across all platforms has exceeded 140 million, connecting more than 5,000 hospitals and nearly 1 million real doctors nationwide, and has helped more than 300 renowned doctors create AI digital twins.

Risk Warning and DisclaimerMarkets have risks, investment requires caution. This article does not constitute individual investment advice, nor does it consider the special investment objectives, financial situation or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suited to their specific situation. Investments made accordingly are at your own risk. ```