Китайский ИИ пошел в провинцию: почему деньги не покупают любовь

Golden rush in the artificial intelligence world has shifted from the glittering skyscrapers of Shenzhen to dusty counties and small towns of China. When growth rates in megacities slowed down, tech giants like Tencent, Alibaba, and ByteDance remembered the old good tactics of market capture — bombard potential users with money. This time the occasion was the Chinese New Year, and the expansion tool was neural networks.

But something went wrong. Instead of mass adoption of technologies, companies faced cold pragmatism of provincial residents who are ready to take money but not ready to spend time talking with algorithms. Context here is more important than the news itself.

In China there exists a concept of "lower market" (xiachen shichang), covering hundreds of small towns and thousands of counties (xiancheng). This is exactly where the numbers for annual reports are forged, if the capital market is already oversaturated. Earlier this scheme worked flawlessly: WeChat Pay and Alipay became giants precisely thanks to "red envelopes" (hongbao) with digital money.

Now they decided to conduct AI models along this path. Alibaba promotes its Tongyi Qianwen, Tencent rolled out Yuanbao, and ByteDance is betting on Doubao. The logic is simple: give a person a couple of yuan for registration, and they will become your loyal user.

However, developers forgot to ask the provincial residents themselves. For a person from a county, AI is not a "personal productivity assistant," but a strange toy that can't plow a garden or fix a tractor. When marketing budgets began to turn into real payouts, it turned out that users perceive these applications as a short-term quest.

Downloaded, got a bonus, deleted. Attempts by neural networks to start a dialog or offer help with essay writing cause local residents, at best, bewilderment. Why write an essay if you just need to check the yuan rate or watch a short video on Douyin?

The problem here is not in technical backwardness, but in a colossal gap between product hypotheses and reality. In Beijing, developers genuinely believe that "smart search" is needed by everyone. In a county, meanwhile, people are used to voice messages and visual content.

Text chatbots for them are a return to the era of button phones, only with an incomprehensible interface. Even ByteDance's aggressive marketing, which embedded its Doubao in every possible nook of its ecosystem, faces the fact that people use AI only for the sake of getting virtual points or discounts. Moreover, a funny irony emerged: neural networks turned out to be too complex for ordinary people.

To get a coherent answer from Tongyi Qianwen, you need to understand how to write a prompt. But a regular user doesn't want to be a computer operator; they want the magic to work by itself. As a result, millions of yuan spent on attracting traffic from the provinces turn into "dead souls" — accounts that will never return to the app after holiday payouts end.

Companies buy installs, but don't buy loyalty. The situation with AI in Chinese counties is an excellent lesson for the entire world. A technological leap doesn't happen just because people got smartphones and an extra five minutes of time.

If technology doesn't solve a basic, down-to-earth problem here and now, it remains an elite entertainment. Until neural networks learn to help with harvest logistics or finding cheap fertilizers better than a village neighbor, no red envelopes will make them mainstream. The industry will either have to simplify the product to the level of one button, or admit that AI is still a story for white-collar workers.

Main point: Flooding the market with money helps inflate statistics for investors, but doesn't create real demand. Will tech giants be able to come up with scenarios for AI that will be understandable outside of coworking spaces?