Does the future of AI lie in social media?

As competition in the AI space intensifies, some of China’s top model developers are making a notable shift: they’re building social platforms. According to Chinese media outlet 36Kr, Moonshot AI is currently testing a content community feature within its AI assistant Kimi. The feature includes interactive functions such as likes and comments, and is now in a gray-box testing phase limited to selected users. The goal is to identify and fix product issues ahead of a broader rollout.

This is part of a growing trend. Chinese tech companies are betting that combining AI with social interaction could unlock a crucial edge: user retention.

AI + community

Earlier this month, Chinese social media platform Xiaohongshu held an AI developer competition, positioning itself as both a feedback engine for product iteration and a launchpad for promotion. For AI companies, social media isn’t just a distribution channel – it’s becoming a strategic layer for both inspiration and retention. 

On the one hand, community interaction offers valuable feedback and creative spark for developers. During Xiaohongshu’s recent AI development competition, many winning projects were born, refined, and promoted entirely within the platform’s social ecosystem. AI innovation, after all, thrives on collaboration and iteration. On the other hand, as large models become more capable, the industry is still searching for sticky, real-world use cases. Most leading AI apps, from OpenAI’s ChatGPT to DeepSeek, still function primarily as tools. That utility-first approach limits emotional engagement and user stickiness.

Chinese news outlet Jiemian cites an industry insider who compared the current wave of AI tools to early mobile-era utility apps. “The problem is that people don’t stay long,” she said. “Monthly active users may look strong, but time spent is still low.” That’s why some AI players are now betting on community features to increase stickiness. Another source cited by Jiemian says the line between tools and communities is blurring: “AI assistants need to shift from being event-driven to relationship-based.”

AI players already tapping into different forms of social networks

Even if it’s not always branded that way, AI companies already benefit from a kind of built-in community. 

ByteDance’s Doubao, for example, likely benefits from access to user behavior data on its short-video platform Douyin, enabling feedback loops that boost recommendation and interaction. Baidu’s ERNIE Bot, meanwhile, draws training data and user engagement signals from Baidu’s search engine, document-sharing site Baidu Wenku, and cloud platform Baidu Netdisk – while not social in the traditional sense, these still provide diverse user scenarios. After integrating DeepSeek’s model, Tencent’s Yuanbao briefly topped Apple’s China App Store, benefiting from its ecosystem, including WeChat’s public accounts and video platform. DeepSeek, on the other hand, is taking a model-first approach with no public signs of a social or community-facing product in development.

What’s next?

It remains to be seen whether AI-native social products will meaningfully challenge existing social media such as Xiaohongshu or China’s X-like microblogging platformWeibo. But the complementary nature of large models and social interaction is becoming more evident. If today’s AI tools want to evolve beyond one-off queries and productivity hacks, they’ll need the emotional hooks, peer sharing, and ambient discovery that social media excels at. For now, AI + community might be a bet. But it’s one that more players are starting to place.