AI Sparks in the Mountains
The Rongjiang Method and a People's Approach to Technology
An Unlikely Crucible of Innovation
In the mountainous terrain of Guizhou, one of China’s southwestern provinces, lies Rongjiang—a county of 385,000 people that was, until 2020, officially classified as a site of deep poverty. It is here, in this seemingly unlikely setting, that one of the most compelling grassroots experiments in digital sovereignty is unfolding. While conversations about Artificial Intelligence (AI) are often dominated by tech monopolies in the Global North, Rongjiang is quietly pioneering a method for the universalisation of AI as a public tool, adapted to the specific needs of a rural community. This is not a story about technology descending from on high, but about a community actively seizing and reshaping digital tools to serve its own developmental agenda.
This AI initiative does not arise from a vacuum. It is built upon a solid foundation of digital capacity forged through the county’s recent history, most famously exemplified by the Cun Chao (Village Super League). This local football tournament became a global social media phenomenon not by accident, but as a result of a deliberate strategy to build a collective, locally-controlled digital commons. By systematically training thousands of residents in the use of social media and live-streaming, the county leadership operationalised a new vision of rural development: ‘let the mobile phone become the new farm tool, let data become the new agricultural resource, and let live-streaming become the new farm work’. It is this pre-existing collective capacity that created the fertile ground for a county-wide leap into the age of AI.
The Method: From National Strategy to a People’s ‘Wisdom Battle’
Rongjiang’s exploration of AI is not an isolated act but a local manifestation of a broader national strategy. The Chinese state has identified AI as a strategic technology for the next stage of its development, a commitment recently reaffirmed in a 2025 collective study session of the Political Bureau of the CCP Central Committee. In his remarks, President Xi Jinping stressed the need to ‘highlight application orientation’ and promote ‘all-society general education’ on AI, underscoring the state’s vision of technology as a tool for public good and national development.
In Rongjiang, this national call was translated into a concrete local agenda with the official publication of the ‘Action Plan for Building Rongjiang into a National Benchmark County for the Universal Learning and Application of Artificial Intelligence’. This plan outlines an ambitious vision to achieve 100% training coverage in government units, 100% AI literacy outreach in all villages, and widespread application across all sectors of the local economy and society. This vision was driven forward by a county leadership that recognised the potential of AI to fundamentally reshape government efficiency and public service delivery.
To turn this vision into a reality, the party committee of Zhongcheng, a township in Rongjiang, devised a unique and highly effective method for mass mobilisation: the ‘Zhi Zhan 14 Tian’ (The 14-Day Wisdom Battle). Rather than relying on traditional, top-down training sessions, which often fail to inspire real engagement, they launched a two-week competition. All government staff were encouraged to use AI tools to solve concrete problems in their daily work, document their process, and submit their best case studies. To overcome the common hurdle of motivation, the prizes were not nominal certificates but practical rewards: one to three days of paid vacation. This simple, clever incentive transformed the learning process from a mandatory task into a popular and creative game, leading to a tenfold increase in the efficiency of tasks like writing reports.
Recently, Rongjiang held a county-wide competition to review the results of this AI adoption drive, which brought to light a remarkable number of innovative applications. The following sections will introduce some of the most representative cases that emerged from this collective experiment, grouped into three distinct scenes of application.
Scene One: AI as the Digital Assistant for Professional and Administrative Efficiency
Case 1: Automated Document and Report Generation
The General Duty Office of the county government demonstrated a highly sophisticated application of AI to solve a common administrative problem: the need to synthesize information from multiple, disparate source documents into a single, coherent meeting plan. The team gathered various background materials, such as briefing notes and participant lists, and uploaded them all simultaneously into the DeepSeek AI model. Their key innovation, however, lay in the prompt. Instead of merely asking the AI to summarise the text, they instructed it to generate the final output in HTML code, referencing a pre-existing document as a formatting template.
This case reveals a critical technique for administrative efficiency, which could be termed ‘code as formatting’. It represents a profound understanding of the AI’s capabilities, leveraging its coding skills to completely bypass the tedious and time-consuming manual labour of document formatting. The AI, acting on a single command, reads and synthesises the content from numerous files while simultaneously structuring it into a professional layout with tables, headings, and lists. The result is a perfectly formatted Word document, ready for download and distribution, produced in minutes rather than hours. This is not simply using AI to write; it is using AI to construct a finished, structured product, thereby reclaiming significant time from bureaucratic labour for more substantive work.
Case 2: Rapid Information Processing and Knowledge Management
A tutorial produced by Yang Xiucheng from the County Party Office addressed a different, though equally pervasive, challenge: information overload. The case demonstrates a workflow for rapidly digesting long-form content from the internet. The process is remarkably simple: the user copies the URL of a webpage—in this case, a lengthy article published on WeChat—and pastes it into the Doubao AI application, using the ‘AI Reads Website’ feature.
The core method here is the use of the ‘URL as the prompt’. This technique transforms the AI from a passive respondent into an active research assistant, capable of independently fetching, parsing, and analysing web content. The true power of this method is revealed in the multi-modal output: with a single click, the user can transform the source text into a concise, bullet-pointed summary for quick understanding, a visual mind map for grasping the conceptual structure of the argument, or a full translation for overcoming language barriers. This workflow effectively turns the AI into a ‘knowledge accelerator’, providing a powerful tool for officials who must constantly learn and adapt in a rapidly changing information environment.
Case 3: Strategic Content Ideation
The County Party’s Information Division showcased how AI could be used not just to process information, but to generate strategic ideas. Their primary challenge was the creative and political difficulty of ‘topic selection’—identifying relevant, timely, and impactful subjects for government reports and public communications. Their solution was to engage the AI as a strategic partner through a sophisticated ‘role-playing’ prompt. They instructed the AI to act as a ‘government affairs information officer’ and, providing it with the context of the current date and recent social hot topics, tasked it with proposing a set of newsworthy article ideas.
This practice elevates the use of AI beyond a mere search engine to that of a strategic consultant. The ‘role-playing’ technique forces the AI to filter its vast knowledge base through a specific professional lens, aligning its output with the user’s specific institutional goals and context. The results are not generic keywords, but fully-formed story pitches, complete with a rationale explaining their relevance and timeliness. This method offers a powerful tool for any organisation seeking to develop a more proactive and strategically-minded communications apparatus, helping them to shape the narrative rather than simply react to it.
Case 4: No-Code Development of Custom AI Tools
Perhaps the most advanced and transformative application was demonstrated by ‘Division One’ of the County Party Office, which used a suite of tools to build their own bespoke AI applications without writing a single line of code. Using the Coze platform, they created a custom AI ‘robot’ designed specifically for the task of proofreading official documents according to their department’s stylistic rules. In an even more complex example, they used Lark Base to build an automated system for agricultural analysis: a field worker can upload a photo of a diseased crop, and the system, powered by an integrated AI model, automatically identifies the disease, writes a diagnostic report, and suggests treatment options in the table.
This case exemplifies the democratisation of AI development. It proves that the power to create specialised AI tools is no longer the exclusive domain of software engineers. Platforms like Coze provide a visual, intuitive interface that allows frontline workers to encode their own domain expertise into a functional AI agent. This workflow, which combines human action (taking a photo) with an automated analytical process, is a model of practical, human-in-the-loop automation. It represents a fundamental shift from being a passive user of technology to an active creator of technological solutions tailored to immediate, local needs.
Scene Two: AI as a Factory for Public and Creative Content
Case 5: The Visual Interpretation of Public Policy
The county’s Supervision and Inspection Centre was tasked with communicating an abstract government policy—‘reducing the burden on grassroots officials by combating formalism’—to the very people it was meant to help. To do this, they used AI to produce a series of engaging, mixed-media Public Service Announcements (PSAs). Each video presented a short, live-action skit depicting a real-world example of a burdensome bureaucratic practice. Immediately following this, a friendly, 3D-animated AI mascot named ‘Xiao Zhi’ would appear on screen, explain which specific regulation the practice violated, and provide a hotline number for reporting similar issues.
This ‘mixed-media’ narrative strategy is a highly effective communication technique. The use of live-action footage grounds the problem in the lived reality of the audience, creating a moment of recognition and validation. The subsequent appearance of the animated AI mascot provides a clear, authoritative, yet non-threatening explanation of the solution. This method makes the often-opaque language of state policy legible and accessible, transforming a dense document into a simple, memorable, and actionable piece of public information.
Case 6: The Storytelling of Bureaucracy and Culture
Several departments demonstrated the power of AI to tell stories and shape institutional identity. The county Finance Bureau and the Party Office’s Clerical Division used AI image and video generation to create short animated films depicting the daily work of their staff. By using prompts that specified an aesthetic similar to that of the Japanese animation house Studio Ghibli, they transformed the mundane reality of office work into a warm and humanising narrative. In another case, a promotional music video for the Village Super League was created by using AI to generate 3D animated mascots and then compositing them onto a real photograph of the local stadium, seamlessly blending the digital and the real.
These cases highlight the role of AI in organisational and cultural branding. The ability to generate content in a specific, recognisable aesthetic allows any department or organisation to craft a unique public persona. By turning the story of their work into a visually appealing and emotionally resonant narrative, these government bodies are able to communicate their purpose and value to the public in a far more effective way than traditional reports or press releases. It is a powerful example of using AI to bridge the gap between the state and the people through the shared language of art and story.
Scene Three: AI as the Digital Archivist for Local Heritage
Case 7: The Digital Restoration of Collective Memory
In a poignant demonstration of AI’s cultural potential, Chen Haonan of the County Party Office undertook a project of digital restoration. He found a low-quality, black-and-white photograph online of Hu Bingduo, a revolutionary hero from Rongjiang county who lived in the early 20th century. Using a simple, smartphone-based AI application, he was able to restore, enhance, and colourise the image, transforming a faded historical artefact into a vivid and lifelike portrait.
This application showcases AI acting as a ‘digital conservator’. In the past, the restoration of historical materials was a highly specialised and expensive process, accessible only to well-funded institutions. This case proves that the tools for cultural preservation are now being democratised. The ability for any individual to reclaim and revivify a piece of their community’s history is a profound development. It makes the past tangible and immediate, allowing for a stronger connection between a community and its heritage, and ensuring that the memory of those who built the present is not lost to the decay of time.
Case 8: The Animation of Personal and Community History
Several individuals, including Peng Min and Chen Nian of the County Party Office, used a simple toolchain of AI applications to turn personal experiences into animated films. The workflow involved using one AI (Doubao) to generate a series of still images from a text script, another (Jimeng) to add subtle motion to those images, and a third (Jianying) to edit them into a final video with sound and text. With this method, they created stories about the personal journey of a grassroots official assigned to work in a village and an allegorical fable about local life.
This final case demonstrates perhaps the most radical potential of grassroots AI: the empowerment of ordinary people to become the storytellers of their own lives. This workflow transforms AI from a tool of professional communication into a medium for personal testimony and cultural creation. It provides the means for individuals and communities to document their own histories, share their own struggles and triumphs, and project their own voices into the public sphere. It represents a powerful liberation of the means of cultural production, enabling a more diverse and democratic narration of the world.
Conclusion: A Method for the Global South
The story of Rongjiang is not a technological fantasy. It is a grounded account of social organisation and collective learning. For many communities in the Global South, the promise of AI can feel distant and abstract, hindered by a set of common challenges.
First is the imagination gap: communities are often unaware of what AI can realistically do for them, and therefore do not know what to ask for.
Second is the participation gap: technology is frequently perceived as the domain of experts, leading to apathy and a lack of broad-based engagement from the community.
Third is the resource and skills gap: there is often a lack of funding, technical expertise, and accessible training.
Rongjiang’s experience offers practical responses to each of these challenges.
To close the imagination gap, they moved beyond abstract lectures and focused on creating a library of hyper-local, tangible video examples of what AI could do—restoring a grandparent’s photo, designing a poster for a local festival, or getting advice on crop pests.
To close the participation gap, they gamified the learning process through the "Wisdom Battle," with pragmatic incentives and a focus on solving participants’ own daily problems, while a "24-hour AI help desk" run by local enthusiasts ensured that no one was left behind.
To close the resource and skills gap, they leveraged state-provided vocational training funds and, most importantly, built their entire ecosystem around free, consumer-grade smartphone applications, proving that a transformative AI practice does not require expensive enterprise software.
Ultimately, Rongjiang's experience offers a simple lesson. Successful grassroots AI adoption seems to depend less on the technology itself, and more on the process of introducing it. By rooting the learning in solving real-world problems and encouraging collective participation, they have shown how any community can begin to adapt these powerful new tools to fit their own unique context and needs.