Smart Earth

In China's changing countryside，young farmers and local officialsareembracingAl to tackleconundrums from the aging population to pest control

當AI開始“下鄉\"

n the peach orchards of Jiangsu's Yangshan town，33-year-old farmerWangHuanwatches asa drone buzzes quietly above the saplings， capturing high-definition images of the field.

These daily flights collect visual data，which， combined with readings from soil temperature and humidity sensors，are entered into an AI algorithmbased on Alibaba Cloud to detect signs ofpestsand disease.If trouble is spotted，another drone will be dispatched to mist insecticide on the affected area.

“It only takes about half an hour for the drone to spraypesticide across mylO mu [about 0.7 hectares] of land，\"Wang says.“IfIdo itbyhand， it would take around five hours.\"

Only7O yearsago，farmersin China were turning to the Xinhua Dictionary for solutions to the fast-changing needs of modern agriculture. Today，artificial intelligence is helping young farmers like Wang overcome the challenges theyfaceina traditionallyexperience-based profession.As China's countryside grapples with anagingpopulationand increasingdemands forproductivity， both farmersand local officials are embracingAI to modernize agriculture and improverural governance.

AI in the fields

In 2021，tired of the stress from his e-commerce job，Wang returned to hishometown to farm.But he quickly realized thathe couldn'tmatch the stamina of veteran farmers，nor their intuitive knowledge of the land.“Older farmers rely on experience passed down over decades，\" says Wang.“Forus [newcomers]，it'snot feasible to wait 10 years to gain the necessary knowledge before putting it into practice.\"

Last October， theMinistry ofAgriculture andRural Affairsannounceda“National Smart FarmingPlan，”calling for theuse of AI tools in agriculture throughout the country. The plan envisions AI tools managing every task from cropraisingand animal husbandryto pest prevention.

InFebruary， the State Council reinforced the effort，advocatingforsmartagriculture through expanded AI applicationswith big data， the Internet of Things （IoT），andother advanced technologies for rural developmentand administration.

Local AI initiativeshavealso ramped up.Last February， the city of Zhaoqing in Guangdong province introduced the Mr.Lan chatbot， which can answer questions about the entire production chain of orchids from cultivation to distribution， and give maintenance suggestions. This February alsosawXiong'an，astate-level newareasouthwest to Beijing，launch its own chatbot，Xiong Xiaonong. Trained on data from both experts and experienced farmers，the chatbot offers guidance on planting and pest control.

For young farmers like Wang，AI offersboth practicaland psychological support.“Youmight knowhow to fertilizea tree，but not exactly how much it needs，\"Wang tells TWOC.With AI analyzing multispectral images captured by \"Compared to traditional methods， this [data-driven method] gives us a more objective way of managing crops and enablesustohandle larger fields.\"

drones，Wang can easily adjust the dosage of potassium to ensure everypeach ripens toa consistent sweetness.“Compared to traditional methods，this [data-driven method] gives us a more objective way of managing crops and enables us to handle larger fields.In the long run， its economic value is relatively higher.\"

AIcanalsoworkwithnear-infrared spectroscopytoassesstheripenessoffruits.Media outlet ShineGlobal reported in February that some farmers have turned to AI to check the ripeness of durians within seconds.This reduced a process that used to take three dayswith manual work to just two hourswith automation，and improved the accuracy of durian ripeness predictions from 50 percent to 9l percent.

Yet，automated agriculture doesn't come cheap.Afull drone setup with software licenses cost Wang over 2Oo，ooO yuan， plus annual software fees of nearly2，OoO yuan. This isa largeinvestment formostindependent farmers inWang'shometown，who farm small plots of justdozensofmu，roughlythesizeofafootball field.“You could hire a few people [to farm] for a fewhundred yuan per day， but it takes longer to break even when you use machines，”Wang tells TWOC.He plans to rent out the spraying drones to other farmers to recoup his costs.

Apart from finances， the average age of China's farming population presents another obstacle to the wider adoption of AI. In Wang's village， around 7O percent of farmersare over the age of 50，and many are reluctant or unable to adopt new technologies.Anarticle fromthe state media outlet Guangming Net，published in February， noted thatfewer than5 percentof farmersin Chinaare capable of using agricultural apps independently.

However， government-backed initiatives like Tencent’s “AI 1Ol forLocals”are trying to make a difference with hands-on training for both farmers and local officials.Recently， in northeastern China， a wall slogan that reads‘Ask Tencent's Yuanbao [AIAssistant] to Look After Your Postpartum Sows” went unexpectedly viral， reflecting the huge potential that AI literacy can have for meeting everydayruralneeds.

AI in office

InApril this year， in collaboration with state media outletsandlocal governments，Tencent launched AI training courses for over 200 grassroots officials in Zijin county， Guangdong. The program taught everything from adjusting pesticide doses to writing legal documents using theAIassistantYuanbao.

Zhao Lin， deputy party secretary of Shuangshi village in western China's Sichuan province， hasalso tried to integrate AI into everyday governance.The 37-year-old official organized a training course for local civil servants thisMarch， focusing on the use of various popular AI assistants likeDeepSeek，Kimi，andDoubao.“Weaimed to make these officials familiarwith new technologies andhelp them connect theirworkwith these tools，\"Zhao tells TWOC.

In Zhao's village， there are 2，947 residents， including five officials who report on household Although Al color-detection can identify serious issues with the crop， it often misses more subtle changes over time that any seasoned farmer would catch with a glance.

subsidies，cropyields，and farmregistrationsto 13 higher-level government offices.“The primary datawegetisauthentic，butitcan oftenbechaotic and nonstandard，\" Zhao laments.

According to Zhao， elderly villagers still keep handwritten notes，makingithard to read their records.Zhao's teamusesDeepSeek and Kimi to photograph and transcribe thesenotes into searchabledigital records.“Weusedan optical character recognition （OCR） software before，but itwasn'talways‘smart，'comparedwith thesenew tools that can extract，analyze，and quickly generate content more effectively，\"he says.

AI has also eased the burden of writing reports， particularly for less-educated village officials.“With DeepSeek， they can just put in the topic，and the AI toolwill generatea draft. They onlyneed to tweaka few details，” Zhao tells TWOC，explaining that this savesalotof time.

But the technology isn't foolproof.Earlier this year，when draftingaproposal fora millionyuan vegetable farmingproject，Zhaoasked an AI model to outline a feasible plan.“It just spit outvague，copy-pasted ideasfromother sources without any citations，\"he recalls.“It was clever. It tried to sound convincing， but turned out to be empty.\"

He notes that agricultural AI still lacks enough high-quality data to improve its performance.‘The more data it gets，the morereliableitbecomes.\"

The AI brain

Limited data，alongside issues like high system costsand low-performing devices，was also cited in April by the Economic Daily as a major hurdle to"China's smartagriculture initiatives.The report stressed thatAIaloneisunable to revolutionize farming—it must work with the“IoT， big data， and modernagricultural machinery.\"

In Zhao's village，AI toolsare built upon a broader smart farming system that was developed afewyearsago.Inthevegetable fields，sensors continuously trackhundreds of environmental indicatorsinreal time，withresearchersfrom Southwest University of Science and Technology analyzingdataonplant growth，climate conditions，andpest resistance.“Based on the numbers collected， these experts can tell us exactly what kind of water， fertilizer， and pesticide to use in each season，\"Zhao explains.‘After the harvest， AI tools like DeepSeek will make calculations on themarketprices to help us sellat thebest rate.\"

Zhao envisions AI as a“smart brain”powering a fully digital village， connected through IoT， 5G，and cloud computing. In the village，sensors gatherdata onsoil，water，and weatherdayand night，feeding it to AI models thatofferactionable insights.“Without data，even the smartestAI is useless，”Zhao says.“When these four systems work together， they generate real value for production.\"

Wang， the young farmer，agrees.He feels that AI servesmore likeanassistant，andis far from ready to replace human judgment or effort.Earlier thismonth，he discovered that althoughAI colordetection can identify serious issues with the crop， it often misses more subtle changes over time that anyseasoned farmerwould catch witha glance. Additionally，while AI excelsat routine tasks like fertilization，pesticide spraying，and weeding， irregular activities like fruit-picking and bagging still require human labor.

Wang has since started feeding data about pesticide use，weather，and soil humidity into his smart farming system to improve its predictions， and makes sure to double-check all the numbers. \"It'snotthatAI eliminatestheneed forfieldwork. You have to learn how to use AI effectively first， and gain experience in planting and related technical skills，”he says.“AI can capture the overall direction，but humaninterventionremains essential forthedetails.\"