AI breeding could run 80% of crop trial work, China Academy says

Chinese researchers say artificial intelligence could take over as much as 80 percent of the workload in certain stages of conventional crop breeding, making the process dozens or even hundreds of times more efficient. The claim comes from Sun Qixin, an academician of the Chinese Academy of Engineering and professor at China Agricultural University, who is using computer simulations to predict plant growth directly from genetic code.

What is changing in the breeding cycle?

For generations, breeders examined plants one by one in the field, measured their characteristics and repeatedly selected those with desirable traits. The method worked, but identifying a small number of superior varieties from thousands or even millions of individual plants is slow and labor-intensive.

Sun's approach replaces much of that field walking with prediction. Researchers teach AI systems to forecast what a plant will look like and how it will perform based on its genes, while also identifying the genetic characteristics behind desirable traits observed in the field. Once these models become accurate enough, they could significantly cut the need for extensive field trials and manual selection.

"Artificial intelligence will certainly bring a revolution in breeding in the long run," Sun said.

Why is this suddenly feasible?

Three foundations have to be in place, according to Sun:

• More comprehensive agricultural data, gathered by drones monitoring fields from above and multispectral equipment at ground level, recording each plant from emergence to harvest.
• Cheaper genetic sequencing. Sequencing a single wheat plant's genome to high quality once cost about 1 million yuan ($150,000); comparable work now costs about 100 yuan — roughly the price of a bag of wheat seed.
• More sophisticated models that reliably connect genetic makeup with observable characteristics.

Sun expects the technical data accumulated on major grain crops and livestock to reach a level sufficient for specialized intelligent breeding models within three to five years. He cautioned that building fully functional intelligent breeding systems will require patience and sustained investment, with about five years before their practical impact becomes widely visible.

What about roots and field machinery?

New imaging technologies and AI algorithms now allow breeders to analyze root systems precisely, identifying traits linked to high yields, efficient fertilizer use and drought resistance. In the past, breeders often had to wait until harvest to judge a variety, with limited ability to observe the roots responsible for its performance. The shift could let breeders design varieties with target characteristics from the outset rather than relying on repeated field selection.

Beyond breeding, the same technology stack — big data, satellite remote sensing, drones and autonomous machinery — is changing nearly every stage of production. On unmanned farms, digital systems coordinate machinery for planting, fertilization, irrigation and harvesting, pushing agriculture toward standardized, large-scale production.

How is policy responding?

The 15th Five-Year Plan for accelerating agricultural and rural modernization, released in June, includes emerging and future agricultural industries in national planning for the first time. It calls for scaling up intelligent breeding, new-energy agricultural machinery, the low-altitude economy in agriculture, agricultural bio-manufacturing and novel foods.

An official from the Ministry of Agriculture and Rural Affairs said initiatives will include tax incentives for scientific and technological innovation and policies supporting the first application of newly developed agricultural equipment, with local governments encouraged to back first-version agricultural technologies and software.

"Such policies could help reduce the risks faced by companies developing new technologies and encourage agricultural producers to adopt innovative products at an earlier stage," said Du Jinyi of the national rural vitalization publicity and education center.

What about training and universities?

Du said rural talent programs must expand beyond conventional planting and breeding techniques to cover smart agriculture, digital villages, green low-carbon production and practical AI applications, with training shifting from classrooms to hands-on, field-based instruction.

China Agricultural University is adjusting its programs for the next five to 10 years, adding degrees in biological breeding, food nutrition and health, green agricultural development, veterinary public health and intelligent science and technology. It plans schools specializing in bio-manufacturing and artificial intelligence, plus expanded research cooperation with Heilongjiang, Jilin and Anhui provinces. "Their value depends not only on scientific breakthroughs but also on whether farmers can use them conveniently and effectively," said university president Chen Wei.

Farmers and breeders should watch the three-to-five-year window: that is when Sun expects enough data on major grain crops to support the first specialized intelligent breeding models at scale.