Chinese researchers are using AI and genetic data to predict how plants will grow, a method that could make some stages of crop breeding hundreds of times faster. The work points to a practical use for AI in agriculture: narrowing the search for better crops before years of field trials do the sorting.
Watch Desk analysis
What happened
China Daily reports that Sun Qixin, an academician and professor at China Agricultural University, is using computer simulations to predict plant growth from genetic code. Sun says AI could eventually take over as much as 80 per cent of the workload in certain stages of conventional breeding, reducing the need for extensive field trials and manual selection. Those are projections, not measured results from a deployed system.
The report describes a striking change in one of the inputs: sequencing an individual wheat plant’s genome, once about 1 million yuan, can now cost around 100 yuan, according to Sun. He says better agricultural data, cheaper sequencing and models that connect genes to observable traits are all needed to make intelligent breeding work.
Researchers are also using drones and multispectral equipment to collect information on plants throughout the growing season. AI and imaging could help identify traits linked to yield, efficient fertiliser use and drought resistance, including in root systems that are difficult to assess before harvest.
Why it matters
Breeding a crop means finding a small number of promising plants among thousands or even millions. If models can make reliable predictions from genetic and field data, breeders could focus physical trials on stronger candidates rather than measuring and selecting every plant by hand. That is the useful promise here, and the reason sequencing costs matter: more affordable genetic data could make it practical to study far more plants.
The report also gives a reality check. Sun says specialised systems are beginning to address particular breeding tasks, but fully functional intelligent breeding systems will take sustained investment, with their wider practical impact potentially about five years away. Lower sequencing bills help; they do not magically supply good data or a model that predicts performance reliably.
Our read
This is a substantial AI application with a clear job to do, not a model looking for a problem. The compelling possibility is faster, more targeted crop selection, especially where drought resistance and efficient use of fertiliser matter. But the hundreds-of-times speed-up is a prospect, not a demonstrated result. The test will be whether predictions lead to better varieties in farmers’ fields, at a cost and level of reliability that make them useful.
What to watch
- Whether researchers publish measured results from breeding trials using these systems.
- Whether genetic and field data improve enough to support reliable predictions across crops and growing conditions.
- Whether the technology becomes affordable and practical for farmers, rather than remaining a promising research tool.
Discussion spark: If AI can narrow the search for better crops, should breeders trust its predictions to reduce field trials, or keep broad testing as a safeguard against missing useful varieties?
Sources and evidence
- AI to drive next farming revolution – Chinadaily.com.cn (5 November 2026, 00:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.