AI in agriculture: 17% of farmers use it, 6% trust it
17% of farmers use generative AI. Only 6% trust it.

17% of farmers now use generative AI for farm tasks.
only 6% name it as a trusted source for decisions. 56% still trust their agronomist.
that gap is the most useful thing to understand about AI in agriculture right now.
adoption is real, trust isn't there yet, and the people who can close that gap are exactly who farms and food companies will want to hire.
What's actually happening with AI in agriculture
the big numbers come from McKinsey's Global Farmer Insights 2026, a survey of 5,500 farmers across 10 countries:
→ 17% use gen AI for farm-related tasks (Claims Journal) → 6% cite gen AI or AI search as a trusted decision source, against 56% for technical agronomists (RFD-TV) → 11% of US farmers say they pay for AI tools → spending intent fell 24 percentage points, which McKinsey linked to higher costs
so farmers are curious, but money is tight. keep that in mind for everything below.
on the machinery side, John Deere launched JD, a generative AI assistant inside its Operations Center, in September:
- farmers ask questions like "how did fuel use in tillage compare to last year?"
- it answers using only that farm's own historical data
- it's free to Operations Center users, with early access this autumn
- Deere hasn't said which AI models sit underneath (and reserves the right to change them)
and the sceptical view is already out there. AgNavigator asked whether AI demand in farm machinery is being overestimated, noting a Canadian maker of low-tech tractors was reportedly sold out through October.
How AI is used in the food supply chain
further down the chain, the wins look less like robots and more like better ordering.
Finnish retailer Kesko says its K-food stores cut food waste by about one million kilograms in 2025, using forecasts built on sales data, campaigns, seasonality and local weather.
interesting detail though: the release never says "AI". it says forecasting.
(sometimes the impressive result is just very good data discipline.)
and globally, the World Bank mapped 60 AI use cases across the agrifood chain, whilst warning that global models often overlook local crops, languages and supply chains.
small-scale producers grow a third of the world's food, so that's not a small gap.
What uni won't teach you about AI in farming
agriculture and food science degrees are great on agronomy, soils and animal science. where they fall short:
→ farm data is treated as someone else's job. yet tools like JD only work if the data going in is decent. → tech is taught as features, not margins. a farmer cutting spend doesn't care about "AI powered". they care about cost per hectare. → trust is skipped. the 6% vs 56% gap is a people problem, not a software problem. nobody teaches you how an agronomist and an AI tool should work together. → context gets flattened. the World Bank point about local crops is a reminder that a model trained somewhere else can be confidently wrong on your farm.
The critical-thinking move: 5 questions for any AI farming tool
use these on placement, in a dissertation or when a sales rep shows up at the farm gate:
- whose data is it, and who else sees it? JD runs on the farm's own records. ask where that data lives, who can access it and what happens if the farmer switches brands.
- what's the margin maths? turn every claim into money per hectare or per tonne. with spending intent down 24 points, a tool that can't pay back this season is a hard sell.
- does it replace the agronomist or back them up? farmers name agronomists as a trusted source nearly ten times as often as AI right now. the tools that work with that trust will win (radiologists have heard this question before).
- was it trained on farms like this one? crop, climate, scale, language. the World Bank warning about local crops applies to every model.
- is it AI, or just good data? Kesko's result came from forecasting. always ask what the simplest method behind the result is. sometimes a spreadsheet does most of the work.
What to do this month
- read the McKinsey survey summary and pull out three numbers you'd quote in an interview. most applicants won't have read it.
- test a gen AI tool on a real farm question. ask it about a crop or livestock issue you studied, then check the answer against your lecture notes or an agronomist. write down where it was wrong.
- build a tiny forecast. take any public crop yield or food price dataset and forecast next season in a spreadsheet. then try a simple model in Python and compare.
- talk to one farmer or agronomist about which digital tools they actually use and which ones they ignore. that conversation is worth more than another lecture slide.
so TLDR: AI in agriculture is spreading fast but trust and margins decide what sticks. learn to ask about the data and the money, not just the features.
if an AI tool and your agronomist disagreed about when to spray, who would you listen to?
Sources
- Global Farmer Insights 2026McKinsey & Company · 08 SEP 2026
- Farmers Are Embracing AI More Than Any Other Tech, McKinsey SaysClaims Journal (Bloomberg) · 09 SEP 2026
- Farmers Adopt AI While Still Trusting Agronomists MostRFD-TV · 22 SEP 2026
- From spreadsheets to AI: Deere gives farmers new features in Operations Center, JDThe Robot Report · 01 SEP 2026
- Where is innovation heading in the ag machinery market? Why the future isn't just AIAgNavigator · 08 SEP 2026
- Food waste reduced by one million kilograms: K-food stores' order volumes managed through forecastingKesko · 23 APR 2026
- Harnessing Artificial Intelligence for Agricultural TransformationWorld Bank · 25 NOV 2025
Get the next issue by email
theVANTAGEd is written by students to bring critical thinking to AI across real industries. One issue every two weeks.