Does AI help farmers get useful advice?
No good evidence. Studies so far point to a loss, but there are too few or too weak to rate it.
We show no impact figure until at least two studies qualify.
There is not enough evidence to rate this yet.
The services in these studies are run by non-profits and research teams; none reports a cost per farmer (checked 16 Nov 2026).
What the studies found
No impact figure yet. Fewer than two studies qualify, so under our rules we do not show how big the effect is. The studies we found are listed below, with what each measured.
How sure are we?
- Starting level0 of 5, from 1 qualifying study: fewer than 2.
- Result0 of 5. There is not enough evidence to rate this yet.
Behind the average
- AI-generated content read in a human voice (arm T2) (Precision Development (2026)): listening rate 3.5 pp lower than status quo (6.3% decrease, p<0.05); duration heard 9.6 s lower (15.9% decrease, p<0.01)
- Advisories of nearly equal length across arms (2-3 messages) (Precision Development (2026)): no notable difference in listening rates or durations heard. The entry contradicts itself on length: Results say AI advisories are '10-13 seconds or 12% shorter' than human ones (mean 107 s), while the abstract says 'AI-generated scripts tend to be longer'
- Farmers who recalled the advisory (phone survey) (Precision Development (2026)): no significant difference in comprehension between AI-generated and human-generated advisories
How, where and why it works
1 of 4. AI can write advice faster or answer farmers' own questions; whether farmers listen and act on it is the open question.
0 of 4. No study yet reports who gains most.
1 of 4. The counted trial used AI to write and voice phone advisories that agronomists had written before.
0 of 4. No study counts cost.
What to do next
Measure your pilot →If you are trying it yourself
- Check AI farm advice against your local extension service.
- Try one change on part of a field before the whole farm.
If you are rolling it out
- Test AI-written advice against your current advice with a random half of farmers.
- Track what farmers hear, remember and change, and later yields.
What it takes
- Have an agronomist check AI-written advice before it goes out.
- Measure whether farmers listen to and remember the advice, not only how much you send.
What can go wrong
- In the counted trial, farmers listened to AI-written advisories slightly less and remembered less.
- Advice that is wrong for local soils, pests or weather can cost a season.
The studies
| Study | Grade | Design | People | AI model (released) | Company role |
|---|---|---|---|---|---|
| Paddy AI A/B test 1: Artificial Intelligence versus Human-Generated Advisories Precision Development (PxD), with the Coffee Board of India, 2026. PxD experiment registry entry (registered trial) | C | field RCT, deployed work | 28,796 coffee farmers in Karnataka randomised to 3 arms (stratified by district and gender); follow-up phone survey of 4,498 farmers. The AI-assisted workers are PxD agronomists producing the advisories. | PaddyAI (PxD assistant that generates, records and translates advisories; underlying model not named) | vendor-run: PxD built PaddyAI and ran and reported the test itself on its own service (Coffee Krishi Taranga) |
Studies we found but do not count
- AI Advisory for Smallholder Farmers: A Randomized Field Experiment in the Philippines and India (Aouad et al. (2026), grade C). Only the abstract could be read: SSRN blocks automated access, and no other copy was found, so attrition and how outcomes were measured could not be checked. It will be counted once the full paper can be read.
Does the rating hold up?
- Run without company-run studies0 studies: strength 0, verdict no good evidence.
- Run without working papers and preprintsNo change: no such studies.
How up to date is this evidence?
Method →Evidence searched to 2 Nov 2026. The studies do not all state model release dates. Rating computed 16 Nov 2026 under Module 15 v0.3 (pilot).
How this page was made: AI research agents searched for studies and read each paper; a second pass checked every number on this page against the paper itself, with the quote and where it appears. The rating is computed by fixed rules, not chosen. No person outside AI Adoption has reviewed it yet. Editorial owner: AI Adoption editors. Spot a mistake? Corrections.
Disclosure: AI Adoption uses AI tools, including Claude from Anthropic, in its research and editorial work. No company pays for or has any say in a rating.