AI for farmingBeginnerLesson 1255 min read

Data, privacy and fairness for farmers

Farm apps collect field locations, photos, yields and phone numbers. Learn who should control that data, what real consent looks like, and how tools can leave small farmers and offline villages behind.

✦ Ask the tutor 🧪 Lab · 20 min 📝 Lesson test

Speed

Lesson in motion

In 60 seconds

Data, privacy and fairness for farmers

Farm apps collect field locations, photos, yields and phone numbers. Learn who should control that data, what real consent looks like, and how tools can leave small farmers and…

1/4

In simple words

If a neighbour borrows your notebook of farming secrets, you want to know why, and you want it back. Farm apps also "borrow" information about your field. You should know what they take, why, and be able to say no.

Farm apps and services can collect a lot about a farmer: name and phone number, the exact location and size of the field, photos, what was sown, how much was harvested, and sometimes bank or land records. This data is useful. It is also valuable to companies, and it can be misused.

Who should control farm data?

The data describes the farmer's land and work. A fair principle is that the farmer should know what is collected, agree to it, be able to see it, and be able to take it back or ask for it to be deleted. India has a data protection law, the Digital Personal Data Protection Act, 2023, about personal data and consent. Check the current rules for how it applies to your service.

What real consent looks like

Fake consent

A long English form nobody reads
"Agree" is the only way to use the app
Data is shared with other companies without saying so
No way to delete your data later

Real consent

Short, clear words in the farmer's language, or spoken aloud
The farmer can use the main service and still say no to extra sharing
Clearly says who will get the data and why
The farmer can see, correct and delete their data

How tools can be unfair

Who may be left out	Why	What builders can do
Small and marginal farmers	Tools may be tested mostly on large, well-recorded farms. Small plots can be too small for satellite pixels.	Test on small plots; check accuracy separately for small farms
Farmers with little or no internet	Apps may need fast data or a new phone	Offer SMS, voice calls, offline modes and help through farmer groups
Women farmers	Land and phones may be in a male relative's name, so they may not be counted as users	Design for shared phones; ask women farmers directly in testing
Speakers of less common languages	The app may only work well in a few big languages	Test in local languages and dialects with native speakers
Farmers growing local crops and varieties	Training data may cover only popular crops	Collect data on local crops with farmers' consent

Note

Fairness is not only about the AI model. It is also about who can reach the tool, who can understand it, and who benefits from the data. A tool that only helps farmers who were already well off can widen the gap.

Watch out

Be careful when a farm data app is linked to loans, insurance or buying and selling. A mistake in the data, like a wrong crop area or a wrong yield estimate, could affect decisions about a farmer. Farmers should be able to see and correct data that is used to make decisions about them.

Real example

A farmer producer organisation in Maharashtra wants to use members' field data to plan group buying and selling. Before starting, they hold a village meeting. They explain in Marathi what they will collect and why, who will see it, and that members can say no. They keep the data within the FPO and share only totals with buyers.

Try this. Open a farm app you or your family use. Can you find what data it collects and who it shares it with? Can you delete your account? Write down what you found.

Watch and read more

Lab

A one-page, plain-language data promise for a farm app, and a list of who the app might leave out.

~20 min

The problem

This is a writing exercise. Imagine you are building a farm app that takes leaf photos and field locations.

Write a "data promise" to farmers in 8–10 short sentences in simple words: what you collect, why, who sees it, how long you keep it, and how to delete it. Write it so it could be read aloud in a village meeting.
List what the app will not do with the data.
List three groups of farmers who might be left out by your app, and one change for each that would include them.
Ask one farmer or family member to read or listen to your promise. Write down one thing they did not understand, and fix it.

You are done when

The promise says what is collected, why, who sees it, how long it is kept, and how to delete it
Farmers can say no to extra sharing and still use the main service
Three groups at risk of being left out are named, each with a concrete change
A real person's feedback was used to make the wording clearer

Hard questions

Try to answer before you reveal. If you can answer these, you understood the lesson.

Q1Pooling data from many farms helps everyone get better pest warnings. How can you get this benefit while protecting individual farmers?Reveal

Ask for clear consent to pooling, and let people say no. Share only what is needed, such as the village or block and the pest seen, not the exact field or name. Publish only totals or maps at an area level. Give the results back to the farmers who contributed, so they benefit too. Keep the raw data secure and delete it when it is no longer needed.

Q2Your app is 90% accurate overall. Why might that hide a fairness problem?Reveal

The overall number can mix groups. If most test photos came from large farms in one region, the app could be very accurate there and much worse for small farms, other regions, other languages or local varieties. Measure accuracy separately for each group, and include those groups in the test data before claiming the app works for everyone.

Please sign in to continue.

Questions people ask

Is sharing farm data always bad?

No. Sharing can help. For example, pooled data from many farms can improve pest warnings for a whole area. The important things are that farmers agree, know what is shared, and also get benefit from it.

What is 'anonymised' data, and is it safe?

It is data with names removed. But exact field locations can still point to a person, especially in a small village. Removing names is not enough on its own. Share only what is needed, and prefer totals over farm-by-farm details.

Can a company sell my farm data?

It depends on what you agreed to, the company's policy and the law. Read what the app says about sharing, and ask your FPO or agriculture officer if you are unsure. Choose apps that let you say no.

How can a startup make its tool fair for small farmers?

Include small farmers in design and testing from the start. Measure accuracy separately for small plots and different regions. Offer low-data and voice options. And keep prices and requirements within reach.

Lesson test

5 questions. Get 3 right (60%) to pass and complete this lesson.