In this episode, Johann Tasker investigates what AI could mean for crop management.

He travels from a Bedfordshire wheat field to the banks of the Thames, an agronomy office in Essex, farm trials in East Anglia and to Cambridge University.

Arable grower Andrew Robinson is already putting AI to the test by allowing it to devise a fungicide programme for his winter wheat.

Machine-learning specialist Freddie Alan Stevens explains why general-purpose chatbots can give convincing but wrong answers – and why specialist agricultural AI could be much more powerful.

Independent agronomist Alex Hall considers where human judgement still matters – and how AI could help human agronomists do a better job.

And Agrii’s Jonathan Trotter and Sam Fordham reveal what happened when they put technology head-to-head with conventional farm decision-making.

Finally, former Mineral chief executive Dr Elliott Grant looks ahead to agentic AI – systems capable of investigating problems, using specialist tools and potentially taking action themselves.

Chapters

00:00 – Can AI manage a wheat crop?
Andrew Robinson explains why he handed the fungicide programme for half a hectare of winter wheat to AI – and why the resulting crop caught his attention.

03:01 – Putting ChatGPT to the agronomy test
Can a chatbot diagnose yellowing wheat? Johann Tasker tests what ChatGPT can – and cannot – do when faced with a real-world crop problem.

05:58 – What is AI actually doing?
Machine-learning specialist Freddie Alan Stevens explains how large language models work, why they hallucinate and the risks of relying on confident-sounding answers.

09:30 – Why specialist agricultural AI matters
From wheat counting and disease recognition to slug prediction and satellite interpretation, could purpose-built models prove more useful than general-purpose chatbots?

12:36 – What does an agronomist bring?
Ceres Rural partner and independent agronomist Alex Hall discusses experience, judgement, risk and the importance of understanding an individual farm business.

16:18 – Can AI give agronomists more experience?
A farmer gets only one harvest each year. Could machines learn from thousands – and help advisers draw on far more information than any individual could absorb?

19:27 – Man versus machine
Agrii’s Jonathan Trotter and Sam Fordham reveal the results of a field experiment comparing technology with conventional decision-making.

22:47 – The connected farm
How sensors, drones, satellites, weather information, research data and AI could increasingly work together to support farm decisions.

24:42 – Bias, trust and ground-truthing AI
Who built the system? What was it trained on? And could commercial interests influence its recommendations? Jonathan Trotter explains what growers must check.

27:45 – The next leap: agentic AI
At Cambridge University’s Whittle Laboratory, Dr Elliott Grant explains the arrival of AI agents capable of reasoning, using specialist tools and carrying out tasks autonomously.

31:36 – What happens to the agronomist?
As routine knowledge becomes easier for machines to reproduce, specialist expertise, proprietary knowledge and an understanding of individual farms could become increasingly valuable.

34:13 – Back to the AI wheat trial
Andrew Robinson reveals how Microsoft Copilot devised its fungicide strategy – including finding the wrong weather forecast – and why the AI programme cost more than his farm standard.

37:11 – Asking AI the right question
Cleanest crop, highest yield, best margin, lowest environmental impact or least risk? The answer an AI produces depends heavily on the objective it is given.

38:30 – Farmer, agronomist and AI
What have the experiments and interviews taught us? Johann brings together the evidence before asking ChatGPT the original question one final time.

This episode is brought to you by LG Seeds UK – providing innovative, reliable seed solutions to farmers in a changing world.

Special Guests:

Andrew Robinson – farms director, Heathcote Farms. Andrew is trialling an AI-generated winter wheat fungicide programme alongside the farm’s conventional approach.

ChatGPT – ⁠the general-purpose AI assistant tested throughout the episode.

Freddie Allen-Stevens – machine-learning and AI specialist at Insight ML, with experience developing agricultural applications including crop counting and slug-risk prediction.

Alex Hall – partner and independent agronomist at  ⁠Ceres Rural in Saffron Walden. Ceres provides independent crop-management and agronomy advice.

Jonathan Trotter – part of  ⁠Agrii’s research and agricultural technology work, exploring how digital technologies and agronomy can be combined in practical crop management.

Sam Fordham – works in Agrii’s digital agriculture operation, including its Contour platform and the integration of farm, crop, weather and remote-sensing data.

Dr Elliott Grant – technology and agriculture entrepreneur who led agricultural AI work at Google X and later became chief executive of Mineral. His work has focused on the intersection of AI, technology and agriculture.

Useful links

⁠ChatGPT – the general puropose large language model from OpenAI.

⁠Agrii digital farming and precision agriculture – including information about Agrii’s RHIZA and Contour digital agronomy services.

Insight LM – machine learning consultancy services to help businesses leverage data-driven insights.

⁠Ceres Rural – information about Ceres Rural’s independent agronomy and crop-management services.

⁠Dr Elliott Grant – background on Elliott’s work in agricultural technology and AI, including his work at Cambridge University's Whittle Laboratory.

This episode of Farm Insight was written, edited, produced, sound-designed and hosted by Johann Tasker for Farmers Weekly.

To get in touch with Johann: linkedin.com/in/johanntasker/

For Farmers Weekly, visit fwi.co.uk or follow linkedin.com/company/farmers-weekly

For sponsor or advertising opportunities, email podcast@fwi.co.uk. In the UK, you can also text the word FARM followed by your message to 88 44 0. 

See omnystudio.com/listener for privacy information.