Exploring the capabilities of the AI regenerative agronomy tool FieldLark (FieldLark.ai) — and why we still need books, magazines and consultants
“Asking the earthworms…”
“Listening to the loam…”
“Drawing from deep roots…”
“Synching with the field…”
This is what users of FieldLark see — instead of the internet’s typical “Be patient!” spinning wheel, after they type in a query. It’s the first hint that FieldLark is trying to be something different: a very modern technology that seeks to not alienate humans from the ecosystems they live in.
FieldLark is a recently released artificial intelligence tool designed to serve regeneratively minded farmers and agronomists. Like ChatGPT, it receives questions written in plain human language and provides answers in common terms. Unlike ChatGPT and other large-language models, FieldLark was trained on data not from across the entire internet but specifically from regenerative sources, including over a decade of data and crop recommendations from Advancing Eco Agriculture. According to AEA, FieldLark is “designed for growers who are exploring or practicing regenerative agriculture and want quick answers to practical questions.”
For full disclosure, AEA’s founder, John Kempf, is the executive editor of this magazine, but he was not consulted for this review. We received early access to FieldLark and have been using it on our own farms for several months now, and our assessment — that it’s a valuable tool that eco-farmers should take advantage of — is purely our own.
Models like ChatGPT are great for a number of things. They can do what Google does, just faster. They can give an intelligible answer to “Create a Dipel spray for cabbage with a 48-ounce sprayer, also including fulvic acid, manganese sulfate, and zinc sulfate.” Dipel is an approved organic Bt spray, and I don’t mind using it on my cabbages when absolutely necessary to save them. But FieldLark, when asked the same question, demurs. It first asks for more information: What specific pests do I have? What stage is the cabbage in? Where are you located? What have you observed about the damage? It then tells me, “Once I have a clearer picture, I can offer more targeted short-term treatments and long-term prevention strategies.”
This is where FieldLark, with its regeneratively focused knowledge base, excels. It seeks to provide long-term solutions — not just the short-term fix the user might want. And these long-term recommendations always consider biology, not just chemistry. Users can input their soil and plant tests to receive specific recommendations or can upload photos for help identifying specific diseases. Using FieldLark becomes somewhat addictive — when you walk through your farm and observe things, you can now get answers to all of the agronomic issues you’re curious about.
FieldLark helps users engage in first principles reasoning, as opposed to analogical thinking. Both first principles and analogical reasoning are important. While first principles reasoning starts with “What do we know for sure, and what can we build from that?” reasoning by analogy builds on precedent and asks, “How has this been done before, and can we copy or adapt that?” If I want to know how to change the oil in my car, I want an analogical answer: it always works to unscrew this drain and add this much oil. But if I need to figure out why my car is making a certain noise, I need to understand how the engine works in the first place. This is why we have mechanics: most people prefer to pay someone to have first principles knowledge of automobiles.
ChatGPT mostly uses analogical reasoning. FieldLark doesn’t just take your question and answer it, though; it explains how the plant-soil-biology world works. It describes the mechanisms that are involved in whatever problem you have. It’s designed to be more concerned about the “why” than the “what.”
An obvious question becomes, “If we have FieldLark, why do we need agronomists — or books — or this magazine?” Although I’m admittedly biased in my assessment that we do still need all of these things — particularly books and this magazine — tools like FieldLark ultimately help growers ask better questions of their agronomists. In a sense, it allows you to ask the question you think might be a dumb question, without the fear of thinking your agronomist will judge you for not knowing! This, in turn, allows the grower to ask even better questions when interacting with their agronomist.
And, in our opinion, traditional resources like books and magazines absolutely retain their value. They provide the long-form, in-depth discussion of topics that 500- or 1,000-word AI-generated responses can’t. And magazine articles in particular expose growers to a variety of topics and ideas that they may otherwise not come across. AI tools like FieldLark are the ultimate in specific knowledge; books, magazines and conferences provide the broad-based knowledge that forms the foundation of our understanding and that enables us to ask more specific questions.
It turns out that earthworms have a lot of wisdom to offer to ecological farmers.
