Iris Holm: Tell me you looked at the actual farmer survey before we recorded this.
Lila Soto: I did, and I've been sitting with the 52% number ever since — this is Purdue University and the CME Group Ag Economy Barometer — 52% of farmers say AI tools offer no meaningful benefit. Not 'I haven't tried it.' Tried it, said no.
Iris Holm: And Congress just passed the FARM AI Act.
Lila Soto: Bipartisan, which is the other thing — Don Davis, Democrat from North Carolina, Zach Nunn, Republican from Iowa, House side. Ted Budd and Adam Schiff on the Senate companion, introduced March 21. And Davis is out there saying this is a national security issue. Which framing is doing a lot of work.
Iris Holm: It's laundering a local ROI problem as a geopolitical one.
Lila Soto: Mm, or — I guess I'd frame it as: two conversations that aren't talking to each other. Davis is asking whether America can compete. A farmer in Iowa is asking whether this software is worth the Tuesday morning it takes to learn it.
Iris Holm: The question is whether a federal bill built around USDA, AFRI grants, and a new AI in Agriculture Advisor position can actually close that gap — or whether it's downstream of a problem nobody in that bill has named.
Lila Soto: And the weird detail that opens that question — eleven percent of farmers don't even know if their tools currently use AI. So we're debating adoption when some of them can't locate the thing we're asking them to adopt.
Iris Holm: That 11% not-knowing detail actually cuts the other way from what you'd expect. It's not evidence of ignorance — it's evidence the question is wrong. Because here's what the Purdue barometer is actually showing: 52% of farmers have evaluated this and said no. That's a verdict after trial, not a waiting room.
Lila Soto: Yeah — and I think the analogy that keeps landing for me is, imagine a tool your smartest neighbor swears by. You try it on your actual plot, your soil, your market. It just doesn't click. That's not ignorance. That's a mismatch.
Iris Holm: Product-market mismatch. Full stop.
Lila Soto: And there's actual research framework behind that feeling — this 2026 Sustainability study, UTAUT and Task-Technology Fit, Midwestern farmers specifically. The primary barriers aren't 'I don't understand AI.' They're perceived benefit — or the lack of it — economic skepticism, high upfront costs, trust deficits. Which is kind of a different diagnosis than what the FARM AI Act is treating.
Iris Holm: Right — the bill builds USDA Extension capacity, funds AFRI research, creates an AgARDA mandate. Supply-side, all of it. None of that touches why a tool fails on a specific farm.
Lila Soto: Mm, and the Frontiers in Artificial Intelligence systematic review names it even more structurally — scale, trust, the digital divide. Those aren't knowledge deficits. Those are infrastructure and relationship problems.
Iris Holm: Which means — actually, wait — the 75% non-adoption number isn't the alarming one. The alarming one is the 52%. Because 75% could be early days. But 52% saying no meaningful benefit after looking? That's the evidence problem the bill doesn't name.
Lila Soto: So the question kind of becomes — are these tools being built with farm-specific conditions in mind, or built first and then sold downward? Because the mismatch you're describing doesn't fix with more outreach.
Iris Holm: Frankly, no. And that's the failure mode before the success case — you can't train your way out of a product that doesn't fit the operation.
Lila Soto: So if you can't train your way out of a product problem — what does the bill actually build? Because I keep reading the mechanisms and they're all... upstream. AFRI gets AI designated as a priority research area. AgARDA expands its mandate. A new AI in Agriculture Advisor coordinates with NIST on standards. That's infrastructure for a product the customer has already returned.
Iris Holm: NIST standards. For a farmer in central Iowa.
Lila Soto: I mean, picture it — late February, she opens a USDA Extension flyer about an AI nitrogen-optimization workshop. She already tried one of these platforms. Documentation was eighty pages. The ROI math didn't close. Another workshop is just... the same pitch with a federal seal on it.
Iris Holm: The cost barrier isn't in the bill at all. Rural connectivity gap — also not in the bill. Those are the two most concrete obstacles for small-to-medium operations and they're just absent.
Lila Soto: Which is — yeah, that's the part that sits uneasily. Because the 2026 Sustainability study on Midwestern farmers is explicit: small-to-medium operations are structurally disadvantaged. The tools scale for large commercial farms. AFRI grants historically go the same direction.
Iris Holm: Look — there is one lever in the bill that's different. USDA Extension. Local relationships, state-delivered, it's not a top-down standards document. That model might actually reach the trust problem.
Lila Soto: Oh, that's the interesting exception.
Iris Holm: But the bill's heavy investment goes to AFRI and AgARDA. Extension is listed as a delivery mechanism, not the center of gravity. So the one piece that fits the actual problem is... a footnote to the supply-side infrastructure.
Lila Soto: Mm — and there's a longer version of this pattern that I kind of want to get into, because the overestimation of precision agriculture adoption goes back over twenty years. Whether the FARM AI Act is just reprinting that same forecast with new language — that's a different and maybe harder question.
Iris Holm: The eighty-page documentation farmer doesn't need better standards coordination. She needs a tool built around her operation — not certified against someone else's.
Lila Soto: And that twenty-year pattern is the thing that breaks the framing open, because there's a 2026 academic study that actually went back and compared input dealers' precision ag forecasts against real adoption numbers across a twenty-one-year period. Not one cycle. Twenty-one years. Persistent overestimation, every single time.
Iris Holm: Every cycle.
Lila Soto: Every cycle. So the Precision Agriculture Adoption Gap isn't a temporary lag waiting for a policy push. It's — I mean, at twenty-one years it starts to look like the actual shape of the thing, not a deviation from it.
Iris Holm: Which means the FARM AI Act may not be responding to a new emergency. It might be reprinting the same forecast with a new acronym on top.
Lila Soto: Hm — and the equity piece cuts even deeper here. The 2026 Sustainability study is pretty explicit that small-to-medium farms carry the sharpest cost burden and the least technical infrastructure. But AFRI grants and a senior USDA AI Advisor — historically those mechanisms pull toward large commercial operations that can actually engage federal research pipelines.
Iris Holm: So the bill most helps the farms that need it least.
Lila Soto: Yeah. And then there's the Nebraska-Lincoln work — they tested ChatGPT-4o as a farm decision-support tool inside the TAPS farm management competition. Which is kind of fascinating as a proof of concept, but — wait, actually, that's a controlled competition environment. A small operation in practice needs something that runs on spotty rural broadband and doesn't require a research team.
Iris Holm: That's the gap. Interesting research. Nowhere near the simplicity that closes it.
Lila Soto: And the Responsible AI framework says tools have to be context-specific, built with farmer participation. NIST setting centralized standards from the other direction — those two things are just pulling opposite ways. I don't know how you resolve that from inside a federal bill.
Iris Holm: The test isn't any of that, though. Not how many AFRI projects launch, not whether AgARDA gets its expanded mandate, not whether the AI in Agriculture Advisor is seated at USDA by Q4. The test is one farmer, eighteen months from now, pointing at a specific problem AI solved — cheaper and simpler than what they were already doing. That's it. That's the whole test.
Lila Soto: And I genuinely don't know if that happens. I mean — I want it to. But if the twenty-one-year precision agriculture pattern just... repeats? The uncomfortable version of that question isn't 'what did the bill miss.' It's whether policy keeps diagnosing farmers as the problem when maybe the tools just haven't earned the answer yet.
Iris Holm: That's the one I can't settle. Frankly.
Lila Soto: Yeah. Me neither. Thanks for sitting in it with me.
