34 Follow us on Twitter @BetterFarmingON Better Farming | February 2024 PREPARING FOR ARTIFICIAL INTELLIGENCE Exploring AI use on-farm. By Dale Cowan crops: yield matter$ It seems everywhere we turn someone is talking about artificial intelligence (AI) and all its various applications. In agriculture, huge investments are being made in AI platforms ranging from equipment companies to crop inputs. ChatGPT is probably the most used and accessible platform to most anyone with a computer or mobile device and couple of dollars of disposable income per month. Ask the platform a question and it will come up with an answer from the Internet on anything that is readily available, archived and can be found. ChatGPT at present lacks any reasoning; it only regurgitates what it has found and does a nice job of presenting its findings, but needs some critical thought before it should be acted upon. What might it actually mean for a farmer if we fast forward a few years? How would AI be used every day or at least weekly on the farm? Let’s ask Auricle AI a question. “Based on my current crop of soybeans (name variety) planted May 15, my present soil test information, tillage practice, ground cover, soil type, microclimate in the field, weather forecast, what is the likelihood of white mould developing? Give me a time frame and select the products required with rates, nozzles and volume to be applied and provide an ROI.” The programming that makes this happen is referred to as a large language model. This supports questions being asked of the data available to the model, which means anyone can ask the question without having to be a computer programmer. In a matter of a few seconds, an answer is returned. Wow, how great would that be? There are companies that believe this is all possible. The answer will come from many linked databases, ranging from your own farm data, fungicide manufacturer, nozzle company, sprayer company, weather station data services, your local input supplier, your agronomist (maybe). All these sources of data need to be seamlessly connected to support a decision. The real question is, will any of them replace your trusted advisor? To make this a reality, we need to examine where we are today and look at the gaps in connected data. To start, the foundational piece is that businesses have customers. They have farms that have fields that need a georeferenced boundary. Without this fundamental piece, nothing gets connected. How many farms – with all of their fields with georeferenced boundaries – are sitting in one database connected to the cloud? Next is a background map of soil type and topography that can be linked to a field boundary seamlessly. After that is the recording of the following field activities: Tillage; fertility; applications of inputs with dates; rates; products and method (banded, broadcasted incorporated and to what depth); hybrid varieties; plant population dropped; plants emerged; growth models tracking crop growth stage; AI may help producers improve efficacy and sustainability on-farm, but take the proper precautions. Bits and Splits - tock.adobe.com photo
RkJQdWJsaXNoZXIy NTc0MDI3