In the rapidly evolving world of agriculture, artificial intelligence (AI) is making significant inroads, promising to revolutionize how farmers receive and implement crop advice. While AI has long been a staple in various industries, its application in agriculture is still a burgeoning field. The question remains: can AI-based advisory tools truly support farmers, and if so, how accurate and reliable are they?
The advent of AI chatbots like ChatGPT has sparked widespread curiosity and discussion. These chatbots, powered by large language models (LLMs), can generate and understand human language, making them potential tools for providing crop advice. However, there are significant caveats to consider. For instance, free versions of these chatbots often use user input to train themselves, raising concerns about data privacy and security. Paid versions, on the other hand, offer more protection but come at a cost.
The history of AI dates back to 1950 when mathematician Alan Turing developed the Turing test. Since then, AI has evolved significantly, with notable milestones including IBM’s Deep Blue defeating chess grandmaster Garry Kasparov in 1997 and the development of Watson, a computer capable of communicating in natural language. Watson’s capabilities were further enhanced when IBM acquired The Weather Company in 2016, using its data for weather forecasting, which is crucial for agricultural businesses.
Chatbots, a form of generative AI, have also been around for decades. Eliza, developed in the mid-1960s, is considered the first chatbot. Today, chatbots are ubiquitous, from virtual assistants like Amazon Alexa and Apple Siri to product recommendations in online shops. ChatGPT, launched in autumn 2022, is perhaps the best-known chatbot, capable of generating text based on user input.
But how useful are these chatbots for crop advice? Ardon Verschoor, an arable advisor at Van Iperen, experimented with ChatGPT but found its recommendations unsuitable for specific crop protection needs. He sees potential in automating office work and assisting with subsidy applications, but not for technical crop advice. Similarly, independent advisor Gerard Meuffels has not seen AI chatbots in practical use and notes that many farmers are unsure about their reliability.
However, there are signs of progress. AI Ag Advisor Norm, developed by the U.S.-based Farmers Business Network, is likely the first sector-specific AI chatbot using large language models. This chatbot is designed to provide more tailored and reliable advice, addressing some of the limitations of general AI chatbots.
Arable advisor Luc Remijn from Delphy has experimented with various prompts, finding that chatbots can provide useful information about insect development and pest descriptions. However, he notes that new internal policies have restricted the use of ChatGPT, leading him to switch to Microsoft Copilot. Gert Sterenborg from Maatschap Sterenborg finds the answers too generic and vague, but sees potential in guiding the prompts for more tailored advice.
The number of specific agricultural applications is increasing, with sector-specific chatbots and models addressing the reliability and data privacy concerns of general AI chatbots. The U.S.-based Farmers Business Network was likely the first to launch such a tool, setting a precedent for more tailored and reliable AI advisory tools in agriculture.
As AI continues to evolve, its role in agriculture is likely to expand. While current AI chatbots may not be ready to replace agronomists, they offer valuable assistance in automating office work and providing general information. The future of AI in agriculture looks promising, with the potential to revolutionize how farmers receive and implement crop advice. However, it is essential to approach AI with caution, ensuring that data privacy and reliability are prioritized.