'Shit in, shit out', AI is coming for agriculture, but farmers aren’t convinced

Summary

Australian farms are at the forefront of a wave of technological change coming to agriculture. Over the past decade, more than US$200 billion (A$305 billion) has been invested globally into the likes of pollination robots, smart soil sensors and artificial intelligence (AI) systems to help make decisions. What do the people working the land make of it all? We interviewed dozens of Australian farmers about AI and digital technology, and found they had a sophisticated understanding of their own needs and how technology might help – as well as a wariness of tech companies’ utopian promises. The future of farming The supposed revolution coming to agriculture goes by several names: “precision agriculture”, “smart farming”, and “agriculture 4.0” are some of the more common ones. These names all gesture towards a future in which the relationships between humans, computing and nature have been significantly reconfigured. Perhaps remote sensing technology will monitor ever more of a farm system, autonomous vehicles will patrol it, and AI will predict crop growth or cattle weight gain. But there’s another story to tell about the way technological change happens. It involves people and communities creating their own future, their own sense of important change from the past. AI, country style Our research team conducted more than 35 interviews with farmers, specifically livestock producers, from across Australia. The dominant themes of their responses were captured in two pithy quotes: “shit in, shit out” and “more automation, less features”. “Shit in, shit out” is an earthier version of the “garbage in, garbage out” adage in computer science. If the data going into a model is unreliable or overly abstract, then the outputs will be shaped by those errors. This captured a real concern for many farmers. They didn’t feel they could trust new technologies if they didn’t understand what knowledge and information they had been built with. A different kind of automation On the other h...