■ MODERATE RISK ■ Agriculture & Environment
Not replaced, but heavily mechanised further — farming has been automating for two centuries and the result each time has been fewer people producing more food. Expect the same again.
“Precision agriculture and robot harvesters are growing. But weather, soil, and instinct still need a farmer.”
Our AI replacement risk score — how we score jobs
Farming is a business of decisions made against biology and weather. When to drill, which variety, how much nitrogen given the price and the forecast, whether to cut silage before the rain, when to sell grain, when to cull, how to service the debt on a combine that runs six weeks a year. Underneath that sits an enormous amount of physical work — machinery maintenance, fencing, livestock handling, harvest logistics — and a farm is also an equipment repair shop with a crop attached.
Precision agriculture is already deeply embedded. GPS-guided and autosteer tractors are standard on arable operations, variable-rate application adjusts inputs metre by metre, drone and satellite imagery flags disease and stress before it is visible from the cab, and computer-vision sprayers target individual weeds and cut chemical use dramatically. Robotic milking parlours run without human presence at milking, and autonomous tractors are commercially available. Harvest robotics for soft fruit and vegetables has been slower and expensive, but labour shortages in horticulture are pushing hard on that problem.
What resists is judgement under uncertainty and the physical unpredictability of biology on uneven ground. Livestock get sick in idiosyncratic ways; a fence gets flattened; the ground is too wet in one corner. Farmers also absorb risk — commodity prices, weather, disease, subsidy policy — in ways no algorithm assumes. The realistic outcome is what has already been happening: fewer, larger, more capitalised farms with fewer workers each, and a farmer whose day includes more screen time than seat time. Our risk score of 35 tracks continued consolidation more than sudden displacement.
Automatability: our editorial assessment of current and near-term AI capability
Transformed but not eliminated by 2040. Autosteer, variable-rate inputs, and imagery-based monitoring are already standard on larger arable operations; robotic milking is mainstream in dairy. Through the 2030s expect autonomous field machinery and vision-based weeding to become normal on well-capitalised farms, and horticultural harvest robotics to mature under labour-shortage pressure. Farm numbers keep falling and farm sizes keep rising — the long-running trend that automation has always accelerated.
They reduce labour per hectare, which is the same thing tractors did to horses and combines did to harvest gangs. The farmer — the person deciding what to grow, when to sell, and how to finance it — remains. Hired field labour is the exposed category, and in most developed markets that labour is already scarce, which is why the machines are being bought.
Viable but capital-intensive and consolidating. Entry without inherited land is genuinely hard, and the economics reward scale. The growth areas are agronomy, agricultural technology, contracting services, and specialist high-value production rather than conventional commodity smallholding. If you love the work, get technical — data and machinery skills are what differentiate operators now.
On narrow, data-rich questions — nitrogen rate, disease risk from weather models, optimal harvest timing — decision-support tools are genuinely strong and often beat intuition. On the messy composite judgements, involving cashflow, a broken machine, a weather window closing, and a contract deadline, no. And nobody has built the model that absorbs the financial risk when it is wrong.
Routine field operations and repetitive manual harvesting, where robotics and autosteer already work or are close. Milking parlour labour has largely gone in automated dairies. Persisting longest are livestock husbandry with real animal handling, machinery maintenance, and the management layer, since farms are becoming more complex businesses rather than simpler ones.