AI isn’t likely to wipe out all farming jobs – but it is changing who bears the risks
- Written by Sophia Duan, Associate Dean, Research and Industry Engagement, La Trobe University
The global economy is bracing for major job disruption as artificial intelligence (AI) advances and spreads across industries. Experts have been warning about[1] this shift for years, and fiercely debating whether the benefits of an AI revolution will outweigh the cost of mass displacement in the workforce.
Few sectors expose this tension as clearly as agriculture. Pressure on farming is intensifying. Global food demand is projected to rise by 35–56%[2] by 2050, driven by population growth, urbanisation and changing diets.
This helps explain why AI is increasingly promoted as a productivity solution[3] to produce more food with fewer inputs, under more volatile conditions.
Yet on farms, enthusiasm for AI is often tempered by caution. And that caution is not simply about whether jobs will disappear. A deeper concern is risk, and who bears responsibility if the technology fails.
Technological change
Agriculture is not a controlled environment. Farming is biological, dynamic and deeply context-dependent, shaped[4] by weather, soils, ecosystems and animal behaviour. Because of this complexity, AI is (and will continue to be) rarely used to replace people outright. Instead, it automates specific tasks.
Automation[5] has been a big part of the farming story for decades, long before AI arrived on the scene. From mechanised harvesting and GPS-guided tractors to automated milking systems and variable-rate fertiliser application, technology has gradually changed how farms operate.
But AI is different. Rather than replacing farmers, AI is mainly being used to support decision-making in environments that are too complex, variable and context-dependent to be fully automated.
Most current uses of AI support monitoring and optimisation[6]: detecting crop stress from satellite imagery, predicting irrigation needs, tracking livestock behaviour or flagging disease and weed risks[7]. Farmers and farm workers still interpret the information and decide how to respond.
A clip from an ABC Science documentary showing AI robots being used in cattle farming.AI is automating tasks, not whole jobs
Our previous research with colleagues from CSIRO’s Data61 explored the future of Australia’s agricultural workforce[8], showing how digital and automated technologies are changing how farm work is done rather than removing the need for people. Demand is growing for skills in decision-making, oversight and technology management, particularly as labour shortages persist. However, adoption of advanced technology in farming remains limited[9], especially among small producers.
It’s a similar story internationally. For example, in the United States, only around 25%[10] of farms used advanced technology by 2019, with adoption of automatic steering and machinery guidance systems more common on larger operations. These technologies automate specific tasks and can reduce fatigue, but they do not eliminate farm operators.
Across other industries, evidence from the International Monetary Fund (IMF) shows about 60% of jobs[11] in advanced economies are exposed to AI.
Separate findings from the OECD indicate AI exposure is primarily at the task level, with only about 27–28%[12] of employment currently in occupations at high risk of full automation.
Uneven gains
The productivity promise of AI and other types of digitalisation in agriculture is genuine. In practice, however, these gains are uneven.
Evidence[13] shows adoption and benefits vary widely by farm size, crop type, region, and access to capital, data and skills. It also risks leaving some farmers behind[14] due to poor connectivity and limited digital access, constraining their use of data-driven and AI-enabled tools.
Read more: Logged out: farmers in Far North Queensland are being left behind by the digital economy[15]
Risk and reward
This is where the core tension lies. When AI-supported decisions succeed, efficiency improves. When they fail, humans carry the consequences.
For example, if an irrigation system mistimes watering, the farmer bears the yield loss or soil damage. If a particular crop disease is missed, a whole season’s income may be wiped out.
AI systems do not absorb financial loss, regulatory scrutiny or reputational damage. Farmers and advisers do. This dynamic is central to our research[18] through the Australian government’s Soil CRC program on how easy it is for farmers to actually adopt these new technologies.
That work shows farmers assess technologies not just on technical performance, but on how they affect business risk, autonomy and accountability.
The future of farming
AI will continue to reshape Australian agriculture. The most important question is not whether it replaces farm jobs, but who carries the risk when AI becomes part of everyday decisions.
If AI is designed to genuinely support human judgement, backed by shared accountability and proper assurance, it can make farming safer, more resilient and more skilled.
If not, it risks quietly increasing exposure for those already operating at the edge of uncertainty.
Productivity gains are possible. But they will only be realised and socially accepted when AI systems are designed not just to optimise outcomes, but to protect the people who live with the consequences.
References
- ^ warning about (www.weforum.org)
- ^ rise by 35–56% (doi.org)
- ^ promoted as a productivity solution (doi.org)
- ^ shaped (www.oecd.org)
- ^ Automation (www.mckinsey.com)
- ^ monitoring and optimisation (theconversation.com)
- ^ weed risks (www.sciencedirect.com)
- ^ Australia’s agricultural workforce (www.researchgate.net)
- ^ remains limited (doi.org)
- ^ 25% (ageconsearch.umn.edu)
- ^ 60% of jobs (www.cef.imf.org)
- ^ 27–28% (www.oecd.org)
- ^ Evidence (www.oecd.org)
- ^ risks leaving some farmers behind (theconversation.com)
- ^ Logged out: farmers in Far North Queensland are being left behind by the digital economy (theconversation.com)
- ^ Photographic Collection from Australia via Wikimedia Commons (commons.wikimedia.org)
- ^ CC BY-SA (creativecommons.org)
- ^ our research (soilcrc.com.au)
