AI won't replace farmers — but it may replace tasks, NFU told

Artificial intelligence is moving from tech buzzword to farmyard talking point, as producers grapple with what data-driven systems could mean for the future of resilient farming.

On day two of the NFU Conference today (25 February), a packed hall heard from technology, academic and commercial livestock experts on how AI might quietly reshape decision-making across the sector.

Chaired by NFU Cymru Deputy President Abi Reader, the session brought together Tim Gordon of Best Practice AI, Professor Jasmeet Kaler of the University of Nottingham and David Speller, chief executive of OptiFarm.

The discussion began by clarifying that this was not about artificial insemination or avian influenza. It was about artificial intelligence — and its potential role in strengthening farming systems under increasing economic and environmental pressure.

The tone was measured rather than evangelical. Producers are more accustomed to debating disease outbreaks, volatile markets and rising input costs than algorithms and machine learning.

Yet speakers argued that AI may sit quietly behind many of the solutions that define the next decade.

Tim Gordon reflected on how quickly the technology has advanced. What felt abstract only a few years ago is now mainstream. The speed of development, he suggested, has been non-linear — and that acceleration is what makes it disruptive.

He compared AI to a highly accomplished agricultural graduate. It can speak fluently about livestock systems, grassland management and disease control. It has absorbed vast quantities of research and data. But it has never walked a yard.

The analogy captured both promise and limitation. AI is informed, but not experienced.

Gordon spoke not only as a technology adviser but as a Cumbrian sheep farmer. Margins are tight, diversification is often essential and every decision carries risk.

For him, AI has acted as a “thinking partner”, stress-testing ideas, interrogating assumptions and modelling scenarios faster than traditional methods allow.

A farmer asked the question many were thinking: “Is this going to replace us?”

Gordon’s response was direct. It will not replace farmers, but it may replace certain tasks. Routine analysis and paperwork could be automated.

The greater risk, he suggested, is that businesses ignoring these tools may fall behind those using them intelligently.

Trust was another theme. “If it has never seen a cow, how do we trust it?” one farmer asked.

AI systems can sound confident even when incorrect. They operate on statistical patterns, not lived experience. Over-reliance carries risk, particularly where legal or welfare decisions are involved. Human judgement, Gordon stressed, must remain central.

David Speller brought the discussion into pig and poultry units, where data collection is already routine.

Feed conversion, mortality, environmental conditions and behaviour metrics generate vast streams of information daily. The opportunity for AI lies in identifying patterns invisible to the human eye.

Early detection of health or welfare shifts can protect both productivity and profitability.

A poultry producer asked whether this means a future of constant surveillance. “Are we heading to a world where every bird is tracked and every movement analysed?”

Speller acknowledged the concern, arguing that analytics should focus attention rather than replace instinct. Used correctly, technology can reduce noise and allow producers to concentrate on higher-value decisions.

Cost inevitably surfaced. “This sounds expensive. Who pays?”

Speakers noted that not all AI applications require heavy hardware investment. Many tools already available can assist with policy interpretation, business modelling and compliance at relatively low cost. As adoption increases, pricing is likely to become more accessible.

Professor Jasmeet Kaler expanded the lens to cattle and sheep systems.

Wearable sensors, behavioural analysis and predictive modelling are being explored as early warning systems for lameness, illness and nutritional imbalance.

In a context of labour shortages, climate volatility and evolving disease dynamics, such tools could strengthen operational durability.

One beef farmer questioned whether producers were “fixing something that isn’t broken”.

Professor Kaler responded that experience remains invaluable, but the operating environment is changing. Technology does not erase instinct; it extends its reach.

Data ownership proved one of the most sensitive issues. “If we are feeding all this information into systems, who owns it?”

Production data carries commercial value. There is unease that aggregated data could strengthen supply chain leverage or be used in ways farmers did not anticipate. The panel agreed that contractual clarity and transparency are essential if trust is to be maintained.

As the session closed, one final question cut through the complexity: “Where should we start?”

The advice was pragmatic. Begin with analytical tools by using AI to summarise policy updates, compare enterprise margins or test diversification strategies.

The mood in hall was neither hype nor hostility. It was cautious curiosity.