Artificial Intelligence, Antibiotic residue tests, feed costs ease

Two announcements this month put artificial intelligence into poultry's most repetitive jobs. Lecturers at FAST UB have developed PIYIKAI, a system that identifies the sex of day-old chicks from images. It trains neural networks on large sets of chick photos to classify them, building on earlier poultry models called YMAD and YBEDI. The pitch is lower labour cost, fewer classification errors, and less handling of chicks, plus an answer to the ethical problem of culling male chicks in the layer sector.

At the Global Animal Production and Veterinary Industry Summit in Putrajaya, Malaysia in July, Prof Dr Goh Yong Meng of Universiti Putra Malaysia set out the other application. Monitoring systems read animal movement, sound and biometric data to flag early disease signs, so farmers treat only sick animals instead of medicating the whole group. He pointed to a pig farm in China where software picks up individual coughing and treatment goes to the affected cages only.

Read the full story · source 2

Both stories are the same bet: a camera and some software replacing a trained pair of eyes on a job nobody enjoys doing twice.

Chick sexing is a hatchery problem, and the welfare argument behind it sits mostly with layer operations, where male chicks get culled. If it works, it lands as a cost line inside the hatchery, and a contract grower will never see the machine, only the chicks that come off it.

Targeted medication is the one that could changes how a house is run. That said, the practical applications of this within large poultry houses seems like more of a theoretical benefit than a reality. I think the more promising version of this technology is one that uses camera monitoring and trained AI models to make equipment and environmental decisions based on bird behaviour. This is likely the direction innovative farms will head, as we will start to see large farm operations becoming increasing automated, with only a few people to verify the outputs and directions provided by a machine.

Regardless, with all new technologies there is a cost benefit analysis that needs to be done before making any implementation decisions. What is the accuracy of these machines? How much do they cost? What is the education gap to effectively use them? In areas with very low labour costs, delegating tasks to machines can often times be a less effective use of capital unless the performance benefits are significant. In high labour countries, this tech is usually quickly adopted.

For now, we watch to see how the technology evolves and monitor the feedback from early movers, rather than the described benefits from those producing the technology.

Royal Veterinary College (RVC) researchers, working with institutions in Bangladesh, India and Vietnam, have produced one of the first residue comparisons across countries using a single testing method. They analysed 558 chicken breast samples collected in 2021-22, at the end of the production cycle and at points of sale, screening for 69 veterinary antimicrobials to European Union (EU) reference laboratory standards. At some points of sale, roughly 4% to 9% of samples exceeded international safety limits, against under 0.1% in the EU over the same period. The likely drivers are treatment close to slaughter without observing withdrawal periods (the gap between the last dose and slaughter), plus possible contamination from feed and drinking water.

The trade consequence is already visible elsewhere. Brazil, the EU's largest chicken supplier at 211,000 tonnes in 2025, has until 3 September to show its meat does not come from animals given antimicrobials reserved for human medicine or used for growth promotion, or it loses its export authorisation.

Read the full story · source 2