TechWatch’s Emily McDaid spoke to Brian Acheson, director of information systems and business solutions at Moy Park, about making use of effective data analytics and the future of AI.
Northern Ireland’s Moy Park is a UK top 10 food company – and the leading European poultry provider – with 12,000 employees and 35m chickens on the ground at any one time. Making effective use of data analytics is important for Moy Park to stay competitive.
How big is Moy Park’s data tracking operation?
Out of more than 12,000 employees at Moy Park, 2,000 are system users. A particularly innovative use of data is by our data analysts, PhD-level specialists who provide statistical and analytical support scrutinising the data offline. The key ‘big data’ mining application in our business is JMP (an SAS-based solution). They can model predictive scenarios, for example, in nutrition or housing conditions for our birds, and then we can apply this evidence-based research to improve our performance.
Across how many sites do you need to maintain one view of your data?
Our objective is ‘enterprise computing’ – one logically coherent dataset for the whole business. We currently have many stand-alone systems, mostly operating from a data centre in HQ. This houses a virtualised environment hosting 250 servers. Virtualisation allows us to maintain many systems on a small number of physical servers and dynamically call up and close down capacity to meet demand (saving energy in doing so).
The scope of our operations is 13 factories and 25 other major locations (offices, stores, hatcheries, feed mills etc), all connected to our wide area network. A large proportion of our poultry suppliers have broadband connections for entering data onto our systems to help monitor birds. We have a variety of secure technologies for sharing data with our suppliers and customers, including giving restricted access to one another’s systems. The main third-party access is with our customers for sharing and entering data regarding future demand and promotions that we plan together.
What kind of data on the provenance of the chickens gets tracked? What other kind of data is tracked?
We have full traceability in our supply chain. Birds are not tracked individually (like cattle) but at a flock level, and move through the supply chain as a flock. On each flock, we know the identity of its parent and grandparent flock – the farms, hatcheries, houses and pens it has passed through.
Typical flock data includes:
- breed of the bird, and the identity of the original batch of eggs purchased
- feed regimes – feed recipes, feed mill sourced from, batch of feed
- care regime – free range, organic
- house conditions – windows, heating system, mean temperatures, cleaning regime
- health – mortality, veterinary records
Where are you using AI and robotics?
Our big data analytics solution uses algorithms to search for statistically significant patterns and relationships in datasets. We use robots in our factories.
Can you describe the robotic system that picks and packages fillets?
The machine is manufactured in Iceland by our equipment supplier, Marel.
- The main piece of machinery is a 1-metre-wide conveyer belt, which moves a very high volume of individual fillets toward a robot.
- The belt (known as a tray-track) has an embedded scale and therefore has information on the weight of each fillet.
- The robot’s packing instructions will be to optimise the selection of fillets into a pack. If, for a 200g pack, it has picked a 91g fillet, it will look for a second one of 109g or more.
The machine is capable of doing all of this faster than the human eye can appreciate.
Any other areas where Moy Park is making use of AI?
We are always investigating innovation solutions, including in technology and ICT, and so AI could play a role in the future.
The primary focus for use of AI-type technologies is in planning. Modern database technologies allow very large datasets to be processed in-memory, and multiple production-planning scenarios run in real time. Planning runs that could once only be run overnight can now be run through many iterations in real time.
Most of our 2,000 users spend large amounts of their time interacting with information. They are knowledge workers for whom time is scarce. The greatest impact AI could have is helping us improve productivity, to make better decisions on how to prioritise time and tasks; an AI solution that helps people align their use of time to their performance objectives and to carry out their work in a collaborative way based on shared/corporate information sources.
When do you think AI will overtake human intelligence?
It has happened already! An AI computing system developed by Google researchers in the UK has beaten a top human player at the game of Go, the ancient Eastern contest of strategy and intuition that has bedevilled AI experts for decades.
While this is a fantastic accomplishment in the field of AI and machine learning, the industry is still in its infancy in terms of personal and business applications.
By Emily McDaid, editor, TechWatch
A version of this article originally appeared on TechWatch