From machine learning to generative AI: What is out there for dairy farmers?

We’ve covered before how Big Data can be leveraged to the benefit of dairy farmers, from tracking calf health vitals​ through real-time monitoring technologies to detecting breaks in the cold chain​ through Internet of Things solutions.

Generative artificial intelligence (AI) and machine learning offer even more tools that can be adapted to a myriad of applications in dairy. But where did it all start?

An invited review published in Applied Animal Science attempted to round-up some of the applications for AI and machine learning related to dairy farming and production going as far back as the 1980s.

The authors from University of Florida Gainesville and the Colorado State University Fort Collins traced the use of AI in dairy as part of so-called expert systems – computer software applications trained to solve problems and carry out functions like those performed by humans. These systems didn’t take root, the authors noted, partly due to the hardware and software limitations of the time. However, scientists started to take advantage of machine learning – also known as a method called ‘random forest’ – to analyze large data sets – a practice that continues today.

Further in the study, which is referenced at the end of this article, the authors suggest new ways in which dairy farmers and food producers can take advantage of AI – for example, to offer real-time translation where employees speak different languages, or to use virtual reality platforms to train workers.

But with so much funding and resources going into AI development and the promise that the technology could revolutionize the dairy sector, what is actually available to producers right now?

Taking advantage of automation and optimization toolsPlatforms that gather and monitor data in order to provide case-specific analysis – for example, to detect early disease symptoms by logging subtle temperature changes in cattle – are prevalent, with solutions designed to optimize manufacturing processes also gaining traction.

It’s only January 2024, but within a week, two commercial solutions aimed at optimizing dairy production have been announced.

Texas-based Ever.Ag has launched a cheese yield optimization tool that leverages AI and machine learning to enable manufacturers improve productivity and reduce waste.

The company says cheesemaking poses inherent challenges and it’s hard to predict what is going to come out of the next vat based on know-how alone; the technology on the other hand promises ‘unprecedented visibility’. “Our Cheese Yield Optimization program digitizes the majority of the cheese production process, analyzes the data to provide actionable recommendations to make operators in real-time and provides suggested recipe changes for tomorrow’s production,” explained Ryan Mertes, head of manufacturing solutions at Ever.Ag. “These recommendations are based on machine learning and AI and quantify the difference between optimum yields and undergrade production to the operations and financial teams.”

He added that the tool ‘learns from existing and new data sets to highlight operational improvements, without taking away the art of making cheese’. “The system does this with recommendations tailored to the user,” said Mertes. “Using existing data sets means customers will receive results in as little as 90 days versus 12 to 15 months.”

These tools would both enhance efficiencies while enabling producers to elevate the quality and consistency of their products, added the company’s Simon Drake, EVP, data science solutions.

Another solution coming from the US is SPX Flow’s Anhydro SmartDry System, which utilizes precision control and automation to improve consistency and control over spray-drying systems and product quality. The technology – which is a small form-factor system that packs a quad-core processor – can provide moisture control optimization for the dryer chamber and multiple fluid beds in order to eliminate moisture variability from production. The system can automatically adjust its settings to maintain production requirements, says the company, and can be set up in several days.

And more work is being done globally.

In the UK alone, funding projects initiated by state agency Innovate UK is allocating £100m/$126.9m to invest in AI innovation in key sectors including agriculture. One of the projects funded through the program was a feasibility study led by environmental control company Galebreaker Ltd alongside IoT specialists Smartbell, which is assessing how dairy cow behavior can be used to optimize barn environment and improve herd productivity and welfare. Another project led by machine learning specialist digiLab is helping farmers to identify and verify carbon capture.

What’s next for automation in dairy? ChatGPT (probably) has the answerAside from machine-learning and automation solutions, the next wave of technology is likely to leverage generative AI.

One of the early examples of a ‘virtual assistant for dairy farmers’ was a project launched by Dutch tech company Connecterra.

The project, which set out to create Ida, an AI-powered assistant for dairy farmers – received in excess of €2.4m/$2.6m in funding including €1.6m/$1.7m from the EU’s Horizon 2020 program. Connecterra went on to develop Ida into a tool that can monitor and compare cow behavior and farm performance against the most efficient farms globally, helping farmers improve their environmental performance. The company has since entered a strategic partnership​ with livestock management solutions firm Datamars, which has acquired Ida, though Connecterra continues to develop AI-powered solutions for farmers. 

According to the Dutch company’s CEO Yasir Khokhar, generative AI powered by large language models (LLMs) has the potential to be an even bigger game-changer for dairy than straight-up machine learning.

“While current language models are trained on human communication, it is also possible to train them on specific knowledge, such as dairy farming,” he explained. The current wave of LLMs are generalists. Their underlying training is based on human communication, text and visual data scraped from the internet. However, it is possible to train these LLMs with specific knowledge leading to the creation of a dairy-trained AI model that can help make complex decisions.

“It is our belief that almost every aspect of the dairy industry will have an AI-driven use case.”

Know about any exciting AI technology developments or feasibility projects related to dairy production or farming? We’d like to hear from you – please get in touch with the editor​ with a brief summary of what the project is about, who is behind it, and what are the intended outcomes.​

Sources:

Invited Review: Examples and opportunities for artificial intelligence (AI) in dairy farms
Authors: De Vries, A., Bliznyuk, N., Pinedo, P.
Published: Applied Animal Science 39:14-22, 2023
DOI: 10.15232/aas.2022-02345