From clucking to squawking and even “grunting,” the meaning of chicken sounds has always been a mystery, even to farmers.
Not anymore, however, as Japanese artificial intelligence (AI) technology has finally managed to translate them, giving a unique insight into a chicken’s well-being.
Experts trained an AI model with around 100 hours of chicken recordings until it could identify with 80% accuracy whether a bird was happy, sad or scared.
The scientists used machine learning (ML), a specific subset of AI that allows systems to learn and draw informed conclusions.
Audio clips played by experts show the wide range of noises birds make – but can you identify a chicken’s emotion as effectively as an AI?
Humans can expect more ‘meaningful’ interactions with chickens thanks to study results, researchers say
The research was led by Professor Adrian David Cheok from the University of Tokyo, known for his expertise in the field of sex robots.
“This is a big step forward for science and it’s only the beginning,” Professor Cheok said.
“We hope to be able to adapt these AI and ML techniques to other animals and lay the foundation for incredible intelligence across different animal-related industries.”
“If we know how animals feel, we can design a much better world for them.”
According to the academic, his team is now planning to create a free app so that farmers in the UK and Australia can use the technology.
It will listen and analyze the chicken’s sounds and determine what the bird is feeling – whether it is happy, sad, anxious or just hungry.
A proof-of-concept study has been published as a pre-print on Research placemeaning it hasn’t yet been peer-reviewed, although the team has submitted it to Nature Scientific Reports.
The study was led by Professor Adrian David Cheok (pictured) from the University of Tokyo, known for his expertise in the field of sex robots.
What is machine learning?
Machine learning (ML) is a branch of AI based on the idea that systems can learn from data, identify patterns, and make decisions.
ML systems can learn to improve their ability to perform a task without being explicitly programmed to do so.
Such systems can detect patterns or trends in data to draw conclusions or help humans make better decisions.
Machine learning systems become more effective over time as they learn.
For the study, Professor Cheok and his colleagues used a total of around 200 hours of chicken recordings, from a sample of 80 chickens.
The vocalizations of birds in various conditions were recorded before being “thoroughly” analyzed, initially by humans.
The team collaborated with eight animal and veterinary psychologists, who provided expert insights into the chickens’ emotional states.
Once the emotions linked to chicken noises were identified by the human experts, the study authors were able to train their model with the data.
The model was fed approximately 100 hours of chicken sounds, comprising short snippets, each labeled with its corresponding emotional state.
Once trained, the team then tested the AI’s accuracy with an additional 100 hours of fresh chicken sounds that had not been used during the training phase.
Researchers found that he was able to discern the meaning of chicken sounds with about 80 percent accuracy for a variety of emotions: happiness, fear, hunger, fatigue and pain.
This graph shows the accuracy level of the AI model in correctly detecting happiness from chicken recordings – just above 80%
New recordings from chickens could be fed into the model to identify what the bird is feeling – potentially in the form of new software. This could be useful to farmers if they hear their birds making certain sounds but have no idea what they mean in relation to their welfare.
The team describes chickens as “highly social animals” with complex cognitive abilities and emotional experiences, which are furiously bred for their meat.
Their cognitive abilities include perception, learning, memory and even problem solving, while emotions range from happiness to anxiety and fear.
The researchers say their “groundbreaking” work “bridges the gap between human and animal communication.”
“This research not only opens new avenues for understanding and improving animal welfare, but also sets a precedent for further studies of AI-based interspecies communication,” they say.
We already know that chickens have emotional intelligence; a landmark 2011 study found that females showed obvious signs of anxiety when their young were in distress.
These signs included increased alertness, higher heart rate, lowered eye temperature and clucking at the chicks.
British research has shown for the first time that birds demonstrate empathy – the ability to feel someone’s pain or see their point of view.
Chickens were worshiped rather than eaten in Iron Age Britain: birds were often buried ceremonially alongside people, study finds
Today, chickens are a staple in the farmers’ paddock and one of the most widely grown meats in the world.
But new research shows that the ubiquitous birds were originally revered and worshiped as companions before being eaten.
Experts analyzed historical chicken remains found at more than 600 sites in 89 countries, including England, Italy, Turkey, Morocco and Thailand.
The findings suggest the bird was first domesticated only 3,500 years ago in Southeast Asia before spreading to western Europe.
Domestication was triggered by the cultivation of dry rice, which attracted wild jungle birds, the ancestor of the modern chicken, down from the trees around 1,500 BC.
Before that, chickens were considered “exotic” – unusual or tropical because they came from another country – and were buried alongside humans.