African elephant expert Joyce Poole has spent 50 years studying how elephants communicate. In the 1980s, she noticed something unusual: when one elephant called a specific family member, only that individual responded, while the others ignored the call. Sometimes, the caller would try again, and a different elephant would answer. This made Joyce wonder if elephants had a way to address each other directly, possibly similar to using names. However, at the time, there was no technology to prove it.
Many years later, she worked with researcher Mickey Pardo to test this idea scientifically. Mickey recorded hundreds of elephant calls while carefully noting who made each call, who it was meant for, and what was happening around them. The acoustic data from these calls was transformed into long numerical strings and then analyzed using a statistical model. The model was able to predict which elephant the call was directed to more accurately than chance. This suggests that African savanna elephants may use specific sounds to “name” each other.
This discovery is exciting because machine learning helped reveal something humans could not detect. Today, artificial intelligence researchers hope to go even further by developing large language models designed specifically for interspecies communication. These systems could help humans better understand complex animal communication in the future.
Vocabulary:
• notice (verb) : to see or become aware of something.
• encode (verb) : to change information into another form.
• acoustic (adj) : related to sound.
• statistical model (noun) : a mathematical tool used to analyze data.
• predict (verb) : to say what you think will happen based on information.
• evidence (noun) : information that shows something is true.
• decode (verb) : to understand or translate hidden information.
Source: Vox