JERUSALEM, Sept. 17 (Xinhua) -- Israeli and German researchers found that artificial intelligence (AI) can recognize patterns in animal sounds but may struggle to understand what those sounds mean, Tel Aviv University said in a statement on Thursday.
In their study, published in Current Biology, the researchers examined a basic problem facing scientists using AI to study how bats, whales, birds, and other animals communicate.
The researchers said that similar sounds can have different meanings, while very different sounds can sometimes carry the same message.
To test the limits of AI, the researchers studied recordings of toddlers who had not yet fully developed speech. The children made sounds while distressed, calling for their parents, or asking for food.
Unlike animal sounds, the researchers could partly determine what the children were trying to communicate.
The team tested the recordings using traditional sound analysis and two advanced AI systems.
Although the AI systems performed better than the traditional method, they still sometimes grouped sounds with different meanings together or separated sounds that conveyed the same message.
The researchers concluded that understanding animal communication will require more than AI, adding that scientists will also need to study animal behavior and how animals hear and respond to sounds. ■
