AI can recognise animal sound patterns but may miss their meaning – study

Israeli and German researchers revealed that while artificial intelligence can successfully recognize animal sound patterns, it frequently struggles to understand their true meaning. Tel Aviv University announced the findings, noting that the technology often groups distinct messages together or separates identical ones.

Scientists turning to artificial intelligence to decode how animals communicate are hitting a fundamental roadblock. A team of Israeli and German researchers has discovered that while advanced algorithms excel at picking up audio patterns, they often fail to grasp what those sounds actually signify, according to a statement released by Tel Aviv University.

The Complex Puzzle of Animal Communication

The researchers explained that similar sounds can carry entirely different meanings depending on context, while drastically different acoustic signals can sometimes convey the exact same message.

Testing Algorithms on Toddler Speech

To rigorously test the limits of modern algorithms, the research team analyzed audio recordings of young children who had not yet fully developed speech. These toddlers produced noises while experiencing distress, calling for their parents, or asking for food. Unlike animal communication, the researchers could partially determine the intent behind the children’s calls, providing a reliable baseline for comparison.

Traditional Audio Analysis Versus Advanced AI Systems

When the team put the recordings through both traditional sound analysis methods and two advanced artificial intelligence systems, the algorithmic approach yielded superior performance. Even so, the technology fell short of genuine comprehension. The AI systems occasionally committed critical errors by grouping sounds with entirely different meanings into the same category, or conversely, by separating distinct audio clips that shared a single message.

What the Findings Mean for Future Research

The published findings demonstrate that software alone cannot bridge the gap in bioacoustics. Decoding the signals used by wildlife will demand a much broader interdisciplinary approach, requiring scientists to combine computational tools with direct observation of animal behavior and a deeper understanding of how different species perceive and respond to sound.

AI can recognise patterns in animal sounds. (Photo
Photo: Eastleighvoice
Photo of author

Sophie Lin - Technology Editor

Sophie is a tech innovator and acclaimed tech writer recognized by the Online News Association. She translates the fast-paced world of technology, AI, and digital trends into compelling stories for readers of all backgrounds.

Asteroid Apophis Set for 2029 Close Flyby Without Collision Risk

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.