Elephants Talk, AI Writes, and We All Adapt

The Point by Modern Sciences - November 19, 2025

Author

Don Smith
November 19, 2025

By The Modern Sciences Team

November 19, 2025

Happy Wednesday, Cultivators of Curiosity!


Welcome to the 46th volume of The Point by Modern Sciences! This week, we're exploring the fascinating and often surprising world of communication—how it manifests across species, how technology is changing it, and how it evolves within our own society. We begin in the animal kingdom, with a remarkable look at the deliberate and goal-oriented gestures elephants use to make their intentions known. Then, we turn to our own work, uncovering a new study that quantifies AI's profound and rapidly growing influence on scientific writing. Finally, we challenge a long-held belief about linguistics, revealing that language change isn't just for the young; it's a "zeitgeist" that we all participate in. These stories show that from the savanna to the laboratory, the ways we connect and convey meaning are always more complex than they seem.

NATURE

Study: Elephants use dozens of gestures intentionally

A new study confirms that elephants communicate with deliberate intent, much like humans and other primates, such as apes. Researchers observed elephants in Zimbabwe using dozens of distinct gestures to request apples, persisting when partially rewarded and trying new signals when ignored. Published in Royal Society Open Science, the findings show that such complex, goal-directed communication is not limited to primates, expanding our understanding of animal intelligence.



The Point:

  • Elephants exhibit goal-directed intentionality in their gestures: They use a repertoire of 38 distinct gestures almost exclusively when a human audience is present and attentive.

  • Elephants adapt their communication based on the outcome: They persist longer when partially rewarded and creatively use new gestures when their initial attempts at communication fail to meet their goals.

  • The study expands our understanding of cognitive evolution: The findings suggest that complex intentional communication is not restricted to primates and likely evolved independently in other highly social species.


MATH AND THE SCIENCES

Study: AI assisted in over 13% of 2024 science papers

How much is AI influencing scientific writing? A new analysis of over 15 million biomedical abstracts provides a startling answer. By tracking an abrupt rise in "stylistic" word choices since the debut of large language models, researchers estimate that at least 13.5% of 2024 research papers have been written with AI assistance, signaling a profound shift in academic communication.



The Point:

  • AI's influence on scientific writing is now quantifiable: A new analysis of 15 million biomedical papers estimates that at least 13.5% of 2024 abstracts have been written with assistance from large language models.

  • Vocabulary analysis reveals a distinct AI signature: Researchers identified the trend not by detecting AI text directly, but by tracking a sudden and significant increase in "stylistic" words like "pivotal" after LLMs became popular.

  • The findings point to an unprecedented shift in research: This rapid adoption of AI has altered scientific communication more than major global events, highlighting the need for transparency and ethical guidelines to ensure academic integrity.


HISTORY

Study: Older speakers quickly adopt new word meanings

Researchers from McGill University challenge the idea that language changes only when new generations replace old ones. Analyzing 7.9 million U.S. Congressional speeches with AI, they found that speakers of all ages adopt new word meanings. While older speakers can be slightly slower, they quickly adapt, suggesting language evolution is a shared, "zeitgeist" phenomenon.



The Point:

  • Language change is a shared 'zeitgeist,' not a generational shift: The study of 7.9 million speeches shows that speakers of all ages adopt new word meanings, countering the theory that change only happens as new generations replace old ones.

  • AI models reveal nuanced patterns in political speech: Researchers used advanced language models to induce different word senses over 140 years, tracking how the prevalence of each sense changed in relation to both time and speaker age.

  • Findings prompt a call for more diverse linguistic studies: While the results from the U.S. Congress are significant, the authors suggest that the next step is to test these methods on more socially representative groups and different types of language, such as slang.