AI × Cetacean Communication Research
What are dolphins
trying to tell us?
We track the science of cetacean communication — from signature whistles and burst-pulse clicks to the latest AI models like DolphinGemma — and publish what we find, openly.
Weekly dolphin science — research updates, AI advances, field notes.
DolphinGemma & AI Models
Google DeepMind's open model — trained on 40 years of Wild Dolphin Project field data — is the most significant AI advance in cetacean research. We track its development and what it can actually do.
Peer-Reviewed Science
Signature whistles, burst-pulse clicks, mirror self-recognition, cultural transmission. We read the papers and publish plain-English briefs so the science stays accessible.
Open Findings
Our analyses, prompt logs, and model outputs are published openly. If you are a researcher or institution with data, we want to collaborate.
Latest from the lab
Research Briefs
AI Models
DolphinGemma: What the Model Actually Does
Google DeepMind released DolphinGemma in April 2025. Here is what it can classify, what it cannot, and what it means for the field.
Bioacoustics
Signature Whistles: The Evidence for Dolphin Names
Decades of research confirm dolphins use individually unique whistles as identity signals. What the science shows — and where it stops.
Cognition
Mirror Self-Recognition: What It Does and Does Not Prove
Dolphins pass the mirror test. That is remarkable — and it's also frequently overstated. A careful reading of Reiss & Marino 2001.
Interaction, not translation.
That is the honest frontier.
We don't claim to decode what dolphins say. We study the patterns, track the science, and build the tools that make genuine two-way interaction — someday — plausible. Follow the research.