'Articulatory Intelligence' Gives a Voice to Glaciers

Climatologists spend their careers interpreting vast amounts of data, including satellite measurements, ocean temperature models, and seismic records. However, translating this expertise into a form that teachers, policymakers, or the general public can easily understand and engage with is an extremely difficult challenge.

Agentic Thwaites Glacier is a project developed by ElevenLabs artist Harry Yeff to overcome these limitations, giving a 'voice' to one of the world's most endangered environments: a glacier. The project is built on the following open source scientific datasets:

• Satellite measurement data from NASA and ESA
• Scientific records from the International Thwaites Glacier Collaboration
• Climate and ocean models from NOAA and the British Antarctic Survey

This data serves as the knowledge base for a conversational AI agent, built with ElevenAgents and custom voice design. Users can ask questions and receive answers grounded in real scientific data, offering a personalized experience infused with poetic sensibility that goes beyond simple information delivery.

Yeff has named this approach Articulatory Intelligence. Rather than simply retrieving information, it refers to the ability to 'articulate' complex and vast data into a form that is conversational and experiential.

The project was introduced at the United Nations AI for Good and was praised for sparking over 3,000 conversations, showcasing the future of human-machine collaboration. It is also being pursued as part of the Agentic Voicing Natures initiative, which is expanding its scope to other endangered species and environments, such as the Bengal tiger and the dodo bird.