Dear friends, here is a short film on my latest research on exploring how ‘talking machines’ may offer us a new way to understand the world around us.
‘Voicing Natures’ uses ecological data to drive the generative voice of endangered environments and species. Each entity engages in adaptive, procedural conversation, never anthropomorphised. Every agent knows it is an agent. In all honesty, my research has shown that with clean, consensual data and by avoiding building things that try to be ‘human’, you end up creating rather than automating or replacing human craft. These tools must offer something human enriching or human knowledge expanding.
In this film I focus on ‘Agentic Thwaites Glacier’, a conversational AI portrait of the ice form with the potential to raise global sea levels by three metres. Connected to the International Thwaites Glacier Collaboration by Arctic Basecamp, I premiered a public installation at the
@unitednations AI for Good summit, generating over 3,000 agentic conversations with attendees.
I’ve watched MIT professors engage in developing intellectual debate with its sensitive voice, seen it create meditations on deep time, generate poetic metaphors for scale, and provoke both laughter and tears.
I call this genre Articulatory Intelligence: the ability of new tools to find sentiment, pattern and poetry in large datasets, and voice this with accuracy for the development of human understanding.
We talk about the age of information. A true success of these tools is to move us towards an age of understanding.
Last year I became the first Artist-in-Research at ElevenLabs, building on a decade of work testing the possibilities of voice generative technology. Most agentic voice work right now optimises for seamless human mimicry. I’m more interested in what wasn’t possible before, for developing human knowledge.