At a global scale, we are in a crisis of collective decision-making, with concepts of democracy, identity, enfranchisement and citizenship under threat. In the context of the on-going Climate Emergency, as AI challenges our understanding of non-human intelligence, and as countries worldwide experiment with giving personhood and citizenship to non-human species, we believe it’s essential to explore new forms of multi-species interaction and conversation, within new open environments for both humans and non-humans to communicate. Collective conversation and decision-making is key to our shared futures.

While we don’t yet fully understand non-human communication, much progress has been made in developing authentic conversational interfaces (and understanding communication) for a broad range of non-human species. But we still aren’t sure what we might talk about, or what it means when we discover their personalities, perspectives and opinions conflict with our own. We have neither the language nor the cognitive skills to make crucial – yet complex – collective decisions about our shared futures; this only gets more acute as non-humans return (for it wasn’t always this way) to the conversation.

We’ve been exploring some of this in our recent project, T(h)ree, a permanent landscape intervention sited within Clarence Gardens, part of Regent’s Park Estate in central London. Co‑created with residents of the estate, the project integrates community knowledge, natural infrastructure and hyper-local AI to give voice to a London plane tree that has witnessed centuries of change, so people can ‘talk’ to the tree, and hear what it has to ‘say’ back. The project merges physical, social, technological and ecological systems to create a more‑than‑human experience.

The title, T(h)ree, references the three genetically identical London plane trees on the site, connected underground via shared root and mycorrhizal networks, reinforcing themes of interconnection and multiplicity. The project is a step towards exploring what it would be like to talk with a tree, to hear what it has to say, and to understand what such an interaction implies. Our previous work has shown that even when people know they aren’t actually talking with non-humans, it permanently transforms their relationship to them as beings, and holds a mirror to their engagement with the world. In many senses this project is less about what the tree says, and more about what we choose to say to the tree.

Our previous work has shown that even when people know they aren’t actually talking with non-humans, it permanently transforms their relationship to them as beings, and holds a mirror to their engagement with the world.

The intervention spans the entirety of the public garden. Its digital and ecosystemic networks (i.e. both the tree roots, and the network cables that drive the project) are inscribed visibly on the surface through cobble sett patterns that extend across grass areas and even the dog run. These pathways blend into the surroundings, reminding us of underlying connections, social, technical and digital. At the heart of the intervention are architectural mounds emerging from the soil, crafted from charred wood and rusted steel, shaped to resemble roots. These mounds anchor both naturally and artificially, embedding the technology that operates beneath the surface, and presenting a tactile manifestation of the digital network along the lines of the tree’s roots.

As visitors approach, they hear an amalgam of voices generated from recordings provided by local residents. Any question triggers a response, generated by a large‑language model (LLM) – much like ChatGPT, but rather than being created in Silicon Valley using copyrighted material owned by others, these are co-created with the community, building on their voices, memories, perceptions and perspectives. The voices are deliberately blended, shifting slightly with each interaction, so the personality and tone are never static and always rooted in the collective voice of the community. This creates a subjective, multi‑layered representation of the tree’s character and history, one that responds unpredictably, yet always draws from the language residents themselves have used. When nobody is around, it occasionally starts humming “Daisy Bell”, a reference to the very first song that a voice synthesis machine sang, in 1961. This is in part an allusion to the way a machine “misunderstands” human categories: ‘daisy’ the flower which surrounds the mounds, versus ‘Daisy’ a person’s name, throwing questions about how AI interprets (and acts upon) the world.

A key part of our practice is to re-imagine, re-design and re-present how recent technology can be created, deployed and used. We hope to show that the dominant Big Tech mode of production has more productive alternatives, ones that demonstrate a way that we can work together as communities to create our own shared futures. In T(h)ree we’ve tried to expose technological infrastructure that is usually hidden. Within each mound, embedded system diagrams map planetary-scale interconnections between the tech stack, LLM components, ecosystem, social interaction, waste outputs and even funding decisions. Through these, visitors can trace broader interdependencies: local creatures and plants; residents’ memories; military‑funded software platforms; copyrighted content that’s used to train AI models; gig‑economy labour needed to refine them; political influence on climate; the broader effect on biodiversity; how that affects residents directly, and therefore their perspective on the environment – and, by extension, their perspective of (and stories about) the tree at the heart of the project. By physically embedding these diagrams into the root mounds, T(h)ree makes visible the normally concealed connections across planetary‑scale systems.

We hope to show that the dominant Big Tech mode of production has more productive alternatives, ones that demonstrate a way that we can work together as communities to create our own shared futures.

There’s a dilemma at the heart of the project. We’re struggling to challenge anthropocentrism, yet the most obvious interface – the ‘voice’ of the tree – is founded on anthropomorphism, raising the potential misrepresentation of non-human perspectives, not to mention making use of AI systems that reinforce human-centric biases. We acknowledge this, but consider this a first step. While work is being done by others to communicate with non-humans, we’re working on what those conversations might look like, an interactive rehearsal, reframing not just how we speak to trees, but how we might respond when they speak back. What happens when we disagree?