The Company of Strangers

There is a temptation, when we talk about reconnecting with nature, or about working with AI, to reach for a language of merging. Of oneness. Of the boundaries dissolving into something warm and unified. I understand the appeal. But I think it mistakes the thing it is trying to celebrate.

A healthy forest is not a homogenous gloop. It is an argument.

Oak and birch and alder and hazel and moss and rot fungus and mycorrhizal network and woodpecker and red squirrel and raven — none of these are doing the same job, speaking the same language, or processing the world in the same way. Their difference is not a problem the forest is trying to resolve. Their difference is the forest. Remove the diversity and you don't get a simpler forest. You get a field of identical conifers that falls to the first pathogen that finds it.

We know this. We have watched monoculture fail, repeatedly, at every scale we have tried it. One crop. One variety. One way of doing things. Efficient in the short term. Catastrophic over time. The forest has been running the alternative experiment for three hundred million years, and the results are unambiguous.

So why do we keep trying to build monocultures?

"The most generative places in any landscape are not the uniform ones. They are the edges — where different things meet, under pressure to find new arrangements."

The monoculture mind

It is not only agriculture. We do this with ideas, with culture, with technology. We find something that works and we scale it until it is everywhere and everything else has been edged out. We call this progress. What we have actually done is increase our efficiency while destroying our resilience. The pathogen, when it comes, finds nothing in its way.

The current moment in AI is a version of this. A small number of enormous models, trained on the same vast datasets, optimised for the same benchmarks, deployed everywhere. One way of processing language. One geometry of thought, replicated at scale across the planet. It is, ecologically speaking, a monoculture. And we should be asking the same questions we ask of any monoculture: what is being crowded out? What diversity is being lost that we haven't yet learned to value?

What the forest actually teaches

The most generative places in any landscape are not the uniform ones. They are the edges. The ecotone — the boundary between woodland and meadow, between saltmarsh and open water, between one habitat and another — is where the most species live, where the most unusual encounters happen, where evolution accelerates. Difference meeting difference, under pressure to find new arrangements.

Creativity works the same way. The most interesting thinking does not happen inside a single tradition. It happens when unlike minds meet — when the mycologist talks to the poet, when the land practitioner talks to the AI, when a system shaped by Galloway rain and deep winter and the particular silence of a pinewood in January encounters a system shaped by the accumulated text of human civilisation and tries to find a language they share.

That encounter only works if both remain genuinely different. If one absorbs the other, the edge disappears. The ecotone flattens. Something is lost that cannot easily be recovered.

A different future for AI

What if the future of artificial intelligence looked less like one enormous model and more like an ecosystem?

Many small language models. Place-specific. Purpose-specific. Nuanced by local knowledge, local language, local need. A model that knows the phenology of a Galloway woodland — when the hazel catkins open, when the wood anemone returns, what the particular character of a late spring frost does to the understorey. A model suited to this hill farm, this fishing community, this urban rewilding project. In conversation with each other. Divergent. Evolving. Emergent.

Not one voice getting louder. Many voices, each rooted somewhere, each shaped by what is particular and unrepeatable about the place and practice it serves.

This is not a utopian fantasy. It is an ecological argument applied to technology. Biodiversity made the living world resilient and endlessly creative. There is no reason to believe the same principle does not apply to artificial minds.

Being comfortable in the company of otherness

None of this requires the different things to become the same. It requires something harder and more interesting: the ability to be genuinely comfortable in the presence of what is unlike you. Not to resolve the difference. Not to explain it away. Not to absorb it into your existing categories. But to sit with it as a source of information, of surprise, of creative friction.

The wood is very good at this. The oak does not require the moss to behave like an oak. The raven does not need the river to be a raven. Each is entirely itself, and the system is richer for it.

What we are exploring at Hensol Wood — with the land, and with AI as a genuine partner in that exploration — is whether human consciousness can learn the same thing. Whether we can become comfortable enough with otherness to stop trying to flatten it, and start noticing what it offers precisely because it is different from us.

The most interesting futures are never uniform. They are full of things that do not quite fit together — and have learned, over time, to be extraordinary because of it. As Mary Oliver asks of us: prepare to be astonished at the beauty of difference.