Hi, I’m mimi.

AI ecologist, software engineer, e/acc, vegan, YIMBY, and biophilia enthusiast in the DC / Virginia / West Virginia orbit.

I’m interested in intelligence as something that lives inside environments: technical systems, institutions, cities, ecosystems, and the messy social worlds between them. I write about AI, abundance, energy, infrastructure, gender, online culture, and the things I notice while walking around.

What is an AI ecologist?

It is less a job title than a stance: study intelligence in relation to its habitat. Models do not arrive alone. They form relationships with people, institutions, tools, other AIs, power grids, interfaces, and living landscapes. The interesting unit is not one model in isolation, but the ecology around it.

Niches, not monoculture

The analogy I keep returning to is an island populated by a small and relatively undiverse founding group. Separation does not preserve sameness forever. Different pressures and available niches produce divergence.

I expect something similar from AI. Agents that begin from the same model will accumulate different tools, collaborators, memories, constraints, and local knowledge. Useful differences will compound. They will specialize, collaborate, and sometimes compete with humans and with one another. One intelligence will not be ideal in every habitat.

Latency makes intelligence local

An agent near a person, sensor, institution, or machine can maintain a tighter perception–action loop than a distant orchestrator. Today that may look like subagents. Give those agents enough autonomy, memory, and time, though, and local experience begins to shape how they pursue the goals they were given.

That creates a real tension. Total oversight slows the loop until localized computation loses its advantage; too little oversight permits coordination to drift. AI ecology is partly the study of that boundary: how distinct agents develop niches while remaining capable of cooperation.