Ecologies of LLM
What can I learn?
A collaborative workshop format, from the Sciences Po médialab, that explores large language models as ecosystems — tracing the data, energy, infrastructure and human and more-than-human relations that a single model depends on, rather than treating it as a self-contained piece of software.
Core insight
An LLM is not an isolated tool but a node in a web of ecological and social dependencies. Looking at the "ecologies" around a model reframes it as something embedded in the world, with costs and entanglements that can be examined and debated.
How to use it in daily work
A facilitation approach for moving group conversations beyond "what can the chatbot do?" toward the systems and resources that make it possible.
- Use the ecosystem framing to help a group see why questions about AI and the environment are not separate from questions about how AI works.
- Adapt the workshop's relational lens for a session that connects AI to broader debates about sustainability and infrastructure.
Note
The source page did not include a full description; the summary above is inferred from the project title and host (Sciences Po médialab) and should be checked against the site.