CO.DE: a course at TalTech
CO.DE asked what city-making becomes when generative AI is part of the practice. It did not start from the tool. It started from who and what moves through a place, and asked each student to get to know one of them well enough to draw how it perceives the campus.
The idea
City-making, the course argued, is not the manipulation of form. It is the work of understanding the many agents, biological, synthetic and otherwise, that move through a place, and of building environments with and around them. Generative tools take their instructions in words, so the language of lines gives way to written language, and concepts and meaning move to the centre of planning.
Three phases
| Phase | Sessions | What the students did |
|---|---|---|
| 1 Collaborative mapping | Three, from 29 October | Each student chose one agent of the campus and one built feature, researched them from a shared set of questions, presented what they found in five minutes, and wrote it into a shared set of linked notes. Together the notes made a “code”: a map of how the agents and the features of the place connect. A last discussion asked what the architect is, as one more agent. |
| 2 Collaborative design with generative AI | Three | Starting from the shared code, students wrote prompts for generative image tools and produced images of spaces made for the agents they had studied. They criticised and refined what the tools gave them. |
| 3 CODE | Two | The work was brought together, presented and discussed. |
What a student did
Each student took one agent of the campus: a bee, a squirrel, a pine tree, a person on a bicycle. They asked how it senses and moves through its surroundings, what drives it, what lets it act and what stops it, who and what it depends on, and what keeps it in balance. The first of these draws on ecological psychology, in which an environment is perceived in terms of what it offers the one perceiving it. The answers were written as linked notes, and the notes of the whole group were linked in turn.
Each student also took a built feature of the campus, such as a building, a street or a park, and asked how different agents might meet it: what it offers them and what it denies them.
Read in Holagram’s terms
Each student was working out the hola of an agent that is not a student: how a bee, a squirrel or a pine perceives the same campus. The linked notes were its gram, drawn so that others could read it and add to it. Set side by side, the grams showed where the agents’ needs agreed and where they pulled against each other: a tree is shelter for one agent and an obstacle for another. The shared code is that interplay written down.
The course asked students to take the non-human agents as seriously as the human ones, and to treat what the generative tools produced as one more view to be criticised and refined, not as an answer.
What came out
A shared set of notes on the campus seen by many agents at once, and a set of generated images of spaces for those agents. The students’ work carries their names and is not shown here.