A0: Applied Exercise
6. TRANSFORM
Move from testing AI to working with AI.
After identifying how the model behaves, use selected artefacts as starting points for a more open-ended human–AI design exploration.
The objective now changes. Rather than keeping the model uninformed, you may deliberately provide context, explain your intentions, introduce references, refine prompts, correct the model, and use other tools introduced in tutorials.
The aim is not simply to generate more alternatives, but to develop a coherent transformation strategy through iteration.
6.1. Develop selected collections
- Select at least 3 collections from your earlier investigations.
- Define one transformation intention / transformation strategy that can be applied consistently across all 3 selected collections: what are you trying to change, improve, explore, or challenge?
- Apply this same transformation strategy to each selected collection.
- Produce approximately 10 variations per collection.
- Develop the alternatives as a coherent comparative investigation rather than as unrelated generations.
- Maintain enough continuity with each starting artefact that the effects of the transformation can be understood and compared across the 3 collections.
6.2. Explore possible transformations
The appropriate transformation will depend on your chosen artefact. For example:
| Artefact category | Possible transformations |
|---|---|
| Bamboo scaffold | Alter bracing; connections; individual members; supports; spacing; overall configuration. |
| Floor plan | Modify circulation; room relationships; openings; unit organisation; core relationships; massing. |
| Façade / building | Vary openings; pattern; articulation; shading elements; material expression; massing. |
| Interior / place | Modify lighting; materiality; furniture; enclosure; spatial organisation; atmosphere. |
| Frame / truss | Alter topology; members; joints; supports; proportions; overall configuration. |
These are starting points rather than fixed requirements. Your transformation should respond to the particular characteristics, problems, or opportunities you identified in the original artefact.
6.3. Work iteratively
For each selected collection:
- Begin separately: Treat each collection as its own design investigation.
- Provide context: You may now explain the artefact, your intentions, constraints, relevant observations, or other information that could help the AI.
- Guide the process: Refine prompts; provide references; identify successful or unsuccessful features; correct misunderstandings; ask for targeted revisions.
- Use different tools: You may combine conversational AI with image generation, image editing, coding, computational tools, or other methods introduced in tutorials where useful.
- Build on previous outputs: Use promising results as inputs for subsequent transformations rather than repeatedly restarting from the original artefact.
- Document the process: Record the starting artefact; transformation intention; important prompts or instructions; intermediate outputs; revisions; and final alternatives.
The following guidelines will help you prepare this part of the assignment:
Read the guidelines on how to interact with LLMs during TRANSFORMS. [link]
Check what evidence, files, outputs, and supporting material you need to produce and retain. [link]
7. SELECT
Apply human judgement.
Generation is not the end of the exercise.
Establish a small set of criteria that reflects what you were trying to achieve. Apply those criteria consistently to your alternatives, compare their strengths and weaknesses, and make a reasoned final selection.
Your choices should reveal what human judgement adds to the generative process.
7.1. Set criteria
Define a small number of criteria appropriate to your transformation intention.
Possible criteria include:
- Addresses the intention: Does the alternative respond meaningfully to the problem, opportunity, or transformation you defined?
- Preserves important characteristics: Does it retain important qualities of the original artefact where appropriate?
- Maintains plausible logic: Does it maintain reasonable spatial, structural, material, construction, or other relevant logic?
- Produces meaningful change: Is the transformation clear enough to matter rather than being merely cosmetic or arbitrary?
- Offers something revealing: Does the alternative suggest an unexpected, useful, provocative, or otherwise interesting direction?
You may modify or replace these criteria where another set is more appropriate to your investigation.
7.2. Evaluate consistently
For each collection:
- Apply the same criteria across the alternatives being compared.
- Identify important strengths, weaknesses, trade-offs, or failures.
- Avoid selecting an alternative only because it appears visually polished.
- Consider whether the output actually responds to the transformation intention and whether its underlying logic is convincing.
7.3. Select & explain
For each collection:
- Select approximately 3 final variations.
- Briefly explain why each variation was selected.
- Identify what changed relative to the starting artefact.
- Explain what the AI contributed to the development of the alternative.
- Identify what required human judgement, correction, interpretation, or decision-making.
- Where useful, note promising ideas that emerged from unsuccessful alternatives.
Your final selection should demonstrate not only what the AI generated, but how you evaluated, directed, and made decisions about what it generated.
The following guidelines will help you prepare this part of the assignment:
Check what evidence, files, outputs, and supporting material you need to produce and retain. [link]