Can designers shape LLM behavior without deep technical knowledge?
Explores whether LLMs can be treated as adaptable design materials that designers can tinker with directly, rather than fixed components handed over by engineers. Matters because it determines whether user-centered judgment reaches model adaptation early.
As LLM-powered products proliferate, the people best positioned to align them with user needs and guard against harms — designers — are often shut out because adaptation reads as an engineering task. Canvil introduces designerly adaptation: engaging the LLM as an adaptable design material rather than a fixed backend. A formative study with experienced designers identifies three characteristics it must have: a low technical barrier, leverage of designers' unique user-bridging perspective, and encouragement of model tinkering. Canvil — a Figma widget — operationalizes this with structured authoring of system prompts to adapt behavior, testing adapted models on diverse user inputs, and integrating outputs into interface designs.
The keeper is the reframing of the LLM as a design material: something a designer shapes through tinkering and user-centered judgment, not a finished component handed over by engineers. Treating model behavior as part of the design surface brings responsibility and user-centeredness upstream, where guardrails are cheapest to add.
This is a service-design-relevant note for Adrian. It sits in tension with Will AI automation eventually formalize designer taste?: Canvil empowers designers to shape model behavior now, while that note warns the judgment they exercise will eventually be captured as automated evals — designerly adaptation is the optimistic near-term practice, taste-reified-into-evals the cautionary trajectory.
Inquiring lines that use this note as a source 2
This note is a source for these synthesized inquiries. Follow a line forward into its question, or open it to trace back to all of its sources.
Related concepts in this collection 2
This note in its neighbourhood — explore the map, then jump to a related concept in the list below.
Click a node to walk · click center to open · click Open in graph to see this note in the full knowledge graph
-
Will AI automation eventually formalize designer taste?
Designers argue taste is the irreducible human element AI cannot replicate. But does the same automation pattern that formalized other skilled work suggest taste itself will become the next layer to be encoded into evaluation systems?
tension: empowering designerly judgment now vs that judgment being captured as automated evals later
-
When should human values enter the LLM development pipeline?
Explores whether human-centered concerns like safety and fairness work better as early design principles throughout development, or as post-training alignment patches. Matters because pipeline placement determines whether human priorities shape the foundation or fight against it.
designerly adaptation pushes human-centering upstream into the design surface
Related papers in this collection 8
Papers most semantically related to this note, ranked by cosine similarity in the embedding space.
- Canvil: Designerly Adaptation for LLM-Powered User Experiences
- Conceptual Design Generation Using Large Language Models
- Opportunities for large language models and discourse in engineering design
- Recommender AI Agent: Integrating Large Language Models for Interactive Recommendations
- Model Swarms: Collaborative Search to Adapt LLM Experts via Swarm Intelligence
- Bridging the gulf of envisioning: Cognitive design challenges in llm interfaces.
- Position: LLMs Can't Plan, But Can Help Planning in LLM-Modulo Frameworks
- Systematic synthesis of design prompts for large language models in conceptual design
Original note title
designerly adaptation treats the LLM as an adaptable design material — designers bring user-centered judgment via system-prompt authoring and model tinkering