Silicon sampling
workshop
Why?
Understanding how technology changes people is difficult — many effects are unexpected and not immediately visible. Large Language Models (LLMs) can serve as a tool to broaden perspectives, using their knowledge of cultural, social, and historical patterns found online. By sampling diverse voices and situations through LLM prompts, participants can surface both likely and unlikely consequences of a technology. These insights feed directly into design choices and support more responsible innovation.
How?
In groups, participants define the technology and describe relevant users and stakeholders. Using an LLM, they generate multiple personas, situations, and future stories that show how people might react, adapt, or be changed. Teams then select, discuss, and map these scenarios, making the potential impacts concrete. The process emphasizes hands-on exploration: not only asking the LLM for outputs, but actively shaping, sorting, and interpreting the results to inspire design directions.
Ingredients
- Basic understanding of Large Language Models.
- Clear stakeholder and user insights.
- Templates or tools for capturing personas and scenarios (storyboards, sticky notes, mapping boards, design fiction).
In practice
Once a technology idea is defined, teams use an LLM to generate a broad range of possible futures. They create dozens of characters, explore different viewpoints, and collect utopian and dystopian scenarios. These are then sorted, evaluated, and translated into design implications. Running the cycle multiple times ensures the design stays grounded in diverse human perspectives and anticipates unexpected consequences.
Phase(s) of use
In the following project phase(s) silicon sampling can be used:
- Realisation
- Analysis
