Community Showcase 1
Date: Monday 2 Nov 2026
Time: 14:40–15:10
Add this session to your calendar, or click a talk’s time to add just that talk.
| Time | Presentation | Abstract | Authors | Contribution type |
|---|---|---|---|---|
| 14:40-14:55 | Do Our Design Frameworks Age? Revisiting Conversational AI Design in the LLM Era | We ran two co-design workshops with 18 psychologists building LLM-powered chatbots using a framework published just before the LLM boom. The talk is about the resulting mismatches there. For example, the framework assumes rule-based systems with fixed capabilities and having a technical team supporting non-technical stakeholders. LLMs generate instead of routing, and now let domain experts design systems directly with no technical intermediaries in the room. We'll discuss what broke (e.g. technical questions arriving too late once roleplaying exposed decisions the group hadn't made yet), what held up (e.g. values-based prioritisation, made even more generative with a single-discipline group), and what we had to adapt to make things work (e.g. adding a "step zero" before prototyping). Our bigger argument is that framework maintenance should count as a genuine HCI contribution, not an afterthought, especially where the underlying technology keeps moving. | Sadek, Peters | Community |
| 14:55-15:10 | How AI became a design material | In 2017, I published an article in ACM Interactions entitled "Intelligence on Tap: Artificial Intelligence as a New Design Material". It predicted that soon, AI would become available for designers to create new products without requiring advanced technical knowledge. This is now happening, but we are still only in the beginning of this revolution. I observe three ways that AI is becoming a design material: First, access to open-weight and inexpensive Large Language Models that can understand context and carry out commands. Second, the Model Context Protocol, which makes it possible for AI models to communicate with external data sources and physical devices in a standardised way. Third, advanced software coding capabilities that lets designers create complete applications based on natural-language prompts. I will discuss the opportunities and risks of this new material, and how it is the responsibility of HCI researchers and educators to master it. | Holmquist | Community |