Plenary session
summary
Conference opens with community news then pivots to an experiment: a chatbot called Gray I Am is trained on 500 messages scraped from Zulip to mimic a prolific contributor. The model answers schedule questions and gives talk recommendations but sometimes invents details, confesses personal crises, and misunderstands program context. Demonstration works intermittently — useful for quick triage and drafting but requires human vetting and careful prompting (use the FHIR label and the ask Graham command).
AI-written summary · about
chapters (7)
- 0:00Welcome and opening
- 1:23Announcements and winners
- 3:28Social event and waffle surprise
- 4:54Fire resource quiz game
- 10:14Introducing Gray I am
- 12:20Testing Gray I am on Zulip
- 15:09How to use Gray I am
related talks
22mConversations With AI: Transforming Conference Learning & Data Engagement
59mMaking the most of artificial and human intelligence for data science
17mUsing bots to Improve Humanity
48mUnlocking LLM Potential: From Exploration to Integration
22mAll Work and No Play? Conversations with a Question-and-Answer Chatbot in the Wild
48mMastering the AI Shift: Adaptive Strategies for Software Evolution
43mAI Has Entered The Chat: Humans, AI, and the Art of Collaboration
27mQ&A #4