Shiny x AI
summary
Workarounds for missing APIs and browser inputs lead to concrete components: an input video clip component and audio spinner that enable a GPT‑4o-style video assistant despite no native video API, plus a multi‑model comparator built in hours. A dashboard uses a chatbot that only writes DuckDB SQL so queries are inspectable, it finds 9 outliers by IQR and shows males tip more in absolute dollars but females tip more by percent, and a screenshot→LLM flow explains Plotly plots. The claim is clear: SQL-as-DSL limits hallucination and makes LLM features auditable.
AI-written summary · about
chapters (10)
- 0:00AI grief stages
- 1:45Shiny and AI tools
- 2:51Video chat assistant in Shiny
- 5:54Multi-model LLM comparisons
- 7:41AI for Shiny dashboards
- 11:42Unexpected LLM dashboard uses
- 14:11Explaining plots with AI
- 15:41SQL and LLMs
- 17:28Coding against LLMs
- 18:52Stay skeptical about AI
related talks
59mMaking the most of artificial and human intelligence for data science
48mMastering the AI Shift: Adaptive Strategies for Software Evolution
22mAI First: Applications of the Future
36mGenerative AI + Python: Unlocking Efficiency, Personalization, and Insight
46mBuilding AI-enhanced User Interfaces for the Web
18mAI and Shiny for Python: Unlocking New Possibilities
24mDemocratizing data access through LLM
28mCreating fluent Frontend Experiences in an age of AI