Making the most of artificial and human intelligence for data science
summary
Practical demos show using ellmer and chatlas to call LLMs from R/Python, register functions as tools, run code remotely and build agents that execute and return results; QueryChat generates constrained SQL, ggbot2 iterates to produce ggplot2 code, and Positron Assistant and Databot run code in a live session (Anthropic API required). Databot surfaces a real data bug (a third of rows misaligned in a TidyTuesday CSV), while warnings about security, privacy and deskilling are detailed.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 2:37AI challenges
- 7:59LLMs for data science
- 13:48Posit’s AI tools
- 25:30Positron Assistant overview
- 35:41AI security and ethics
- 47:04Exploratory data analysis with Databot
- 52:32Q&A
related talks
24mDemocratizing data access through LLM
36mGenerative AI + Python: Unlocking Efficiency, Personalization, and Insight
32m"How Smart is AI? – Real projects to build real understanding"
48mUnlocking LLM Potential: From Exploration to Integration
41mBuilding LLM Agents – Challenges and Learnings
48mMastering the AI Shift: Adaptive Strategies for Software Evolution
21mUI in The Age of AI
10mMeet the Speakers