Democratizing data access through LLM
summary
Claims that LLMs let non‑engineers ask data in plain language and get real queries back, demonstrated by a Telegram social‑welfare chatbot called bandy and a minute of meeting pipeline that indexes audio into searchable text; shows RAG workflows using a vector DB to fetch documents into prompts; demos include image classification on a 2019 MacBook in 5.5s and ChatGPT hitting 100M users in 2 months; warns about hallucinations and 18‑day data onboarding bottlenecks and demands guardrails.
AI-written summary · about
chapters (6)
- 0:00Introduction to democratizing data
- 3:50LLM impact on accessibility
- 9:41How LLM models work
- 14:48Limitations of LLM technology
- 18:25Data engineering challenges
- 22:02Key takeaways and next steps
related talks
15mFrom Words to Data: Using LLMs to Turn Questions into SQL
38mLLMs for Self-serve Analytics - Lessons Learned
33mOpening Keynote: LLMs, between fantasy and reality - the poin...
01h 49mAI engineering with open access LLMs that lie, curse, and steal
33mDesigning LLM Applications for Massive Datasets
17mArchitecting LLM-Based Chatbots using free open-source LLMs
41mEnhancing LLMs with Graph Technology
59mMaking the most of artificial and human intelligence for data science