Panel
summary
Panel reports concrete experiments: cheap semantic search becomes possible with embeddings and Cloudflare plus vector stores, GPT‑3 → 3.5 → GPT‑4 show measurable stylistic drift, Llama 3 and local runs enable novel integrations, and LangChain/Open Router and Simon Willison’s llm get used as plumbing. QA pipelines chain models, double‑check each step and sometimes burn tens of thousands of dollars to reach >95% accuracy. Demos include agents that need on-device models and LLM semantic search work but still fail edge cases.
AI-written summary · about
chapters (19)
- 0:00Introduction
- 0:24AI risks and consequences
- 1:05Bias in language models
- 2:09Comparing AI model behaviors
- 3:32LLMs versus search engines
- 4:56Semantic search and knowledge
- 6:19AI for learning and models
- 7:24Synthesizing research with AI
- 8:26Small language models
- 9:51Open source fine-tuning
- 11:54AI model QA and validation
- 13:17AI agent tooling
- 14:45AI agent reliability
- 16:30AI alignment risks
- 17:13Scaling AI agent systems
- 18:36Siri agent and guidance
- 20:00AI model dimensions
- 21:45AI in physical devices
- 23:14Closing
related talks
37mExpanding the Capabilities of LLMs Through Community Collaboration
49mBen Hall & Barbara Fusinska
48mMastering the AI Shift: Adaptive Strategies for Software Evolution
21mChatGPT, LLMs, and LangChains - A Beginner's Guide
01h 50mChatGPT & LLMs: A Practical Starter's Guide to Language Chains
01h 00mUnleashGPT: Mastering the Implementation of Large Language Models
45mPanel Talk
01h 59mLLM applications with LangChain