How Does LLM and Generative AI Really Work?
summary
Explains that Gemini offers a 1,000,000 token context window and multimodal input, contrasts that with PaLM / Text‑Bison used in a real project, and shows a pipeline: speech transcription → semantic parser that emits JSON order intents → embedding matcher → TTS. Claims a Wendy’s drive‑thru automation built with those components ran in Ohio, handled slang and noisy audio, and went viral, so viewers can judge whether the demos, latency, error rates and deployment details match the bold claims.
AI-written summary · about
chapters (9)
- 0:00Generative AI and LLMs intro
- 3:25What is generative AI
- 6:50How LLMs predict text
- 9:42Transformer model impact
- 11:29Google's Gemini model
- 13:34Gemini features and integrations
- 15:41Generative AI use cases
- 16:47Wendy's drive-thru automation
- 24:53Closing and future of LLMs
related talks
19mLLM-Enhanced Multimodal AI: Revolutionizing Audio/Video Interaction
27mThe Cutting Edge of AI: From LLMs to Multimodal Intelligence
40mPannel Discussion | From Hype To Value: Real World Applications Of Generative AI And LLMs
29mBeyond Solo: Enhancing LLMs through Collaborative Dialogue with GAN-Inspired Approaches
51mGemini, Google's Large Language Model
56mBuilding Multimodal Apps with Gemini 2 0 and Gemini SDK
01h 39mMastering LLMs: From Prompt Engineering to Agentic AI
31mExploring LLMs across clouds – A Year in the Life