EP36: ChatGPT Vision Road Tested, AutoGen Cheese Test & Anthropic's Break Through

This Day in AI Podcast

Join the discord: https://discord.gg/bb6VZHks

This mind-blowing episode explores the shocking capabilities of GPT-4 vision, including how it identified Mike's exact location just from a traffic photo. We dive into the cheese-filled insanity of using multiple AI agents together with AutoGen, and discuss Anthropic's groundbreaking research into neural superposition. Don't miss our dramatic exposé of Meta's new lobotomized AI chatbots - this episode takes you on a wild ride through the cutting edge of AI!

If you like this episode please consider subscribing and leaving a comment or review.

CHAPTERS
=====
00:00 - Shakespearean AutoGen DRs
00:29 - ChatGPT Vision Road Tested, Augmented Intelligence & Comparison to LLaVA Vision
31:25 - The Cost of AI: Is There A Business Model With Margin?
36:52 - The Value of AI is in Productivity: Discussion of Business Models
43:19 - AutoGen Agents Cheese Test: Do Multi-Agents Perform Better?
58:09 - Anthropic's AI Break Through: Decomposing Language Models Into Understandable Components

SOURCES:
=====
https://llava.hliu.cc/
https://www.wsj.com/tech/ai/ais-costly-buildup-could-make-early-products-a-hard-sell-bdd29b9f
https://techcrunch.com/2023/10/09/chatgpts-mobile-app-hit-record-4-58m-in-revenue-last-month-but-growth-is-slowing/
https://www.wired.com/story/generative-ai-chatgpt-is-coming-for-sales-jobs/
https://twitter.com/amasad/status/1712276988403294610?s=46
https://www.anthropic.com/index/decomposing-language-models-into-understandable-components
https://twitter.com/julesterpak/status/1711766882934534631?s=20

PAPERS:
=====
https://transformer-circuits.pub/2023/monosemantic-features/index.html
https://arxiv.org/pdf/2304.08485.pdf