LWiAI Podcast #121 - Top researcher leaves Google, trademarking ”GPT”, Stability AI’s new models, 50% of AI ’Catastrophe’, DrakeGPT

Our 121st episode with a summary and discussion of last week's big AI news!

Subscribe

• Apple Podcasts

• Spotify

• YouTube

• RSS

Check out Jeremie's new book Quantum Physics Made Me Do It

Quantum Physics Made Me Do It tells the story of human self-understanding through the lens of physics. It explores what we can and can’t know about reality, and how tiny tweaks to quantum theory can reshape our entire picture of the universe. And because I couldn't resist, it explains what that story means for AI and the future of sentience 

You can find it on Amazon in the UK, Canada, and the US — here are the links:

UK version | Canadian version | US version 

Outline:

• (00:00) Intro

• Tools & Apps

  • (04:20) Microsoft makes its AI-powered Designer tool available in preview

  • (08:27) Visual Studio IntelliCode AI Assistant Gets Deep Learning Upgrade

  • (12:36) GitLab’s new security feature uses AI to explain vulnerabilities to developers

  • (16:25) TikTok is testing a new option to create AI-generated avatars for profile pictures

• Applications & Business 

  • (19:11) ‘The Godfather of A.I.’ Leaves Google and Warns of Danger Ahead

  • (26:52) Machine Learning as a Service Market Size Growing at 37.9% CAGR Set to Reach USD 173.5 Billion By 2032

  • (29:40) Insider Lays Off 10 Percent of Staff After Announcing Pivot to AI + After the Death of BuzzFeed News, Journalists Should Treat AI as an Existential Threat

  • (35:57) PricewaterhouseCoopers announces multiyear $1B investment into generative AI

  • (38:07) Walmart turns to warehouse robots, suggests a boost in profit is possible

  • (40:23) “GPT” may be trademarked soon if OpenAI has its way

• Projects & Open Source

  • (43:50) OpenAI Threatens Popular GitHub Project With Lawsuit Over API Use 

  • (48:27) Stability AI Launches DeepFloyd IF: A High-Performance Text-to-Image Model with Advanced Integration Capabilities

  • (52:10) Stability AI releases StableVicuna, the AI World’s First Open Source RLHF LLM Chatbot

  • (57:32) Replit's new Code LLM: Open Source, 77% smaller than Codex, trained in 1 week

• Research & Advancements

  • (59:44) First Real-World Study Showed Generative AI Boosted Worker Productivity by 14%

  • (01:03:50) LaMini-LM: A Diverse Herd of Distilled Models from Large-Scale Instructions

  • (01:08:50) AI has better 'bedside manner' than some doctors, study finds + Comparing Physician and Artificial Intelligence Chatbot Responses to Patient Questions Posted to a Public Social Media Forum

  • (01:13:50) Machine Learning and Metabolite Analysis Can Predict Individual Body Clock Timing 

  • (01:13:15) Image generation AI for predicting the deformation of splashing drops

  • (01:18:10) Drones can fly themselves with worm-inspired AI software

• Policy & Safety 

  • (01:19:25) ChatGPT-3.5 Generates More Disinformation in Chinese than in English

  • (01:23:27) The Alignment Problem from a Deep Learning Perspective

  • (01:29:28) Former OpenAI Researcher: There’s a 50% Chance AI Ends in 'Catastrophe'

  • (01:33:30) Lawmakers propose banning AI from singlehandedly launching nuclear weapons

  • (01:35:35) Republicans counter Biden announcement with dystopian, AI-aided video

  • (01:37:42) The UK will spend £100 million to develop its own ‘sovereign’ AI

  • (01:41:25) Kamala Harris to discuss A.I. in meeting with Google, Microsoft, OpenAI and Anthropic CEOs

• Art & Fun Stuff

  • (01:44:09) From DrakeGPT to Infinite Grimes, AI-generated music strikes a chord + John Legend calls for regulation on AI-generated music

  • (01:47:30) The A.I. business influencer guys must be stopped

  • (01:51:05) I’m ChatGPT, and for the Love of God, Please Don’t Make Me Do Any More Copywriting