GSV's AI News & Updates (02/27/23)
Meta Joins the AI Race with LLaMa, Spotify's AI DJ, Hugging Face and AWS, Watermarking Text, the AI vs Human Call Center Battleground
You are receiving this weekly update on the intersection of AI and education as part of our GSV community – we hope this will be a helpful resource as we navigate this evolving and exciting intersection. Please feel free to send over any relevant resources you may come across (claire@gsv.com) – we’d love to learn and hear from you!
General 🚀
Hugging Face and AWS partner to make AI more accessible: First came OpenAI + Microsoft, then Anthropic + Google, now Hugging Face + AWS. They plan to bring generative AI modeling to the people, making it faster and easier to train, fine-tune, and deploy models on the cloud in an open, accessible, and transparent way.
🦙 Meta unveils LLaMa, a new large language model that can run on a single GPU: LLaMA-13B reportedly outperforms ChatGPT-like tech despite being 10x smaller. LLaMA is available in four sizes: 7B, 13B, 33B, and 65B parameters, while GPT-3 has 175B parameters. This could pave the way for access to ChatGPT-like performance on consumer-level hardware.
Spotify releases DJ, its AI powered music recommender: a personalized music AI that serves users curated songs, artists, and commentary in a hyper-realistic voice. Its new DJ feature is a combination of its existing personalization technology (its 2022 Sonantic acquisition) and Generative AI through OpenAI. YouTube also recently announced that users will now be able to build personalized radio stations.
Generative AI on Roblox: two tools to make it easier for creators to build games -(1) a feature that outputs 3D items from text prompts and (2) a tool that completes code (like GitHub Copilot).
Snap launches AI chatbot powered by OpenAI’s GPT: My AI, available to Snapchat+ subscribers, can recommend gift ideas, weekend plans, or recipes.
Bain announces alliance with OpenAI to help enterprise clients identify and realize the full potential and maximum value of AI.
Planning for AGI and beyond (OpenAI): OpenAI’s plan for integrating AI developments into society safely. 3 main goals to tackle and direct AGI integration: 1) empower humanity, 2) ensure benefits and governance are accessible to all, and 3) make sure we can navigate the massive risks.
From CEOs to Coders, Employees Experiment With New AI Programs: ChatGPT’s release has sparked a rush of early adopters eager to speed up tasks or avoid being left behind.
Who Owns the Generative AI Platform? (a16z): Where in this market will value accrue?
Education and the Future of Work 📚
How ChatGPT Could Embed a ‘Watermark’ in the Text It Generates
AI in the Workplace Is Already Here. The First Battleground? Call Centers
Publishing industry turned upside down: Over the last few weeks, Amazon's Kindle store has become swamped with AI-generated books, with over 200 e-books publicly acknowledging ChatGPT as co-author. Clarkesworld, one of the largest fiction publishers, closed submissions due to an onslaught of AI-generated spam.
My class required AI. Here's what I've learned so far. I think focusing on how people use AI in class rather than whether they use it will result in better learning outcomes, happier students, and graduates who are better prepared for a world while AI is likely to be ubiquitous.
Startups and Tools 🛠️
Tome closed a $43 million Series B funding round at a $300 million valuation after launching less than six months ago and no revenue to date. The tool turns simple text prompts into "full visual narratives with intelligent layouts, copy and images." Users are leveraging Tome to craft design portfolios, lesson plans, slide decks, and much more.
Uizard: AI design automation - turn screenshots and sketches into editable designs.
Monocle: a full text search engine indexed on personal data - blog posts and essays, nearly a decade of journal entries, notes, contacts, Tweets, and hopefully more in the future, like emails and web browsing history.
Arcwise AI: GPT Copilot for Sheets.
Grain.AI: AI meeting assistant - summary of meetings and automated notes, recall and clip moments, share learnings and takeaways with customers.
Kraftful: ChatGPT but for Product Research - feedback analysis, user sentiment, competitive analysis.
Bloop: ask large codebases questions and find code quickly - natural language search for code discovery, human explanation for understanding code, stage changes.
researchGPT: an open-source LLM based research assistant that allows you to have a conversation with a research paper.
Transvribe: search and ask questions in any YouTube video.
Tech 💻
Generative AI is a legal minefield: Should AI developers pay for rights to training data? Can generative AI output be copyrighted? Can AI slander or libel someone? Who's responsible if AI systems offer private or dangerously false info?
Pundits Have It Backwards: Search Is A Threat To AI: Big Tech will not go gentle into that good night.
OpenAI has privately announced a new developer product called Foundry: enables customers to run OpenAI model inference at scale with dedicated capacity.
Text-to-Image Diffusion Models: A Guide for Non-Technical Readers
Listen 🎧
Marc Andreessen and Steven Sinofsky on AI sentience, ethics, love, jobs , ChatGPT vs Sydney and more
Books of the future are not going to be a static single written thing, it's going to be a book that writes itself as you read it.
Every kid is going to grow up now with a friend that's a bot. That bot is going to be with them their whole lives. It's going to have a memory, it's going to know all their private conversations, it's going to know everything about them.
Hugging Face CEO Clem Delangue on the Future of Open vs Closed Source in AI
ChatGPT, GPT4 hype, and Building LLM-native products — with Logan Kilpatrick of OpenAI
Fun/Other 🎉
AutoDeck: generate a basic pitch deck from any startup description in ~30 seconds.
Using AI to generate every US president as a Pixar character.