Meta launches foundational media model, ChatGPT gets better at coding and writing
what to know for now
🎬 Meta's launches new AI video generation suite. Meta Movie Gen combines image, video, and audio generation capabilities, outperforming industry models in human evaluations. It enables text-to-video generation, personalized videos, precise editing, and audio synthesis, potentially aiming to insert itself into the creative workflows of filmmakers and content creators. Read more
📝 ChatGPT launches canvas, enhancing AI-assisted writing and coding workflows. Canvas introduces a separate window interface for collaborative projects, offering inline feedback, targeted edits, and version control. GPT-4o powers this feature, trained to trigger appropriately, generate diverse content, and provide precise suggestions across writing and coding tasks. Read more
🎪 OpenAI's DevDay 2024 showcases developer-centric AI innovations. Annual conference unveils four major updates: Vision Fine-Tuning, Realtime API, Model Distillation, and Prompt Caching. Event marks strategic shift towards empowering developers and reducing AI costs, contrasting with previous year's consumer-focused announcements. Read more
🎨 Microsoft Copilot's radical redesign streamlines AI interactions across platforms. The overhaul introduces a more intuitive interface, integrating voice commands, visual processing, and enhanced Windows functionality. This redesign aims to make Copilot more accessible and useful for everyday tasks and complex queries alike. Read more
🧪 AI Research of the Week
Movie Gen: A Cast of Media Foundation Models
From MetaJake’s Take: Movie Gen is a set of AI models that can generate high-quality videos, personalized videos, edited videos, and matching audio from text descriptions. It uses very large neural networks trained on massive datasets of images, videos, and audio to create realistic and customizable media content.
This technology could dramatically accelerate and lower the costs of video production across many industries, potentially disrupting traditional filmmaking, advertising, and content creation. However, the models’ likely use of copyrighted images, videos, and audio raises ethical concerns, especially considering the technology’s potential replacement of media production jobs.
what to know for later
💰 OpenAI secures $6.6B funding, valuation soars to $157B. Thrive Capital leads round with $1.3B investment. Microsoft contributes $750M. Deal positions OpenAI among top three venture-backed startups globally. Funding aims to advance AI research and expand computing capacity. Read more
🤖 California governor vetoes landmark AI safety bill, citing innovation concerns. Governor Newsom blocked legislation requiring safety testing and kill switches for advanced AI models. The bill faced opposition from major tech companies, who warned it could hinder crucial technology development in Silicon Valley.
🧠 Liquid AI debuts LFMs, outperforming traditional language models. Liquid Foundation Models use minimal memory while delivering exceptional computing power. Built on Liquid Neural Networks, LFMs efficiently handle up to 1 million tokens without impacting memory usage, outperforming GPT and Gemini in benchmarks. Read more
🧫 Generative AI enlisted to combat antibiotic-resistant superbugs. HHS funds TARGET project using AI to accelerate discovery of novel antibiotics. Phare Bio, MIT, and Harvard collaborate to identify 15 promising antibiotic leads, addressing 35,000 annual deaths and $20 billion healthcare costs from resistant infections. Read more