Anthropic’s latest model claims to be more powerful than GPT-4o
Plus Ilya Sutskever launches his new company
Today’s Highlights:
📰 News: Anthropic releases latest model, Claude 3.5 Sonnet + Ilya Sutskever announces his new company after leaving OpenAI
💰 Funding: Nvidia passes Microsoft + Apple to become the world’s most valuable public company thanks to AI boom
⚡️ Top News Stories:
1. Anthropic releases its next model Claude 3.5 Sonnet, which outperforms previous models, including Claude 3 Opus, in various AI benchmarks for text, image analysis, and code generation.
A new feature, Artifacts, allows users to edit and collaborate on AI-generated content directly within the Claude interface, enhancing the model's utility beyond basic chatbot functions.
Claude 3.5 Sonnet is twice as fast as its predecessor, handling complex instructions and visual data more accurately, which is crucial for app developers.
It claims to outperform OpenAI's GPT-4o and Google's Gemini across various tasks, and is now available to developers and users.
Anthropic aims to turn Claude into a comprehensive business tool for centralizing knowledge and documents, competing with platforms like Notion and Slack.
Despite incremental progress, Anthropic plans to continue improving its models, with features like web search and preference memory on the horizon.
2. Ilya Sutskever, having left OpenAI, has now co-founded Safe Superintelligence Inc. (SSI), focused solely on developing safe and powerful superintelligent AI.
SSI aims to create beneficial superintelligent AI, prioritizing pure research over commercial products, and insulating itself from short-term business pressures.
The company is co-founded by Daniel Gross(former Apple AI lead) and Daniel Levy(ex-OpenAI engineer), with undisclosed funding but a strong backing anticipated.
SSI will have offices in Palo Alto and Tel Aviv, maintaining a lean team dedicated to AI safety through engineering breakthroughs.
Sutskever’s new venture addresses the challenge of creating a superintelligent AI that is safe for humanity, a critical focus in the AGI race.
3. Color Health and OpenAI partner to help doctors make evidence-based cancer treatment decisions, using GPT-4o to create a copilot application that extracts, processes, and normalizes patient information, generating personalized screening and diagnostic plans reviewed and modified by clinicians before use.
4. Runway releases a new AI tool, Gen 3 Alpha, designed to generate hyper-realistic human faces and expressions, enhancing control over structure, style, and motion of video content. Here are some samples:
5. Meta FAIR (Fundamental AI Research Team) releases new AI models and datasets to advance open research and responsible AI innovation.
6. Google DeepMind releases V2A tech that generates soundtracks from video pixels and text prompts, enhancing the realism of generated movies with synchronized audio.
7. Roblox is developing 4D gen-AI to enhance virtual interactions by assisting in the creation of interactive and functional 3D objects that behave realistically.
8. Snap introduces real-time GenAI-powered AR experiences for Snapchat, enabling instant transformations based on user input.
9. Dell Technologies and Super Micro Computer (SMC), in collaboration with Nvidia, is helping build an "AI factory" to power the next version of xAI’s chatbot Grok.
10. Silicon Valley companies, including Google and OpenAI, are intensifying staff screening to counteract the threat of Chinese espionage, especially for those handling sensitive technologies.
11. Japan is aiming to become the "most AI-friendly country" by focusing on industry-led oversight and minimal regulations, designed to attract investment from global tech companies, and leveraging its pro-business stance to lure investments away from Europe which has a comprehensive AI Act.
12. Forbes accused Perplexity AI of "willful infringement" for copying text and images without attribution, leading to AI-generated content outranking Forbes' originals on Google, an issue CEO Aravind Srinivas acknowledged and promised to address by improving source highlighting.
13. Apple is seeking a Chinese partner to introduce AI services in China due to the unavailability of Western AI models like OpenAI’s ChatGPT.
14. Target will introduce a GenAI chatbot, Store Companion, to nearly 2,000 stores by August to assist team members with process questions and store operations.
💰 Top Funding News:
1. Nvidia became the world's most valuable public company, surpassing Microsoft and Apple, with its market value surging from $1T to over $3T within a year, driven by the booming demand for AI-related chips. Nvidia controls over 80% of the market for GPUs used in AI systems, with a significant customer base in major tech companies.
2. Hugging Face acquires Spanish AI software startup Argilla for $10M to enhance their dataset capabilities.
3. Decagon, which leverages gen-AI to provide human-like customer support, raised a $35M Series A, led by Accel and a16z, w/ several notable angel investors.
4. Factory, which has developed AI-driven tools, called Droids, to automate various aspects of the software development lifecycle, raised a $15M Series A, led by Sequoia Capital, with participation from Lux Capital and Mantis VC, bringing the company's valuation to $120M.
5. Finbourne, which has developed a platform to help financial companies organize and utilize their data for AI and other models, raised a $70M Series B, co-led by Highland Europe and AXA Venture Partners.
6. Daydream, which utilizes gen-AI, ML, and computer vision to provide personalized shopping results, raised a $50M Seed Round, co-led by Forerunner Ventures and Index Ventures, with participation from Google Ventures and True Ventures.