OpenAI and Jony Ive in talks to raise $1bn from SoftBank for AI device venture [Link]
OpenAI, former Apple designer Sir Jony Ive, and SoftBank’s Masayoshi Son in talk to launch a venture aimed at creating the “iPhone of artificial intelligence.” OpenAI CEO Sam Altman has engaged Ive’s company, LoveFrom, for the development of OpenAI’s first consumer device. The collaboration aims to make interactions with AI more natural and intuitive, similar to how the iPhone revolutionized the mobile internet. The venture is in its early stages and has received significant financial backing from SoftBank, which is considering investing more than $1 billion. Discussions are still in progress and a formal announcement could take several months. No final agreement has been reached. The article suggests that any hardware developed will likely take years to come to market.
Google adds a switch for publishers to opt out of becoming AI training data [Link]
Google has introduced a new tool called Google-Extended that allows website publishers to opt out of having their data used for training Google’s AI models, while still being accessible via Google Search. This tool aims to give publishers more control over how their content is used, specifically for Google’s AI chatbot, Bard, and Vertex AI generative APIs. Google-Extended is implemented through robots.txt, a text file used to inform web crawlers about which sites can be accessed. The tool comes as AI applications are growing and questions around data usage are becoming more prevalent. Already, multiple websites have blocked OpenAI’s web crawler to prevent data scraping for training ChatGPT. The New York Times and other sites have taken legal measures to prevent Google from using their content for AI training, due to the difficulty of completely blocking Google’s web crawlers without impacting search engine indexing.
Polycam just shipped a Gaussian splat viewer & creator. [Link]
With only about 20 images you can generate & explore 3D scenes in real-time at 60fps.
3D Gaussian Splatting is a rasterization technique used for 3D reconstruction and rendering. It represents a 3D scene using millions of particles, known as 3D Gaussians, each having attributes like position, orientation, scale, opacity, and view-dependent color. To render these particles, they are “splatted” into 2D space and organized for efficient rendering. The technique is different from traditional methods that use polygonal meshes for 3D data representation.
A breakthrough paper from researchers at Inria and the Max Planck Institute outlines the data generation process for 3D Gaussian Splatting. To create a Gaussian Splat, one needs an image set ranging from 20 to 200 images in either PNG or JPG format. It’s important for the images to be crisp, well-lit, and free of motion blur for high-quality reconstructions.
While most current 3D software doesn’t support Gaussian Splatting, some plugins exist for Unity and Unreal Engine. The technique is considered to have a relatively easier pathway for integration into existing renderers compared to other radiance-field-based techniques like NeRFs.
The technology is still experimental, and users can report bugs or issues through the project’s Discord server. Native mobile support is in the pipeline, but for now, desktop browsers are recommended for viewing splat renderers.
First Interview in the Metaverse | Mark Zuckerberg & Lex Fridman [Link]
The following is a conversation with Mark Zuckerberg, Inside the Metaverse. Mark and I are hundreds of miles apart from each other in physical space, but it feels like we’re in the same room because we appear to each other as photorealistic Kodak Avatars in 3D with spatial audio. This technology is incredible and I think it’s the future of how human beings connect to each other in a deeply meaningful way on the internet. These avatars can capture many of the nuances of facial expressions that we humans use to communicate and motion to each other. Now, I just need to work on upgrading my emotion-expressing capabilities of the underlying human. This is the Lex Fridman Podcast. And now, dear friends, here’s Mark Zuckerberg. This is so great. Lighting change? Wow.
Chat with Enron’s Past: Navigating the Enron Email Corpus with RAG and semantic search [Link] RECOMMENDED
AI language models can exceed PNG and FLAC in lossless compression, says study [Link]
NeuRBF: A Neural Fields Representation with Adaptive Radial Basis Functions [Link]
How do you use AI to get things done faster? [Link] RECOMMENDED
Default No to AI Training on Your Stories [Link]
– allows for multiple agents that align to one goal
– human input allowed
Prompt of the Day 🍭: Chain of Thought
Hope you enjoyed today’s newsletter
⚡️ Join over 200,000 people using the Superpower ChatGPT extension on Chrome and Firefox.