Farhan Hammad 一 14 Jan 2024
Pilot Edition #4
Note: This newsletter is published under the pilot edition until all segments are integrated.
New Sections: Highlights & Under The Radar (one project worthy of recognition)
Today’s Edition at a glance:
CES Recap - AI Electronics Show 2024
World’s first Large Nature Model
GPT Store Launched
Read Time: 6 Minutes
1. CES Overview
Jan 12th, 2024 - In the realm of cutting-edge technology showcased at CES 2024, several pioneering advancements have taken center stage, giving us a glimpse into what’s to come in this year after the recent leaps in artificial intelligence. Among the highlights is the Rabbit r1, a companion device utilizing a unique LLM known as the Large Action Model to comprehend language requests, providing a novel approach to interaction.
Samsung took center stage with the unveiling of Ballie, an AI Robot companion equipped with a projector on wheels. Powered by an LLM, Ballie can perform various tasks, from smart device control to facilitating video calls and offering a portable entertainment experience by streaming movies and YouTube videos on the go.
LG contributed to the AI revolution with its Smart Home AI Agent. Functioning as a personalized interface, this agent seamlessly connects and manages LG smart home products, enhancing user experience and home automation.
Swarovski showcased the AX Visio, AI-powered binoculars at CES 2024. Designed to identify birds and anything else within its dataset, these binoculars mark a unique fusion of AI and optical technology, catering to both birdwatchers and tech enthusiasts.
The ROG Phone 8 Pro, powered by Qualcomm Snapdragon 8 Gen 3 processor was shown offering new AI features such as Semantic Search, AI noise-canceling for calls, and an AI wallpaper generator. Although, Asus is the first to demonstrate the chip’s capabilities but it will certainly not be the last.
CES 2024 exhibition unveiled a broader shift towards a deeper integration of AI into our lives, promising a future where AI becomes an integral and seamless part of our lives. It all starts this year.
2. Dataland Introduced World’s First Nature Model
Jan 13, 2024 - Refik Anadol Studio is spearheading a groundbreaking initiative called DATALAND, introducing the Large Nature Model. As the world's first open-source, generative AI model dedicated to nature, it is trained on an extensive and ethically sourced dataset of the natural world, diverging from traditional Large Language AI Models.
Beyond serving as a repository or research initiative, the Large Nature Model is envisioned as a tool for insight, education, and advocacy for the global environment. By combining art, technology, and nature, the model aims to raise awareness of environmental issues, inspire innovative solutions, and foster connections among isolated archives. Venturing to 16 unique rainforest locations, Refik Anadol Studio utilizes various data collection technologies, including LiDAR and photogrammetry, to capture ambisonic audio and high-resolution visuals of diverse ecosystems.
3. OpenAI Launches GPT Store
Jan 10, 2024 - Open AI released the highly awaited GPT Store this week, announced back in November 2023. The store allows enterprises or individual builders to list their custom GPTs for others to use and interact with. OpenAI has promised to feature the leading GPT releases each week in a highly curated featured section.
Currently, the featured GPTs include a GPT by AllTrails that can recommend personalized trails for morning hikes, CanvaGPT that can design images and presentations, amongst many others. The company is also working on a revenue sharing model on a pay per use basis to antice more independent contributors.
AMD Unveiled AI-Powered Desktop Chips at CES 2024. AMD introduced new desktop chips with embedded AI capabilities at CES 2024, enhancing productivity and enabling on-device AI features. Partnering with Microsoft, AMD aims to bring localized experiences, such as Copilot, to PCs powered by their chips, offering advantages like lower latency, reduced costs, and enhanced privacy by processing data locally.
Meta's MAGNeT: Transformed Text to Music with AI. Meta released MAGNeT, a text to music and sound model that can generate high quality audio samples using a single non-autoregressive transformer.
The team at Google released PALP (Prompt Aligned Personalization of Text-to-Image Models) that promises superior personalization compared to existing models in a single prompt. Utilizing methods like DreamBooth-LoRA and TextualInversion, PALP has achieved remarkable results, raising the bar for text-to-image personalization.
The company behind TikTok, ByteDance, introduced MagicVideo-V2, a new text-to-video generation model. The model has outperformed market competitors, Pika 1.0 and SVD-XT, based on human evaluations. Have TikTokers accidentally helped ByteDance train the model?
Android Auto with AI: Google is integrating new AI assistant features into the newest version of Android Auto. It will generate summaries for longer text messages, and suggest replies and actions to text messages while you’re on the go.
Curated List of Projects Released This Week
PhotoMaker: Project demonstrating a new personalized text-image generation method allowing rapid customization of photos without needing additional training and fine tuning of the model.
Maybe Finance: A personal finance and wealth management platform that was shut down in 2023, now reborn as an open source project. The team behind Maybe was laid off last year after spending over a $1M in building the platform. The application is packed full of features to manage investment portfolios, savings and retirement plans. The application is now fully open-source, offering an indepth look into the code behind a scalable application.
Copilot to ChatGPT: A self-hosted, deployable server to access ChatGPT for free through Github Copilot.
Jan: An offline alternative to ChatGPT that uses the Llama model for inferencing. The application is supported on all architectures & is currently in active development.
Surya: A new multilingual OCR toolkit offering multilingual and highly accurate line-level text detection, performing much better over engines performing word and character level OCR models. It outperforms the popular Tesseract by Google, in benchmarks for precision and character recognition time.
I hope you enjoyed reading this week’s edition as much as I enjoyed writing it. The newsletter is still in its early days and we would love to hear your feedback and suggestions for future publications. You can reach out to us on our Twitter.
Thanks for reading.