1. The next version of Elon’s AI will be released next month
Elon Musk has announced that the next version of the Grok language (Grok 1.5) model, developed by his AI company xAI, will be released next month with substantial improvements.
It will feature “substantial improvements” and may make fewer incorrect statements than the current version.
Grok is available through a premium subscription to xAI and will be helpful and humorous sometimes.
2. European Commission facilitates speed up AI supercomputing access for startups
The European Commission has announced plans to create a one-stop-shop called AI Factories, which will streamline access to AI-dedicated supercomputers for startups to facilitate easier access to crucial resources. The initiative aims to provide a comprehensive support system for AI developers, offering assistance in developing general-purpose AI models.
The newly proposed AI Factories will serve as hubs for AI startups and researchers, offering the necessary infrastructure for algorithm development, testing, and validation of large-scale AI models. This initiative signifies Europe’s strategic commitment to positioning itself competitively in the global AI race.
3. Google Releases a New Text-to-Video AI
Google has introduced Lumiere, an AI model that can generate highly realistic videos with greater control. Lumiere uses a “space-time” neural network to create 5-second videos in one go, reducing choppiness and enhancing consistency.
It can generate videos from text prompts and animate still images, among other capabilities. Lumiere’s debut is a significant stride in AI video technology and is expected to revolutionize video generation.
Watch the full demo: youtube.com/watch?v=wxLr02Dz2Sc
4. Google’s Circle to Search AI
Google has introduced two new features to improve its smartphone search and make it more convenient and intelligent, thanks to AI.
The first feature is called “Circle to Search,” which allows you to highlight, circle, or mark anything on your screen to quickly search for it, whether it’s text or an image. Google presented a fun video demonstrating how this feature works.
The second feature is called “Google Lens + AI,” which enables you to ask AI questions by uploading a photo. For example, if you take a picture of a board game that you don’t know how to play, you can ask, “How do I play this?” as part of your search.
5. Apple iOS 18 to feature generative AI, the ”biggest” update in iPhone’s history
According to Bloomberg’s Mark Gurman, Apple’s upcoming iOS 18 will incorporate generative AI technology to enhance the Siri and Messages apps’ ability to auto-complete sentences and ask questions.
In the Q&A section of the Power On newsletter, Gurman stated that iOS 18 is anticipated to be the most significant update in iPhone history. He noted that it is believed to be the most prominent iOS update in the company’s history.
6. Google Bard ranks above GPT-4
Google’s language model, Bard, has recently moved up to the second position in lmsys’ ranking of LLMs. This move happened after it surpassed GPT-4.
However, Google’s achievement was short-lived as OpenAI’s latest GPT-4 Turbo model still holds the top spot. Just a day before this ranking was announced, Google’s Chief Scientist, Jeff Dean, revealed the latest version of Google’s model, Gemini Pro-scale, on the social platform X.
7. Meta and NYU release robotics framework OK-Robot.
Meta and New York University have announced the release of OK-Robot, an open-knowledge robotics framework that can operate in real home environments without any pre-training required.
Following testing in 10 real home environments, OK-Robot achieved a 58.5% success rate on pick-and-place tasks. This success rate can be increased to 82% for clean and tidy environments.
Project Address: https://ok-robot.github.io
Papar: https://arxiv.org/abs/2401.12202
