Today:

Hey Folks!

In today's newsletter, we're featuring a remarkable YouTube video that's a must-watch. 📺 We're diving into something absolutely mind-blowing. 🤯 Google has just unveiled its latest masterpiece, Lumiere – an AI that's changing the game in video creation. Google's paper spills the beans on the tech behind it, calling it a "Spacetime diffusion model." Basically, Lumiere dreams up the whole video at once, not frame by frame, making for smoother, more consistent shots.

Compare this to other AI tools out there – Lumiere's way ahead in creating more lifelike, consistent videos. It's got this neat trick of maintaining style across a video, something others struggle with.

WATCH THE VIDEO ON YOUTUBE

Google announces new AI-powered features for education

Google's rolling out cool new features for schools, making it easier for teachers to run their classrooms and plan lessons. They're adding AI to YouTube, so it can suggest questions for videos. This tool was first teased last year, but it was limited to access. Now, they've got this thing called Practice sets in over 50 languages, where AI makes up answers and tips. Teachers can even turn a Google Form into a practice set.

They're also launching a Resources tab to keep track of all these interactive tools. Plus, their Duet AI can now help teachers whip up lesson plans.

Classroom management getting a boost too. Teachers can make different groups in Google Classroom and give them unique assignments. They've also added a feature to Google Slides that lets teachers narrate their presentations.

For getting stuff signed, teachers can now use eSignature in Google Workspace. And Google's sprucing up Classroom analytics, so teachers can track how well students are doing on assignments.

They're also making Chromebooks and Chrome more accessible. ChromeOS will read text from PDFs for people using screen readers. And Chrome's got this new feature where it reads highlighted text out loud. Google Meet's getting closed captions in 30 languages and lets you pin more than one host.

TECHCRUNCH

Microsoft Forms Team to Make Cheaper Generative AI

Microsoft is stepping up its AI game by creating a new team focused on developing AI that's more affordable and uses less computing power than OpenAI's tech. This team, called GenAI, is pulling in top AI experts from Microsoft's research division. 

Their goal? To make conversational AI that's lighter on resources but still packs a punch. This move shows Microsoft's commitment to advancing AI technology, making it more accessible and cost-effective.

THE INFORMATION

Europe launches AI one-stop shop with supercomputers for startups

Europe is upping its AI game by launching a hub, dubbed "AI Factories," to give startups easier access to top-notch supercomputers. This move is to boost AI development, allowing for the creation and testing of large-scale AI models. It's a strategic play by the European Union to stay competitive with the US and China in the rapidly evolving tech landscape. 

The initiative stems from recent EU rules aimed at governing powerful AI systems. In short, it's Europe's big push to be a major player in AI, blending resources like computing power, data, algorithms, and talent.

REUTERS

Microsoft’s Journey to $3 Trillion Aided by Cloud Growth, AI Optimism, and Big Deals

Microsoft's market value briefly hit $3 trillion, a milestone for this nearly 50-year-old tech giant. Its stock peaked at a record $402.46 but needed to reach $403.65 to sustain the $3 trillion mark. The company, which became a billion-dollar business in 1986, achieved $1 trillion in 2019 and doubled that in just two years.

This rapid growth is largely thanks to its booming Azure cloud services and the buzz around its potential gains from generative artificial intelligence, positioning Microsoft as a major player in these advanced tech fields.

THE WALL STREET JOURNAL

Nvidia, Microsoft, Google, and others partner with US government on AI research program

The National Science Foundation (NSF) is joining forces with big tech players like Amazon, Microsoft, Nvidia, and others for the National Artificial Intelligence Research Resource (NAIRR) pilot. This program aims to keep the U.S. at the forefront of AI research and innovation. It's a two-year project involving heavy hitters in AI and federal agencies like DARPA and NASA. Born from President Biden's AI executive order, NAIRR will help shape future government and private sector advancements in AI.

Tech companies are pitching in big time with resources and support but aren't getting paid for it. For instance, Amazon's offering support for 20 research projects, and Nvidia's throwing in $30 million, including $24 million in computing power. This move is a game-changer, especially for smaller colleges and rural institutions, giving them access to top-notch AI tech and resources.

YAHOO! FINANCE

High Can See Clearly Now: AI-Powered NVIDIA RTX Video HDR Transforms Standard Video Into Stunning High Dynamic Range

NVIDIA's kicking things up a notch with their new AI-powered RTX Video HDR. This cool tool turns average videos into eye-popping HDR ones, especially on HDR10 displays. For PC gaming modders, the NVIDIA RTX Remix open beta's a game changer with its full ray tracing and AI texture tools. Plus, the GeForce RTX 4070 Ti SUPER is out now, boasting more power for creatives and gamers alike.

This week, 3D artist Vishal Ranga's showing off his skills with a stunning scene in ‘In the NVIDIA Studio.’ RTX Video HDR needs an RTX GPU and an HDR10-compatible display to work its magic. Lastly, Vishal Ranga, a pro in game level design, shares his process and encourages aspiring artists to keep grinding and look after their well-being.

NVIDIA

🧠RESEARCH

GALA: Generating Animatable Layered Assets from a Single Scan

 "GALA: Generating Animatable Layered Assets from a Single Scan" presents a new framework for creating detailed, multi-layered 3D human models from a single scan. This technology addresses the limitations of existing methods which often overlook the complexity of human features like hairstyles and clothing. GALA efficiently separates and reconstructs these layers, including occluded areas, enabling more versatile and realistic human avatars for various applications. The approach involves advanced 3D segmentation, geometry synthesis, and texture completion, significantly improving the process of generating animatable 3D models.

Adept Fuyu-Heavy: A new multimodal model

Adept introduces Fuyu-Heavy, a top-notch multimodal model for digital agents. It's third in its class, ace at UI understanding and various benchmarks. Despite focusing on both text and image data, it rivals similar models in text-based performance. After overcoming challenges in model scaling and data handling, Fuyu-Heavy is ready to boost Adept's enterprise products, setting the stage for even more advanced successors. 

UltrAvatar: A Realistic Animatable 3D Avatar Diffusion Model with Authenticity Guided Textures

"UltrAvatar" is a breakthrough in 3D avatar creation. It focuses on producing highly realistic and animatable 3D avatars with superior texture quality, free from undesired lighting effects. Using a novel approach, UltrAvatar extracts true diffuse colors and applies authenticity-guided texture diffusion for enhanced detail alignment with 3D mesh geometry. This method surpasses existing technologies in generating more lifelike avatars with detailed features. 

WorldDreamer: Towards General World Models for Video Generation via Predicting Masked Tokens

"WorldDreamer" is a cutting-edge model that redefines video generation. It excels in understanding and predicting world dynamics, not just in specific scenarios like gaming or driving, but in diverse, real-world environments. Inspired by large language models, WorldDreamer treats world modeling as an unsupervised visual sequence challenge, transforming visuals into discrete tokens and predicting masked ones. It shines in various tasks, from text-to-video conversion to video editing, showcasing its effectiveness in capturing dynamic elements in general world settings.

Scalable High-Resolution Pixel-Space Image Synthesis with Hourglass Diffusion Transformers

"Hourglass Diffusion Transformer" (HDiT) is an innovative image generative model. HDiT uniquely scales linearly with pixel count, facilitating training directly in pixel-space at high resolutions, such as 1024x1024. It bridges the gap between the efficiency of convolutional U-Nets and the scalability of Transformers, without requiring typical high-resolution training techniques. HDiT shows competitive performance on ImageNet 256^2 and sets a new standard in diffusion models for FFHQ-1024^2. 

Understanding Video Transformers via Universal Concept Discovery

"Understanding Video Transformers via Universal Concept Discovery" explores a new method for making video transformer models more interpretable. It introduces the first algorithm for discovering key concepts in video data, helping to explain how these models make decisions. This approach is unique in its ability to identify dynamic concepts in videos over time, revealing the models' object-centric representations and spatio-temporal reasoning. This research is significant for improving video transformers' interpretability and performance in fine-grained tasks. 

🛠️TOP TOOLS

Yuna - AI-powered mental health app offering 24/7 support, focusing on emotional clarity and anxiety relief. It uses cognitive behavioral therapy techniques, ensures privacy, and provides affordable wellness options.

AI Lawyer - virtual legal assistant offering affordable and fast legal support for both consumers and lawyers. It automates legal research and documentation, simplifying complex legal texts and routine tasks. 

Reading Coach - a tool designed to help improve reading skills. It offers personalized assistance and feedback, focusing on enhancing reading fluency and comprehension.

Winston AI - tool designed to detect AI-generated content, including that from ChatGPT and other large language models. 

Photify AI -  transforms a single selfie into hundreds of different looks. It offers features to explore various hairstyles, genders, skin tones, outfits, and scenarios, all through advanced AI technology. 

📲SOCIAL MEDIA

🗞️MORE NEWS

Google settles AI-related chip patent lawsuit that sought $1.67 bln

Google settled a patent lawsuit with Singular Computing, dodging a $1.67 billion claim over AI chip tech. The suit alleged Google used Singular's tech in AI features across its services. Details of the deal are hush-hush, but Google insists they didn't step on Singular's toes. Singular's founder had shared his inventions with Google earlier, sparking the dispute. Despite the clash, Google maintains its AI chips, vital for functions like speech recognition and content generation, were independently developed and don't infringe on Singular's patents. REUTERS

Apple boosts plans to bring generative AI to iPhones

Apple is making a series of acquisitions, hiring staff, and updating hardware to bring artificial intelligence (AI) to its next generation of iPhones. The company has acquired 21 AI start-ups since the beginning of 2017, focusing on running AI through mobile devices. Apple's goal is to operate generative AI through mobile devices, allowing chatbots and apps to run on the phone's own hardware and software. The company has also unveiled new chips with greater capabilities to run generative AI, and its upcoming mobile software, iOS 18, is expected to enable generative AI features. FINANCIAL TIMES

MLCommons wants to create AI benchmarks for laptops, desktops and workstations

MLCommons, an AI industry group, is creating benchmarks for AI performance on consumer PCs like laptops, desktops, and workstations. Their first focus is on text-generating models, notably Meta's Llama 2. The benchmarks, driven by real-world scenarios, aim to guide buyers on the AI capabilities of devices. Key tech players like AMD, Intel, and Microsoft are involved, but Apple is notably absent. These benchmarks could become crucial in future tech purchasing decisions, potentially expanding to mobile devices too. The initiative is still in its early stages. TECHCRUNCH

SAP announces $2.2B restructuring program that’ll impact 8,000 jobs

SAP, a German tech giant, is launching a €2 billion ($2.17 billion) restructuring initiative, affecting 8,000 jobs, as it shifts focus to Business AI. The plan, part of SAP's strategy for growth in key areas, aims at transforming operations and boosting future revenue potential. This change might not lead to a net decrease in jobs due to voluntary leave programs and internal reskilling. Similar to other tech firms, SAP's restructuring aligns with the industry trend of adapting workforce composition to meet evolving AI demands and maintaining operational efficiency. SAP also reports strong cloud growth, emphasizing its commitment to innovation and AI. CIO

Etsy launches ‘Gift Mode,’ a new AI-powered feature that generates 200+ gift guides

Etsy's new AI feature, "Gift Mode," is a digital quiz that helps you find perfect gifts based on who you're shopping for and their interests. With over 200 tailored guides, it recommends gifts from Etsy's huge range of products. Plus, you can send a sneak peek of gifts by email. Developed using AI and human curation, it aims to make gift shopping stress-free. Etsy's also boosting its focus on AI in search and shopping experiences. TECHCRUNCH

North Korea's AI development raises concerns, report says

North Korea is developing AI and machine learning technologies for various applications, including COVID-19 response, nuclear reactor safety, military simulations, and government surveillance. Despite international sanctions impacting their ability to acquire AI hardware, North Korea is making strategic investments in AI to strengthen its digital economy. The country has been collaborating with foreign scholars, particularly in China, and using AI in innovative ways during the pandemic. Concerns arise over the potential use of AI in cyberattacks and the implications for international sanctions. REUTERS

Shall Google compare thee to a Claude Monet?

Google's Art Selfie feature, now enhanced with generative AI, lets users create personalized images with references to renowned artists and historical art periods. This feature, part of the Google Arts & Culture app, not only generates artistic selfies but also educates users about the art styles or periods depicted. You can imagine yourself in settings like Rousseau's jungles or as a character from Shakespeare's plays. THE VERGE

💰FUNDING

Legal AI startup Spellbook raises $20 mln as sector draws more investments

Artisse AI raises $6.7M for its ‘more realistic’ AI photography app

Norm AI Raised $11.1 Million With Aim to Aid Compliance Chiefs