- NATURAL 20
- Posts
- NVIDIA and Jeff Bezos JOIN FORCES to take the Industry by STORM | Rabbit R1 SECRET Revealed…
NVIDIA and Jeff Bezos JOIN FORCES to take the Industry by STORM | Rabbit R1 SECRET Revealed…
PLUS: Chrome Gets AI Upgrades, Chinese AI Startup Impresses with Open Source Model and more.
Wes Roth
January 24, 2024
Sponsored by
Today:
🚀Join this 3-hour power-packed workshop (worth $99) for FREE and learn hacks to 10X work output and grow your business.
With AI & ChatGPT, you will be able to:
✅ Make smarter decisions based on data in seconds using AI
✅ Automate daily tasks and increase productivity & creativity
✅ Skyrocket your business growth by leveraging the power of AI
✅ Save 1000s of dollars by using ChatGPT to simplify complex problems
NVIDIA and Jeff Bezos JOIN FORCES to take the Industry by STORM | Rabbit R1 SECRET Revealed…
Aravind Srinivas, a former OpenAI employee, who left right before ChatGPT became a big deal, despite initial doubts about his career move, his new venture is now booming, valued at half a billion bucks. Big names like Jeff Bezos and NVIDIA are investing in it.
They've teamed up with Teenage Engineering to create the Rabbit R1, a new kind of mobile gadget that's all about natural language interaction. This gadget is powered by Perplexity's AI, delivering real-time answers without any subscription needed.
Google Chrome gains AI features, including a writing helper, theme creator, and tab organizer
Google Chrome's stepping up its game with some cool AI tricks. They're rolling out three fresh features for both Mac and Windows users. First up, there's a smart writing assistant to help you type out stuff like online reviews or party RSVPs. It's like Google's experimental AI search thing, SGE, which helps draft emails in different styles.
Next, there's this handy tab organizer that'll group your open tabs like a pro, saving you the headache of sorting through them yourself. It's perfect for when you're juggling a bunch of stuff online.
Lastly, they're bringing in a new way to customize your Chrome look with AI. You can create your own browser theme using text-to-image AI, choosing the vibe, style, and colors you want.
These new features will be available in the U.S. soon. The writing assistant is coming next month, while the tab organizer and theme creator are rolling out in the next few days.
This Chinese Startup Is Winning the Open Source AI Race
A Beijing startup, 01.AI, led by AI bigwig Kai-Fu Lee, is shaking up the AI world. Their new open-source model, Yi-34B, trumps Meta's Llama 2, nailing top spots on AI rankings. They've just launched a fancy model that chats about pictures.
Unlike others, 01.AI's handing out its AI tech free, hoping to build a fan base of developers. They're flush with $200 million from Alibaba and others. Lee, a big name in AI, sees this as his dream project, making machines that get our chit-chat, not just codes. He's gunning to make the next big AI apps, like how Uber and TikTok changed the game. They're already a hit in the West, with experts like Jerermy Howard singing praises.
Lee's journey from Taiwan to top AI guru is a tale of ambition, from Apple to launching Microsoft's Beijing lab and leading Google China. His company Sinovation Ventures backed hot AI startups in China. Despite US-China tech tensions, Lee's bridging gaps, hoping 01.AI's tech will be a global game-changer in AI.
More Google DeepMind Staff Depart to Launch an AI Startup
Three Google AI experts have left the tech giant to start their own AI venture, Uncharted Labs. They were part of a team at Google DeepMind that worked on AI for creating images and music. Frustrated by the slow pace and red tape at Google, they're now riding the wave of investor interest in AI startups.
Based in New York, Uncharted Labs has already bagged $8.5 million, aiming for a $10 million initial funding. Big names like Andreessen Horowitz are in talks to invest. This move is part of a growing trend where top AI talent is ditching big tech for the startup scene.
Google’s new Gemini-powered conversational tool helps advertisers quickly build Search campaigns
Google's got this new tool, Gemini, jazzing up how ads get made in Google Ads. Now, advertisers can chat with Gemini to whip up Search ad campaigns in a flash. It's pretty slick – you give it your website, and it churns out ad content, picks keywords, and even suggests images using AI.
This cool feature is now up for grabs for English-speaking advertisers in the US and UK, with plans to roll it out globally soon. Google's VP, Shashi Thakur, says it's a real time-saver and ups the quality of Search campaigns.
Andreessen Horowitz’s Connie Chan to Leave as Consumer Focus Shifts to AI
Connie Chan, a well-known partner at venture capital firm Andreessen Horowitz, is leaving. The firm, famous for investing in big consumer hits like Airbnb, is now focusing more on consumer apps using artificial intelligence. This shift comes with changes to their consumer team. Chan, who was big on spotting internet trends from China, might start her own fund. Anish Acharya, previously into enterprise and financial tech, is taking over her role. Chan's departure reflects a cooling interest in Chinese tech investments among U.S. venture capitalists. She's felt out of sync at Andreessen Horowitz, as the firm's interest in China-related stuff has dwindled.
🧠RESEARCH
"Binoculars," a novel method to distinguish between human and machine-generated text uses two similar language models to calculate a score, effectively spotting texts made by AI like ChatGPT. Remarkably, Binoculars works without training data, and it accurately identifies over 90% of AI-generated content with a very low false positive rate. It's a significant step forward in detecting AI-written texts.
"Lumiere" by Google is a groundbreaking text-to-video diffusion model. Unlike previous methods, Lumiere generates entire video sequences in one go, ensuring high-quality and temporally consistent videos. It leverages a Space-Time U-Net architecture and a pre-trained text-to-image model for this task. The model excels in creating realistic, diverse motions and offers applications in video editing like video inpainting and stylized generation.
"Meta-prompting" is a technique enhancing language models (LMs) by breaking complex tasks into smaller subtasks. Each subtask is handled by a specialized instance of the same LM, improving efficiency and accuracy. This method simplifies user interactions and integrates tools like Python interpreters, showing significant improvements over standard prompting methods in various tasks.
"OK-Robot: What Really Matters in Integrating Open-Knowledge Models for Robotics" addresses the challenges in the field of robotics. Despite advancements in vision, language, and robotics, general-purpose applications of robotics still face limitations. The paper introduces the OK-Robot framework, which combines Vision-Language Models (VLMs) for object detection, navigation primitives for movement, and grasping primitives for object manipulation. OK-Robot is designed to perform pick-and-drop operations without requiring specific training and was evaluated in 10 real-world home environments, achieving a 58.5% success rate in open-ended tasks. The paper emphasizes the importance of nuanced details when combining open knowledge systems with robotic modules.
Enhancing text-to-image diffusion models using Reinforcement Learning (RL) to align them with human preferences and ethics by applying RL across millions of images with diverse reward functions, the authors significantly improved the quality and diversity of images generated by pretrained Stable Diffusion models. Their approach led to human-preferred images 80.3% of the time, compared to the base model, demonstrating improved composition and diversity.
"DiffusionGPT" combines the prowess of Large Language Models (LLMs) with domain-specific generative models for next-level text-to-image generation. It's a clever system that understands diverse prompts and selects the best model for each task, thanks to its unique Trees-of-Thought approach. This allows for more flexible inputs and top-notch performance across various domains. It's further enhanced by Advantage Databases, which use human feedback to fine-tune model selection. This innovative approach demonstrates significant strides in image synthesis across diverse scenarios.
🛠️TOP TOOLS
Superhuman - AI-powered email platform for teams and individuals, designed to enhance productivity. It features AI-assisted email writing, advanced email management tools like snoozing and follow-up reminders, team collaboration features, and integration with social media for insights on contacts.
Katana - cloud-based inventory software designed for seamless management of stock levels, production, and sales across various channels, tailored for small to medium-sized businesses.
Check Get AI - browser extension that integrates ChatGPT, allowing users to utilize AI assistance on any website. It offers features like summarizing, translating, correcting text, and more
Meco - newsletter aggregator designed to declutter your email inbox. It organizes newsletters into a single app, making it easier to read and manage them.
Nerf Studio - provides a modular framework for developing Neural Radiance Fields (NeRFs), designed to make the technology more user-friendly and interpretable.
💰FUNDING
Startup AiDash raises $50 million for tech using AI, satellites to spot wildfire risk
AiDash, a California-based startup, has raised $50 million for its technology that combines AI and satellites to identify wildfire and weather risks on power lines. The funding reflects Silicon Valley's focus on products aiding energy companies in climate change adaptation. AiDash's AI scans satellite images to spot changes and problems along power lines, like encroaching vegetation. This tech also supports biodiversity reporting regulations. The funding, led by Lightrock, includes participation from National Grid Partners and other investors. REUTERS
Cybersecurity automation firm Torq lands $42M in expanded Series B
Torq, a cybersecurity startup specializing in "hyperautomation," has raised an additional $42 million in its Series B funding round, bringing its total funding to $120 million. The startup has seen significant growth, with a 300% revenue increase and 500% client base growth in 2023. It currently serves around 100 enterprise customers, including high-profile brands. The company plans to use the new funding for platform expansion, AI capabilities enhancement, international growth, and increasing sales channel presence. Torq competes with firms like Fortinet, Tines, and Swimlane and is set to grow its 150-employee workforce by 30% by the end of 2024. TECHCRUNCH
Open source vector database startup Qdrant raises $28M
Qdrant, a Berlin-based startup, has raised $28 million in Series A funding led by Spark Capital. Specializing in open-source vector databases, Qdrant focuses on aiding developers in managing unstructured data, crucial for generative AI. The company's recent advancements include a new compression technology called binary quantization, significantly boosting efficiency. Qdrant plans to expand its business team and has recently launched an on-premise edition for enterprises, alongside its cloud services on Azure, AWS, and Google Cloud Platform. The funding underlines the growing interest in vector databases amid the AI revolution. TECHCRUNCH
🗞️MORE NEWS
Google aims to deliver "world's most advanced AI" in 2024, and it certainly has a long way to go
Google's gunning for the top spot in AI in 2024, shooting to create the world’s most advanced, safe, and responsible AI. They're mixing AI into their current products, like business apps and phones, but haven’t hit the jackpot with a solo AI product like ChatGPT yet. They’re also trailing behind Microsoft in cloud biz, partly because of their AI lag. Despite fears, AI hasn’t messed up Google's search game yet, but they’re still battling AI spam. DECODER
Alphabet cuts ties with Australian AI firm that helped train Bard and Google Search
Alphabet has severed ties with Appen, the AI data firm crucial in training Google’s Bard chatbot and enhancing Google Search. This move significantly impacts Appen, as Alphabet represented about a third of its revenue. Appen, known for its work with major tech giants like Microsoft and Apple, is facing a tough road ahead with declining revenues, despite the rising demand for AI training data. Issues like labor disputes and a struggle to adapt to generative AI have also challenged the company. The termination of this partnership marks a major shift in Alphabet's AI strategy and could have broader implications for the AI industry. CNBC
Meet your AI assistant for education: Microsoft Copilot
Microsoft Copilot, a generative AI tool, is set to transform education by assisting educators in creating personalized content, lesson plans, quizzes, and more. It uses advanced AI to generate text, images, code, or audio, learning from existing data and language context. Copilot can tailor content for individual student needs, brainstorm ideas, plan lessons, provide feedback, and offer quick answers with source links. Educators are advised to craft clear, specific prompts for effective guidance of AI. Copilot also has a feature for creating images from text descriptions. Microsoft ensures data protection and privacy in Copilot, aligning with their AI principles and Responsible AI Standard. This tool aims to streamline educational tasks, enhance learning experiences, and is part of a broader range of Microsoft's AI educational resources. MICROSOFT
Kin.art launches free tool to prevent GenAI models from training on artwork
Kin.art, a platform for managing art commissions, has launched a free tool aimed at preventing generative AI models from using artists' work without permission. The tool employs methods like image segmentation and tag randomization to disrupt AI training processes. Developed by Kin.art's co-founder Flor Ronsmans De Vry, this tool aims to ethically balance traditional and generative art. While the tool is free, artists must upload their work to Kin.art's portfolio platform to use it, potentially leading them to the company's fee-based services. TECHCRUNCH
Los Angeles Startup Uses AI To Reduce The Carbon Footprint Of Concrete
Concrete.ai, a Los Angeles-based startup, is leveraging AI to revolutionize concrete manufacturing, aiming to reduce its carbon footprint. Utilizing UCLA research, their AI software has shown a 30% reduction in emissions and a cost saving of over $5 per cubic yard in field tests. The startup's approach involves optimizing concrete mixes with less cement but maintaining strength, addressing cement's significant greenhouse gas emissions. They plan to announce the commercial availability of their technology, Concrete Copilot, at the World of Concrete event in Las Vegas, targeting a substantial reduction in global carbon emissions through optimized concrete formulations. FORBES
Sevilla FC Transforms the Player Recruitment Process with the Power of IBM watsonx Generative AI
IBM and Sevilla FC have introduced "Scout Advisor," a new AI tool built on IBM's watsonx platform, to enhance their player recruitment process. This tool integrates with Sevilla FC's extensive database, utilizing natural language processing and large language models for detailed analysis of potential recruits. It aims to combine traditional scouting methods with advanced data analysis, providing comprehensive player evaluations. This collaboration is part of Sevilla FC's broader strategy to use data and AI in football operations, enhancing their decision-making and competitive edge in player recruitment. IBM
New study on coding behavior raises questions about impact of AI on software development
A new study by GitClear, a Seattle-based developer analytics tool, raises concerns about the impact of AI on software development. The research analyzed 153 million changed lines of code from 2023, noting an increase in "code churn" and "copy/pasted code," which suggests potential drawbacks in code quality. The study warns of "AI-induced tech debt" as AI tools are adept at adding code quickly but may not integrate well into broader projects. This trend could influence software development's future quality and how engineers are compensated. GEEKWIRE
The Pentagon is already testing tomorrow’s AI-powered swarm drones, ships
The Pentagon's cutting-edge AI and drone experiments are revolutionizing military tactics. The Navy and Air Force are testing swarms of autonomous air and sea drones, like at the Unitas exercise, to enhance reconnaissance and attack capabilities. These systems are expected to play a vital role in future conflicts, especially in jamming and electronic warfare scenarios. The U.S. aims to set global standards for AI in military use, focusing on safety and reducing miscalculation risks. This strategy sets the U.S. apart from rivals like Russia and China in autonomous weaponry development. DEFENSE ONE
UAE President announces the establishment of the AI and Advanced Technology
The President of the UAE, HH Sheikh Mohamed, has announced the establishment of the Artificial Intelligence and Advanced Technology Council (AIATC). This initiative demonstrates the UAE's dedication to embracing technological advancements and fostering innovation. The AIATC will play a pivotal role in shaping the country's digital future, with a focus on research, infrastructure, and investments in artificial intelligence and advanced technology in Abu Dhabi. This move aligns with Abu Dhabi's strategy to become a global hub for investments, partnerships, and talent in the AI and advanced technology sectors, contributing to the post-hydrocarbon economy's growth and prosperity. CIO
Anthropic’s Gross Margin Flags Long-Term AI Profit Questions
New data reveals that the profit margins of AI startups like Anthropic may be lower than those of established enterprise software companies. Specifically, Anthropic's gross margin, which is the gross profit as a percentage of revenue, was reported to be between 50% and 55% in December. This figure is considerably lower than the average gross margin of 77% observed in cloud software stocks, according to Meritech Capital. The lower profit margins in the AI industry are attributed to the costs associated with customer support and the infrastructure required to power AI systems. THE INFORMATION
Robotic Breakthrough Mimics Human Walking Efficiency
Researchers at Tohoku University have achieved a groundbreaking advancement in robotics by replicating human-like variable speed walking using a musculoskeletal model controlled by a reflex control method akin to the human nervous system. This development sets new standards for robotic technology and enhances our understanding of human locomotion. An innovative algorithm was employed to optimize energy efficiency across various walking speeds, offering potential benefits for bipedal robots, prosthetics, and powered exoskeletons. This breakthrough could improve mobility solutions for individuals with disabilities and advance everyday robotics. The team aims to further refine the reflex control framework for a broader range of human walking speeds and movements. NEUROSCIENCE NEWS
What'd you think of today's edition? |
What are MOST interested in learning about AI?What stories or resources will be most interesting for you to hear about? |
Reply