• NATURAL 20
  • Posts
  • NVIDIAs new 'Foundation Agent' SHOCKS the Entire Industry! | Dr. Jim Fan and agents for any REALITY

NVIDIAs new 'Foundation Agent' SHOCKS the Entire Industry! | Dr. Jim Fan and agents for any REALITY

PLUS: ElevenLabs Raises $80M Series B, AI Art Defender Nightshade and more.

Today:

NVIDIAs new 'Foundation Agent' SHOCKS the Entire Industry! | Dr. Jim Fan and agents for any REALITY

Dr. Jim Fan, a senior AI researcher at Nvidia, discusses the Voyager project and advancements in AI. Minecraft's popularity is due to its open-ended nature, allowing players to explore and create freely. Voyager, without human intervention, can play Minecraft extensively, continuously learning and mastering skills. 

Metamorph, another initiative, can control thousands of robots with varying configurations. ISAC Sim accelerates physics simulations, potentially enabling agents to generalize across virtual and physical worlds. Dr. Fan envisions a future where autonomous AI agents, like those in movies, all link to a single Foundation agent. This agent learns across countless realities, revolutionizing AI capabilities.

ElevenLabs Releases New Voice AI Products and Raises $80M Series B

ElevenLabs just bagged a cool $80 million in Series B funding, backed by big names like Andreessen Horowitz and Sequoia Capital. They're making waves in voice AI, launching new stuff like a Dubbing Studio and a Voice Library marketplace. They've even got a Mobile Reader App cooking. Their tech's big news in various sectors - think education, entertainment, you name it - and it's already a hit with a bunch of Fortune 500 companies. This cash injection is all about upping their game in voice AI and making sure it's used responsibly. Plus, they're rolling out features to make sure AI-generated content is easy to spot. Their goal? Stay on top as the top dog in voice AI.

Nightshade, the free tool that ‘poisons’ AI models, is now available for artists to use

Nightshade, a fresh and free tool for artists, is now out. Made by University of Chicago geeks, it messes with AI trying to learn from artists' stuff. It's like a digital disguise for images, fooling AI into seeing something else. Think of it as a trickster tool, making AI models see a purse instead of a cow. 

Artists need a Mac or a Windows PC to use it. It's like a digital shield and sword combo against AI copying styles. Some dig it, others say it's a cyberattack on AI. It's the latest drama in the art world's battle with AI scrapers.

Tesla finally releases FSD v12, its last hope for self-driving

Tesla's latest software update, FSD Beta v12, is making waves as the company's final shot at fulfilling its self-driving promises. This update shifts the control of the car from traditional programming to AI-driven decision-making, using a neural network trained on millions of video clips. This big leap replaces over 300,000 lines of C++ code, aiming to mimic human driving more closely.

Despite a delay into 2024, v12 is now rolling out, starting with early Tesla FSD Beta testers. These initial users, known for their familiarity with the system and less critical stance towards Tesla, are the first to experience the update.

Google will pitch Bard Advanced as providing ‘complex, better responses’

Google is set to launch Bard Advanced in early 2024, featuring Gemini Ultra, an AI model touted to outperform GPT-4. Bard Advanced aims to provide more complex and superior responses compared to its current version, Bard, which operates on Gemini Pro, a competitor to GPT 3.5. 

Google's latest AI model is still undergoing extensive safety checks and refinements, with Samsung being one of the first to test it. This upgrade reflects Google's continuous efforts to advance its AI capabilities.

LEAK: EU Commission prepares ‘strategic framework’ to boost AI start-ups, generative AI uptake

The European Commission is reportedly preparing a strategic framework to support AI start-ups and promote the adoption of generative AI in key sectors. This initiative, set to be presented at the end of January, aims to establish "AI Factories" as open ecosystems centered around European public supercomputers to facilitate the development of generative AI models and applications. 

These AI-dedicated supercomputers will be available to start-ups and researchers, provided their work meets ethical standards. Additionally, the strategy includes support services for AI start-ups and connections between start-ups and traditional industries. Funding for these initiatives is expected to come from Horizon Europe and Digital Europe programs, as well as EU member states.

🧠RESEARCH

Yi-VL-34B by 01.AI is a groundbreaking open-source vision-language model excelling in bilingual (English and Chinese) comprehension and conversation about images. It's the first of its kind at 34 billion parameters, offering high-resolution image understanding and trained through a meticulous process for seamless multimodal interaction. Despite its top-notch performance in benchmarks, it's still evolving and has limitations like handling only one image at a time and potential inaccuracies in complex scenes​​.

"Depth Anything" by tiktok presents an innovative approach for estimating depth from single images. It uses a huge dataset (~62M images) without specific labels to improve its accuracy. The model stands out in its zero-shot capability, performing well on diverse datasets and real-world photos. It sets new benchmarks when fine-tuned with specific depth data, showcasing its versatility and effectiveness in depth estimation. 

The "Old Photo Restoration" by Alibaba is a unique tool designed for restoring and enhancing old photographs. It utilizes advanced AI technology to rejuvenate damaged or degraded photos, bringing them back to life with improved clarity and quality. This service is particularly useful for preserving historical images or personal memories, providing an easy-to-use online platform for photo restoration. 

"ActAnywhere: Subject-Aware Video Background Generation" by Adobe introduces ActAnywhere, a generative model that automates the creation of video backgrounds tailored to the motion and appearance of the foreground subject. This model, leveraging large-scale video diffusion models, significantly eases the traditionally manual and tedious task of aligning backgrounds with foreground subjects in videos. It performs exceptionally in generating coherent videos with realistic interactions between the foreground and background, showing great potential in various out-of-distribution samples, including non-human subjects.

InstantID is an AI-driven tool designed to identify and analyze objects and texts in images. It operates online, allowing users to upload pictures for quick, automated recognition and interpretation. The service is particularly user-friendly and efficient, streamlining the process of understanding visual content through AI technology​.

🛠️TOP TOOLS

Modelize.ai - simplifies complex tasks for small businesses, automating steps and organizing them, boosting efficiency and ease of use.

Lume - automates data mapping using AI, speeding up data pipeline creation by 10x. It simplifies complex data mappings, ensuring seamless integration and maintenance of data pipelines for businesses, enhancing efficiency and data handling.

Tango - a tool for digital agencies, streamlining client interactions from proposal to payment. It offers easy-to-create, sign-ready proposals, auto-enforcing smart contracts, simplified billing, and flexible payment options, all aimed at accelerating deal closures and payments.

Mentalport -  AI-powered personal coaching for mental well-being. It provides adaptive support, tailored strategies, and gamified interactions to help individuals cope with stress, achieve goals, and lead a fulfilling life, all within a secure and private environment.

Dgm -  a free online diagramming tool designed for web and AI applications. It offers a variety of smart shapes and pre-built libraries for creating diagrams, supports AI-assisted code generation from diagrams using OpenAI's GPT-4 Vision, and provides cloud storage for easy access. 

📲SOCIAL MEDIA

🗞️MORE NEWS

Fake Joe Biden robocall tells New Hampshire Democrats not to vote Tuesday

In New Hampshire, a fake robocall, mimicking President Biden's voice, falsely told Democrats not to vote in the presidential primary, claiming it aids Republicans and Trump's re-election. The state's Attorney General is probing this illegal voter suppression attempt. The call, seemingly using AI, urged saving votes for November instead. A prominent Democrat, whose number was used, filed a complaint, pushing for full legal action against this attack on democracy. This incident raises concerns about AI's impact on elections and democracy. NBC NEWS

Exclusive: Say hi to Ohai.ai from Care.com founder

Ohai.ai, a new startup from Care.com founder Sheila Lirio Marcelo, aims to simplify family schedule management and task delegation through its virtual assistant platform, "O." The platform combines artificial intelligence and human assistance to provide services like schedule summaries, calendar management, and appointment bookings. With a focus on easing family communication and coordination, Ohai.ai, which currently has a team of about 20, including former Care.com leaders, has raised $6 million in seed funding. This venture reflects Marcelo's long-standing commitment to supporting caregivers and families. AXIOS

AI’s latest task is helping supermarkets safely unload soon-to-expire food

Too Good To Go (TGTG), a Danish company known for addressing restaurant food waste, is now focusing on supermarket food nearing expiration. Its new AI-powered tool helps stores manage expiry dates more efficiently, suggesting optimal discount rates and reducing manual checking. TGTG's solution, starting with the SPAR supermarket chain, aims to reduce food waste and increase revenue by selling near-expiry products at discounted prices. This approach addresses the growing demand for cheaper, short-dated goods and contributes to environmental sustainability by reducing food waste. FORTUNE

The Molecular Twin artificial-intelligence platform integrates multi-omic data to predict outcomes for pancreatic adenocarcinoma patients

Molecular Twin, an AI platform developed to predict outcomes for pancreatic adenocarcinoma patients. This innovative platform integrates a wide range of clinical and molecular biomarkers using advanced machine-learning models. It analyzes a vast dataset, including multi-omic molecular features, to accurately predict disease survival. This approach demonstrates significant potential in enhancing clinical care and democratizing precision cancer medicine globally, making it a notable advancement in the field. NATURE

Using AI to break the opioid addiction cycle

Ben Brown, a research assistant professor at Vanderbilt University, is focusing on the opioid crisis by examining the molecular interactions between painkillers like oxycodone and proteins in the body. Awarded a $1.5 million grant from the National Institute on Drug Abuse, Brown is developing AI to analyze billions of potential opioid drugs. His research targets Mu-opioid receptors, which are key in pain modulation and addiction. This innovative study could pave the way for non-addictive painkillers, potentially revolutionizing pain management and addiction treatment. NEWS MEDICAL

New MIT CSAIL study suggests that AI won’t steal as many jobs as expected

MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) conducted a study to determine the impact of AI on jobs. Contrary to some predictions, the study found that most jobs at risk of AI displacement are not economically feasible to automate currently, with only 23% being cost-effective for automation. The study focused on jobs involving visual analysis, such as inspecting product quality. It suggests that the AI disruption to jobs will likely be gradual and provides insights for policymakers and AI researchers to prepare for the future. The study acknowledges some limitations, including not considering AI's ability to augment human labor.

Cops Used DNA to Predict a Suspect’s Face—and Tried to Run Facial Recognition on It

Police departments in the United States are using DNA to predict a suspect's face and then attempting to run facial recognition on it to solve cold cases. In one case, detectives sent genetic information collected from a crime scene to Parabon NanoLabs, a company that claims to turn DNA into a face. Parabon generated a 3D rendering of the suspect's face based on genetic attributes found in the DNA sample and published it in an attempt to solicit tips from the public. WIRED

Parents worry AI-generated influencers are promoting unrealistic beauty standards to kids

Over 12,000 parents want TikTok to clearly label AI-generated influencers who look real, fearing they promote unrealistic beauty standards and body issues for kids. Nonprofit ParentsTogether initiated this push for transparency. While TikTok requires creators to label realistic AI-generated content, it's not consistently applied. Some accounts hint at being virtual influencers in their bios or hashtags, but kids may not grasp this. TikTok has removed some flagged accounts and prohibits content promoting dangerous weight loss behaviors. The concern is especially significant on platforms popular with kids, like TikTok and YouTube, as virtual influencers blur the line between AI and reality. NBC NEWS

Despite OpenAI Claiming New York Times Lawsuit Is 'Without Merit' The AI Giant Tells UK Regulators A Different Story

The New York Times sued Microsoft and OpenAI, accusing them of using its articles to train ChatGPT. This sparked a debate on AI's role in journalism, focusing on copyright issues. OpenAI called the lawsuit baseless but admitted using copyrighted content for training is necessary. They're telling UK regulators a different story, acknowledging problems with their model, like repeating exact phrases from sources. This conflict highlights broader concerns about AI's impact on various creative fields. Despite the controversy, investment in AI remains strong, with a significant portion of venture capital flowing into the sector. The case could reshape AI-journalism relations and Microsoft's role in the AI market. YAHOO! FINANCE

What'd you think of today's edition?

Login or Subscribe to participate in polls.

What are MOST interested in learning about AI?

What stories or resources will be most interesting for you to hear about?

Login or Subscribe to participate in polls.

Reply

or to participate.