• Bot Eat Brain
  • Posts
  • NVIDIA unveiled Project GR00T, its foundation model for robots

NVIDIA unveiled Project GR00T, its foundation model for robots

PLUS: YouTube "helps" you bend the knee


Good morning, human brains, and welcome back to your daily munch of AI news.

Here’s what’s on the menu today:

  • Did you just GR00T yourself? 🤖 💨

    NVIDIA unveiled GR00T, its foundational AI model for humanoid robots.

  • Create high-quality, consistent 3D videos 🎥 🌐

    Stability AI released a text-to-3D-video tool called Stable Video 3D.

  • YouTube forcibly “helps” you follow new regulations 👺 🫣

    It launched a new tool that requires you to disclose AI content.


Rootin' tootin' GR00T-in 🤖 💨

On Monday, NVIDIA announced Project GR00T. It’s a foundational AI model for humanoid robots.

What is it?

Project GR00T stands for Generalist Robot 00 Technology. It enables humanoid robots to understand language, emulate human movements, and learn from observing demonstrations.

Why bother?

NVIDIA claims the goal is to bridge the AI-physical world gap and advance artificial general intelligence (AGI).

How will NVIDIA accomplish this?

NVIDIA partnered with leading robot companies like Figure AI, Boston Dynamics, Apptronik, and more. It rolled out a new virtual training environment, new simulation and training management tools, and more.

Those robotics companies sound familiar.

Well. Back in October, we covered Boston Dynamics’ Spot. It’s a four-legged robot that can walk, run, jump, dance, parkour, and communicate with you.

In January, we reported on Figure and BMW’s partnership. Figure deployed humanoid robots in BMW’s South Carolina plant.

Yesterday, we covered Apptronik’s partnership with Mercedes-Benz. The goal is to integrate Apptronik’s robots into the automobile manufacturing process.

Get a website designed by AI that doesn’t look like it was designed by AI. In 30 seconds or less.

Durable is the AI website builder and platform that makes running a business easier than having a job.

Build a fully functioning, beautiful website in 30 seconds.

Then, start attracting and booking more clients with integrated CRM, SEO, invoicing, and marketing tools.

Try it for free, no credit card is required.


Generate high-quality 3D videos 🎥 🌐

On Monday, Stability AI released Stable Video 3D. It leverages video diffusion models to create 3D videos from an image or text prompt.

What is it?

It’s a 3D video generation model that allows you to create high-quality, consistent videos from any angle. It outperforms current 3D video models in realism, controllability, and more.

Why do I care?

You can use it to create VR/AR experiences, product visualizations, and more.

How can I use it?

There are two versions: SV3D_u for orbital videos from single images and SV3D_p for video creation along specified camera paths. You can use them commercially with a Stability AI membership, or non-commercially through Hugging Face.

What else has Stability AI done?

Back in September, we reported on Stability Audio. It generates complete musical audio, individual stems, and sound effects from text prompts.

In January, we covered Stable Code 3B. It’s Stability’s coding assistant, which you can run offline on your laptop.

Last week, we covered Stability and Midjourney’s drama. Midjourney banned Stability AI employees for alleged data scraping.


Want a byte-sized version of Hacker News? Try TLDR’s free daily newsletter.

TLDR covers the most interesting tech, science, and coding news in just 5 minutes. No sports, politics, or weather.


YouTube “helps” you follow its rules 👺 🫣

On Monday, YouTube announced a new AI disclosure tool for creators. It requires you to notify viewers when your video contains AI content.

Do I have to?

It’s required if your content contains realistic portrayals of people or locations. You don’t have to disclose AI content if it’s animations, special effects, or any content identified as “unreal.”

How does it work?

You can label your content as AI-generated or synthetic during the upload process in Creator Studio. Labels will then be displayed in your video description or directly on the video player.

Why YouTube, why?

YouTube claims it’s to balance creator empowerment with viewer access to transparent content info. It collaborates with the Coalition for Content Provenance and Authenticity (C2PA) to set industry standards.

They empower us with regulations?




Let’s Build AI — a community of AI experts, enthusiasts, and novices.

Speck AI — a simple, intuitive way to automate web workflows.

LLM Pricing — compare the prices of different AI models.

Kater — a data agent that optimizes for AI to reduce workloads.

Think Pieces

Just how “open” is Grok-1? You can download, use, modify, fine-tune, and distill it. Plus, a look at what this means for the AI community.

China has 12 times more working robots than experts originally thought. Where it dominates in quantity, it lacks in robotics innovation.

Will OpenAI’s chips be made in Abu Dhabi? MGX, a state-backed investment fund, is in talks to fund chip production.

Startup News

Apple unveiled MM1. It’s a new suite of state-of-the-art Multimodal Large Language Models (MLLMs).

xAI reportedly pays the highest salaries for AI developers. This, plus the opportunity to work with Musk, makes it a powerful threat against competitors.

NVIDIA is in negotiations to acquire Run:ai. It’s a billion-dollar Israeli startup that allocates computing resources for AI clouds.


VideoAgent — leverages an LLM as an agent to find, comprehend, and organize data to understand large video sequences efficiently.

RAFT — (Retrieval Augmented Fine-Tuning) is a new method to enhance the question-answering capabilities of LLMs.

MusicHiFi — a method of to enhance stereophonic vocoding in music/audio generation models.



AI agents make mistakes and require constant feedback, but they are capable of completing complex tasks.

Tag us on Twitter @BotEatBrain for a chance to be featured here tomorrow.


Until next time 🤖😋🧠 

What'd you think of today's newsletter?

Login or Subscribe to participate in polls.