Apple Vision Pro Now Controls Real Robots
Plus: OpenAI hoards AI detection tools, Nvidia delays new AI chip, Google's Gemini 1.5 Pro leads AI chatbot rankings, and more.
Hello Engineering Leaders and AI Enthusiasts!
This newsletter brings you the latest AI updates in just 4 minutes! Dive in for a quick summary of everything important that happened in AI over the last week.
And a huge shoutout to our amazing readers. We appreciate you😊
In today’s edition:
🤖 Nvidia AI powers robots with Apple Vision Pro
🕵️♂️ OpenAI has a secretive tool to detect AI-generated text
👀 Tesla's AI gives robots human-like vision
💻 Nvidia delays new AI chip launch
🚀 Google's Gemini 1.5 Pro leads AI chatbot rankings
🧠 AI turns brain cancer cells into immune cells
📚 Knowledge Nugget: AI #75: Math Is Easier by
Let’s go!
Nvidia AI powers robots with Apple Vision Pro
Nvidia introduced a new tool suite for developers to control and monitor robots using Apple's Vision Pro headset. The MimicGen NIM microservice translates user movements captured by the Vision Pro into robot actions, enabling intuitive control of robotic limbs.
Additionally, Nvidia's Isaac Sim can generate synthetic datasets from these captured movements, which reduces the time and cost of collecting real-world data for robot training.
Why does it matter?
This advancement is a practical application of teleoperation. It can lead to more intuitive and effective ways for humans to interact with and control robots and improve their usability in various fields such as manufacturing, healthcare, and service industries.
OpenAI has a secretive tool to detect AI-generated text
OpenAI has been sitting on a tool that can detect AI-assisted cheating for nearly a year. Using an invisible watermarking technique, the company has developed a tool that can detect ChatGPT-generated text with 99.9% accuracy. However, internal debates about user retention, potential bias, and distribution methods have kept this technology under wraps.
Meanwhile, educators are desperately seeking ways to detect AI misuse in schools. A recent survey found that 59% of middle- and high-school teachers were confident some students had used AI for schoolwork, up 17 points from the previous year.
Why does it matter?
This tool could preserve the value of original thought in education. However, OpenAI's hesitation shows there are complex ethical considerations about AI detection and unintended consequences in language communities.
Tesla's AI gives robots human-like vision
Tesla's latest patent introduces a vision system for autonomous robots, particularly its humanoid robot Optimus. The end-to-end AI model uses only camera inputs to create a detailed 3D understanding of the environment, without using expensive sensors like LiDAR.
By dividing the space into voxels (3D pixels), the system can predict each spatial unit's occupancy, shape, semantics, and motion in real-time. It has already been implemented, with Tesla's manufacturing team training and deploying the neural network in Optimus for tasks like picking up battery cells on a conveyor belt.
Why does it matter?
The development of such AI-driven perception technologies could lead to progress in autonomous systems for more sophisticated and reliable operations.
Nvidia delays new AI chip launch
The Information reports that design flaws could delay the launch of Nvidia's next-gen AI chips by three months or more. This setback could affect giants like Microsoft, Google, and Meta, who have collectively placed orders worth tens of billions of dollars for these chips.
Despite the rumored delay, Nvidia maintains that production of its new Blackwell chip series is on track. The company also reports strong demand for its Hopper chips and says a broad sampling of Blackwell has already begun. However, sources claim that Microsoft and another major cloud provider were informed of production delays just this week.
Why does it matter?
A slowdown in chip availability could hamper the development and deployment of new AI technologies, affecting everything from cloud services to generative AI applications. It also highlights the delicate balance and vulnerabilities in the AI supply chain.
Google's Gemini 1.5 Pro leads AI chatbot rankings
Google has launched Gemini 1.5 Pro, an experimental version available for early testing. It quickly claimed the top spot on the LMSYS Chatbot Arena leaderboard, outperforming OpenAI's GPT-4o and Anthropic's Claude-3.5 Sonnet. With an impressive Elo score of 1300, Gemini 1.5 Pro excels in multilingual tasks, technical areas, and multimodal capabilities.
The model builds on the foundation of Gemini 1.5, boasting a massive context window of up to two million tokens.
Why does it matter?
Google's decision to make the model available for early testing reflects a growing trend of open development and community engagement in the AI industry. The company's focus on community feedback also reflects its move toward responsible AI development.
AI turns brain cancer cells into immune cells
Researchers at the Keck School of Medicine of USC used AI to reprogram glioblastoma cells into cancer-fighting dendritic cells. It increased survival chances by up to 75% in mouse models of glioblastoma, the deadliest form of brain cancer in adults. The technique cleverly bypasses the blood-brain barrier by converting cancer cells within the tumor itself, a major hurdle in traditional glioblastoma treatments.
The approach greatly improved survival rates in animal models when combined with existing treatments like immune checkpoint therapy or DC vaccines. The research team aims to begin clinical trials in patients within the next few years
Why does it matter?
The technique offers new hope for patients facing this aggressive disease. Moreover, the approach's application to other cancer types suggests a broader impact on cancer immunotherapy, transforming how we approach cancer treatment in the future.
Enjoying the latest AI updates?
Refer your pals to subscribe to our newsletter and get exclusive access to 400+ game-changing AI tools.
When you use the referral link above or the “Share” button on any post, you'll get the credit for any new subscribers. All you need to do is send the link via text or email or share it on social media with friends.
Knowledge Nugget: AI #75: Math Is Easier
In this detailed and interesting article,
talks about rapid AI advancements, including DeepMind's silver medal performance at the International Mathematical Olympiad and OpenAI's launch of SearchGPT. The article also delves into debates on ongoing AI safety regulations, particularly California's SB 1047 bill.Anthropic's letter proposing significant changes to the bill and OpenAI CEO Sam Altman's op-ed framing AI development as a geopolitical race underscore the complexity of AI policy and development. The article also touches on the debate over the reliability of AI existential risk probability estimates and the increasing integration of AI into everyday technologies and services.
Why does it matter?
While the AI's performance is impressive, the article also notes the need for human assistance in translating problems and the extended time the AI takes compared to humans. These points of skepticism are important for realistic expectations and guiding future improvements.
What Else Is Happening❗
🔄 OpenAI's co-founder John Schulman has left for rival Anthropic and wants to focus on AI alignment research. Meanwhile, another co-founder and president of OpenAI Greg Brockman, is taking a sabbatical.
🤖 Figure, an AI startup backed by OpenAI, teased its latest “the most advanced humanoid robot on the planet” Figure 02.
🎤 Meta is offering Judi Dench, Awkwafina, and Keegan-Michael Key millions for AI voice projects. While some stars are intrigued by the pay, others disagree over voice usage terms.
⚖️ YouTube creator David Millette sued OpenAI for allegedly transcribing millions of videos without permission, claiming copyright infringement and seeking over $5 million in damages.
🤝 Google hired Character.AI's co-founders Noam Shazeer and Daniel De Freitas for the DeepMind team, and secured a licensing deal for their large language model tech.
🖼️ Black Forest Labs, an AI startup, has launched a suite of text-to-image models in three variants: [pro], [dev], and [schnell], which outperforms competitors like Midjourney v6.0 and DALL·E 3.
🗣️ OpenAI has rolled out an advanced voice mode for ChatGPT to a select Plus subscribers. It has singing, accent imitation, language pronunciation, and storytelling capabilities.
📝 Google's latest Gemini ad shows a dad using Gemini to help his daughter write a fan letter to an Olympian. Critics argue it promotes lazy parenting and undermines human skills like writing. Google claims the ad aims to show Gemini as a source of initial inspiration.
➡️ Stability AI has introduced Stable Fast 3D which turns 2D images into detailed 3D assets in 0.5 seconds. It is significantly faster than previous models while maintaining high quality.
🔍 Google's "About this image" tool is now accessible through Circle to Search and Google Lens. With a simple gesture, you can now check if an image is AI-generated, how it's used across the web, and even see its metadata.
New to the newsletter?
The AI Edge keeps engineering leaders & AI enthusiasts like you on the cutting edge of AI. From machine learning to ChatGPT to generative AI and large language models, we break down the latest AI developments and how you can apply them in your work.
Thanks for reading, and see you next week! 😊