The AI-ronman 🚀

Hello and welcome to your essential AI update! Whether you’re a seasoned expert or an eager newbie, there’s something here for everyone. Let’s explore together! 🌍

Quick Takes ⚡

1. SenseNova 5.0 surpasses GPT-4 Turbo in latest benchmarks

2. Adobe's VideoGigaGAN can improve video resolution by 8x

3. Apple releases OpenELM: small, open source AI models designed to run on-device

4. OpenAI's new 'instruction hierarchy' could make AI models harder to fool

5. xAI targets $6 billion funding boost to enhance Grok chatbot, rivaling ChatGPT

6. Sanctuary AI’s new humanoid robot learns faster and costs less

7. The Ray-Ban Meta Smart Glasses have multimodal AI now

Deep Dive 🔍

• SenseNova 5.0 surpasses GPT-4 Turbo in latest benchmarks 💹 
  
  Chinese technology company SenseTime has recently introduced SenseNova 5.0, an updated version of its large-scale model. This iteration surpasses GPT-4 Turbo in almost all significant benchmarks with a ~600 billion parameter setup and a 200,000 context window. Trained on over 10TB of predominantly synthetic data, SenseNova 5.0 excels in areas like knowledge acquisition, mathematical operations, reasoning, and programming. The model, which is soon to include a text-to-video feature with stylistic consistency, has performed extraordinarily well, hinting that China may soon rival U.S. AI technology. However, with limited public access, the model's full capabilities remain unverified.
  
  

  

• Adobe's VideoGigaGAN can improve video resolution by 8x 🎥 
  
  Adobe is elevating the video enhancement landscape with its latest innovation, VideoGigaGAN. This AI model, capable of upscaling low-resolution videos by as much as 8x, leverages a generative adversarial network known as GigaGAN to minimize common upscaling issues like artifacts. Highlighted during Adobe MAX 2023, the model could soon be integrated into Adobe’s suite of products. While still under development, VideoGigaGAN has demonstrated promising capabilities in enhancing video quality, although refining long videos and detailed imagery continues to be a challenge. Check the paper, here.
  
  

  

  
  
  

• Apple releases OpenELM: small, open-source AI models designed to run on-device 📱
  
  Apple has launched OpenELM, a suite of open-source large language models optimized for on-device operations, available via Hugging Face. The suite consists of eight models, ranging in size from 270 million to 3 billion parameters, inclusive of both pre-trained and instruction-tuned variants. Designed for on-device execution, OpenELM emphasizes privacy and potentially superior performance compared to traditional cloud-based solutions. Apple's dedication to on-device AI technology is further demonstrated by the release of the models' weights and training checkpoints under a sample code license. Benchmark tests show that the 1.1 billion parameter model surpasses similar models in performance while requiring less pre-training data. OpenELM is seen as a crucial step toward more private and distributed AI architectures, possibly giving Apple a distinctive advantage in the market.
  

  

• OpenAI's new 'instruction hierarchy' could make AI models harder to fool 🛡️
  
  OpenAI has unveiled a new security measure called 'instruction hierarchy' aimed at enhancing the defense of Large Language Models (LLMs) against prompt injection and other manipulation tactics. This strategy gives precedence to system instructions from developers rather than user inputs or external tools. By employing synthetic data and a method known as 'context distillation', the models are trained to distinguish and prioritize crucial instructions over conflicting ones. Initial evaluations with GPT-3.5 show a notable increase in resistance to prompt extraction by up to 63%, and a reduction in jailbreaking efforts by 30%. Despite occasional user input rejections, the overall stability of the model is maintained.

• xAI targets $6 billion funding boost to enhance Grok chatbot, rivaling ChatGPT 📈
  
  xAI is actively seeking $6 billion in new funding, a substantial increase from its previous target of $3 billion. Major financial support is expected from Elon Musk, along with new key investors like Sequoia, Valor Equity Partners, and Vy Capital. This infusion of capital will be directed towards enhancing the capabilities of Grok, xAI's advanced AI chatbot, positioning it as a stronger competitor to OpenAI's ChatGPT. In addition to funding, xAI is attracting top engineering talent from prestigious companies like Tesla, aiming to accelerate development and innovation.

• Sanctuary AI’s new humanoid robot learns faster and costs less 🤖
  
  The latest release from Sanctuary AI introduces the Phoenix Gen 7 humanoid robot, equipped with substantial upgrades across the board. This new generation features longer operating times, streamlined production, and notably lower costs. Improvements in the robot’s physique enhance dexterity and durability, while upgrades to visual and sensory capabilities push the boundaries of what humanoid robots can perceive. Additionally, the Phoenix’s 'Carbon' AI system now learns new tasks in under a day, significantly speeding up its adaptability. These advancements suggest that the integration of such robots into everyday life is becoming increasingly imminent.
  
  

  

• The Ray-Ban Meta Smart Glasses have multimodal AI now 👓️ 
  
  Ray-Ban's Meta Smart Glasses have enhanced their functionality with a multimodal AI feature that processes different types of data, including images, sounds, and text. This feature, now available to all, enables users to snap photos and receive immediate AI-powered insights. Commands like 'Hey Meta, look and...' allow users to identify various items or translate text directly through the glasses. Although the AI may occasionally err in its identifications, the glasses remain a practical tool for real-time information retrieval in a portable, hands-free format.

GlazeGPT just announced their industry-leading innovation - text-to-MongoDB queries, a revolutionary feature that translates natural language into precise MongoDB queries.🔥This latest update empowers users to effortlessly interact with complex MongoDB datasets and generates complex queries in seconds while understanding your business glossary.

Level up your skills with this 14-day free trial - don't miss out! ⚡️

Thank you for spending this time with us. Keep exploring, learning, and leading in the AI world. We look forward to bringing you more insights next time. Happy innovating! 🌟

Ciao for now!

Author: Poonam 👧 

Karan 😎 🚀 

CEO, GlazeGPT.com