First-ever World Model for Autonomous Drive

Today’s top AI Highlights:

1. First-ever Generative World Model for Autonomous Driving

2. Efficient, Eco-Friendly and Quality Text-to-Image

3. Meta’s Generative AI Features for Branding

4. Google Announces Major AI Upgrades in Pixel, Android and Hardware

5. Runway’s Gen-2 in Canva

& so much more!

Read time: 3 mins

Latest Developments 🌍

Generative World Model for Autonomous Driving 🚘

Wayve, the leading self-driving technology company, unveils GAIA-1, the first-ever generative world model with 9 billion parameters, designed exclusively to enhance and accelerate the training of end-to-end AI software for autonomous driving.

Key Highlights:

• GAIA-1 leverages video, text, and action inputs with state-of-the-art generative AI techniques, creating simulated driving videos mirroring real-world conditions.

• Its world modelling task, focused on the next token prediction within the context of videos, shares the scaling behaviors that have become a hallmark of LLMs.

• GAIA-1's impressive ability to predict diverse futures, behaviors, and traffic scenarios, while being controlled through text prompts, opens new possibilities for enhancing autonomous driving technology.

Efficient, Eco-Friendly and Quality Text-to-Image 🌿

Researchers introduce PIXART-α, a transformative text-to-image model that combines state-of-the-art image generation quality with remarkable training efficiency, reduced costs, and significantly less environmental impact.

Key Highlights:

• PIXART-α dramatically reduces training time, consuming only 10.8% of Stable Diffusion v1.5's training time, saving approximately $300,000 and slashing CO2 emissions by 90%.

• This model employs a strategic decomposition of training tasks, efficient T2I Transformers, and utilizes highly informative data to optimize pixel dependency, text-image alignment, and image aesthetic quality.

• PIXART-α, when combined with Dreambooth, can generate high-fidelity images with natural interactions, precise color modifications, and customized extensions.

Meta's AI-Powered Ad Toolkit 📺

Meta is introducing its first generative AI features for advertisers, allowing them to leverage AI for:

• creating backgrounds for their product images,

• expanding images to fit various aspect ratios for advertising formats, and

• generating multiple versions of ad text based on their original copy.

Meta will introduce more AI features, including generating ad copy to highlight selling points and creating generative backgrounds with tailored themes, and enable businesses to use AI for messaging on WhatsApp and Messenger for e-commerce and support.

Bard in Google Assistant, & more! 💁

Google released new Pixel devices, AI chip and the new Android with tons of AI features:

• Bard is integrated into Google Assistant to handle a wide range of tasks and provide contextually aware assistance across Google services like Gmail and Docs.

• Pixel 8 and Pixel 8 Pro:

  • Powered by the third-generation Tensor G3 chip.

  • AI-driven camera upgrades - Improved low-light photography, better Macro Focus, and advanced telephoto capabilities.

  • Additional AI feature - Webpage summarization, enhanced Call Screen with spam call detection, Video Boost for superior video quality, and Recorder to transcribe spoken words accurately.

• Tensor G3 Chip:

  • Significant hardware upgrades, including the latest ARM CPUs and an upgraded GPU.

  • Allows on-device generative AI, which is 150 times more complex than the most complex model running previously.

• Google Photos: Best Take, Magic Editor for photo editing, and Audio Magic Eraser for video sound refinement. Zoom Enhance lets you crop and zoom into photos after the fact and use generative AI to fill in pixel gaps.

• Android 14: AI-generated personalized wallpapers and AI-enhanced lock screen for displaying prominent information.

Tools of the Trade ⚒️

• n8n: Build complex automations quickly without extensive coding, with deep data integration, flexibility, user-friendly UI, and the ability to connect APIs.

• Interactive Scenes by Luma AI: Create high quality embeddable, and universally shareable 3D assets quickly from a single image, powered by Gaussian Splatting.

• Text-to-Video in Canva: Runway and Canva have partnered to bring Runway's Gen-2 text-to-video generation model into Canva's platform.

• Induced AI: Employ AI agents with human-like reasoning capabilities to automate tasks, delegate workflows, activate them remotely, and run multiple tasks concurrently.

• AI Emojis: Generate emojis from text, free and opensource.

Hot Takes 🔥

• Very concerning that most public conversations today in AI are happening with non-scientific terms with lose definitions like "frontier AI", "AGI", "AI safety", "proliferation",... that are biasing the whole debate with very little to no science associated to them. These are serious topics that should be led by rigorous science papers and processes. ~ Clement Delangue

• Surprised nobody has done a generative AI insurance company using the float to fund GPUs ~ Emad Mostaque

Meme of the Day 🤡

That’s all for today!

See you tomorrow with more such AI-filled content. Don’t forget to subscribe and give your feedback below 👇

Real-time AI Updates 🚨

⚡️ Follow me on Twitter @Saboo_Shubham for lightning-fast AI updates and never miss what’s trending!!

PS: I curate this AI newsletter every day for FREE, your support is what keeps me going. If you find value in what you read, share it with your friends by clicking the share button below!