DALL·E 3, AlphaMissense and Improvements to Google Bard – Live and Learn #25
Welcome to this edition of Live and Learn. Lots of amazing stuff has been announced in the last two weeks: Better features for Google Bard, supreme music generation capabilities from Stability AI, and some serious improvements to image generation with DALL·E 3. As always I hope you enjoy this edition of Live and Learn.
✨ Quote ✨
It seems that the more places I see and experience, the bigger I realize the world to be. The more I become aware of, the more I realize how relatively little I know of it, how many places I have still to go, how much more there is to learn.
– Anthony Bourdain
Links
DALL·E 3 Announcement by OpenAI. Dalle-3 has been announced by OpenAI and WTF are their demos mindblowing. If it is anything like they show in their announcement it will be absolutely amazing. Dalle-3 will be integrated into ChatGPT and will be available for subscribers starting in October. This new model can alter images via an interactive text prompt, which means you can edit and fine-tune your image with further descriptions. Unlike Midjourney, Dalle-3 will be able to put accurate text into the generated images too. This will truly be a game-changer for generative art and I can't wait to try it out.
Improvements to Bard by Google. Bard finally gets good integrations to the outside world, similar to ChatGPTs plugins. Google is a bit late to the game with this development, but I am still excited to try it out and see how easy it makes working with Google products. Integrating and connecting Bard into everything Google has (Gmail, Drive, Photos, Docs, etc.) will be a huge opportunity and I am glad they are finally doing it. My brain goes crazy with excitement about all the potential for automation and improvement such integrations open up. AI assistants, deeply integrated into our daily workflows and lives will soon be a thing and I am super hyped for Google leading efforts on this front.
Stable Audio by Stability AI. StabilityAI is famous for its open-sourced and easily accessible Stable Diffusion model. They kickstarted a whole host of innovation in the open-source image generation community, back when they released Stable Diffusion to the public. And now they have also entered the generative audio game: Their Stable Audio model is available here, and will help generate beautiful music with AI. Right now tracks are limited to 45 seconds in the free plan, and 90 seconds in their paid version but I think this will improve drastically in the future–enabling a whole new way of creating music for games, TV shows, YouTube videos, and even Spotify.
AlphaMissense Genetic Mutations Catalogue by Deepmind. Genetic Mutations can cause serious diseases like sickle cell anemia or even cancer. But for most of the mutations we didn't know if they were likely benign or damaging. In this work, the DeepMind team extended their work on AlphaFold and used it to predict the harmful effects of these missense mutations. They basically charted whether or not a mutation is likely to be "ok" or damaging. This way researchers can now focus their efforts on understanding the most likely disease-causing candidates–a huge win.
🌌 Traveling 🌌
I am roaming around the Himalayas in Northern India's Ladakh right now and it's unbelievably beautiful here. Prayer flags waving in the wind, rivers rushing into valleys, amazing people and good food. That's all I need to feel alive and happy.
🎶 Song 🎶
Morning Matters by Yazmin Lacey
That's all for this time. I hope you found this newsletter useful, beautiful, or even both!
Have ideas for improving it? As always please let me know.
Cheers,
– Rico