Sunday Rundown #59: More From Anthropic & a Sexy Scientist
Sunday Bonus #19: Four fun Midjourney style references you can use.
Happy Sunday, friends!
Welcome back to the weekly look at generative AI that covers the following:
Sunday Rundown (free): I go through this week’s AI news and share an AI fail for your entertainment.
Sunday Bonus (paid): I share a goodie with my paid subscribers (a guide, tip, walkthrough of a tool, etc).
Let’s get to it.
🗞️ AI news
Here are this week’s AI developments.
👩💻 AI releases
New stuff you can try right now:
Anthropic is on a roll these days:
You can now publish, share, and remix any Artifacts you make with Claude. (Like the interactive courses I showcased.)
When using the Anthropic Console, Claude can now help you generate prompts, create test inputs, and evaluate outputs side by side.
Stability AI has also woken up:
The new license update means more people can use Stability AI products for free and with fewer limitations.
Stable Assistant users got a bunch of new features including “search & replace” for objects inside images and 3-minute track generations using Stable Audio (which was featured in my last “Battle Of the Bands.”)
Poe launched Previews, which work a lot like Claude Artifacts, letting you see and interact with applications generated by any models inside Poe.
Google made its newest small model Gemma 2 (7B and 27B versions) available inside the Google AI Studio.
🔬 AI research
Cool stuff you might get to try one day:
Meta AI will soon make it possible for WhatsApp users to reply to and edit AI-generated images inside the app.
Odyssey wants to deliver Hollywood-grade video AI that lets creators control characters, lighting, and scenes with unprecedented precision.
A startup called Phenomenal AI is India’s first text-to-video platform, now in closed beta. (The demo outputs look quite basic compared to many existing AI video players.)
YouTube is working on an improved Erase Song feature that lets creators automatically strip copyright-infringing music from their videos using AI.
📖 AI resources
Helpful stuff that teaches you about AI:
FoxVox - free Chrome extension from Palisade Research that showcases how AI can be used to manipulate opinions by subtly tweaking content like news, etc.
What is AI? - a fantastic long read from MIT that dives into the history of AI, the different factions in this space, and much more.
🔀 AI random
Other notable AI stories of the week:
OpenAI announced new partnerships aimed at broad societal impact:
Partnership with Arianna Huffington to fund and develop a personalized health coach called Thrive AI Health.
Partnership with Los Alamos National Laboratory to study how AI can be safely used to advance bioscience research.
Microsoft is giving up its observer seat on OpenAI board. (Apple won’t be getting one, either.)
🤦♂️ 10. AI fail of the week
I asked for a scientist pulling himself up by his collar. This…is not that.
Anything to share?
Sadly, Substack doesn’t allow free subscribers to comment on posts with paid sections, but I am always open to your feedback. You can message me here:
💰 Sunday Bonus #19: Four fun Midjourney style references for you to use
I explained how Midjourney’s --sref parameter works back in February.
In a nutshell, --sref applies the style of an uploaded image to your prompt to give it the same look and feel.
But there’s also an option called --sref random that lets you roll the dice and stumble upon completely random style references. These take the form of a number string you can apply to future prompts using --sref [number string].
Many of them are meh, but some of them are a lot of fun.
Today, I’ll share four of the fun ones.
Simply copy-paste their --sref strings and have fun!