Art by arvizu_la
GM. This has been a crazy week. We got our first sponsor, Enterpix! 🥳
And we collaborated with Bojan Tunguz from Nvidia to help with his presentation for Kaggle Days in Paris. Big things are on the way.
In today's issue:
💡MOST INNOVATIVE PRODUCT: Metaphor
📰 AI NEWS: The storm around DeviantArt's new AI tool, Notion has big competition, Galactica can summarize academic literature, and GitHub Copilot is getting voice controls. Plus: Startups that raised capital this week
🤯 COOLEST THING: Enterpix
🧠THINGS I LEARNED: GPT-4 could change everything, audio diffusion is worth keeping an eye on, and Lexica is building something awesome
✨FEATURED ARTISTS: arvizu_la,WeavingWithAI,Gothlytical Art, and shaktidreams.
💡MOST INNOVATIVE PRODUCT
Search engines don't change much over the years. They may experience the odd facelift or tweaks to their ranking algorithms, but the underlying methodology remains the same — so that's what we've come to expect.
Metaphor is a new internet search platform that feels a bit like a warm hug. Instead of returning text and image results, it predicts links using the same ideas behind DALL-E and Stable Diffusion.
It feels like Metaphor actually understands what you're looking for, and satisfies the intent behind your search possibly even better than Google. 👇
If you think Metaphor sounds cool, you might also like Andi, which we mentioned briefly in last week's Further Reading. This one feels a lot like Siri, a more collaborative way of searching the web and asking questions.
For our reluctant wordsmiths, check out Jasper below. It's a great timesaver when writing copy.
❤️ We've partnered with Jasper to offer you 10,000 free credits if you signup using our affiliate link. Bonus: it helps support our newsletter and community.
This week, we thought it'd be cool to highlight some new startups in the AI/creative space that are shaking things up with a radical approach.
For our technical readerbase: take a look at Pipeline.ai (Serverless GPU inference for ML models), Inferrd (deploy custom ML models with 1 line), Replicate (run open-source models with a cloud API) and Exafunction (optimize your deep learning inference workload).
And just for fun: Supermeme.ai lets you generate original memes powered by AI in 110+ languages.
The storm around DeviantArt's new AI tool, Notion has big competition, Galactica can summarize academic literature, and GitHub Copilot is getting voice controls.
Last week, DeviantArtannounced a new AI tool called DreamUp (based on Stable Diffusion) for prompt-based image generation. People are pretty angry about the web-scraping nature of Stable Diffusion which resulted in their artworks being used to train the model, and the controversy prompted DeviantArt to publicly clarify its stance on the matter. It also brought out new controls which allow users to choose the amount of AI art that appears in their feeds, and announced that all deviations are now automatically labeled as NOT authorized for use in AI datasets.
New AI-powered Notion rival Mem.ai raised $23.5 million in a funding round led by OpenAI. This latest investment values Mem at $110 million — and is a big indicator that the future of work and productivity tools is firmly anchored in AI.
Galactica, a large language model for science, was released by Papers with Code. They claim it can summarize academic literature, solve math problems, generate Wiki articles, write scientific code, annotate molecules and proteins, and more.
GitHub announced they are expanding GitHub Copilot to be purchased by organizations, and they're running an experiment called 'Hey, GitHub!' which is voice-powered code generation. I love this concept.
Who got funding this week
🤯 COOLEST THING
This week we dive into Enterpix, an AI-Generated Image Search Engine.
Enterpix is sort of similar to Pinterest for AI images. The gallery provides endless prompt inspiration as you scroll, and if you can't find what you're looking for, you can generate your own images for free in square, portrait or landscape resolutions.
If you're totally new to AI art, you can also try out an auto-generated random prompt to get the wheels turning.
To group specific images together for a theme or subject, Enterpix lets you create mood boards called Collections. It's a great place to find, generate and store AI art for easy reference when you need it.
What else we found this week
Need a fun way to learn something new? Whatonearth.xyz generates a paragraph on any topic you ask for, wrapping the facts into a short story that makes it easier for your brain to understand and remember.
Hey Elon, save the bots! Or at least this one. Give it a prompt (text or image), and it will reply back with four AI-generated images.
But if the prompt is vaguely adult-themed... you'll get kittens instead. We think that's a win.
🧠THINGS I LEARNED
GPT-4 could change everything, audio diffusion is worth keeping an eye on, and Lexica is building something awesome
GPT-4 is coming, and it could change everything
There's a huge buzz in Silicon Valley regarding the upcoming release of Open AI's GPT-4, expected to land in the next 1-3 months.
It's been described as "the most anticipated AI model in history" — which isn't too surprising after the massive success of GPT-3.
What do we expect from it? Based on unconfirmed rumors, GPT-4 will likely be multimodal, accepting text, audio, image, and possibly video inputs. It's also expected to be notably cheaper to use.
I've heard from friends that it's just as big of a leap forward as GPT-3 was compared to GPT-2. I'm so excited to see what is possible with GPT-4.
Audio diffusion is worth keeping an eye on
If you're a technical person, here's a quick read. If not, here's what it means in actual English: Platforms like Mubert let you create original audio using a generative music algorithm. This means you can create absolutely unique music — that you own and can use commercially — with just a few clicks.
What are people saying about it? According to Neural Dreams, Mubert is "perfect for background music stuff [such as YouTube videos or podcasts] but not a sound track" just yet. Which means you probably wouldn't expect it to top radio charts as a standalone track.
Lexica has something cooking
One thing that spread fast online in the AI art community this week was tweets from Lexica founder Sharif. He's been tweeting stunning AI-generated photos of house interiors and furniture, leading to much speculation about what they're up to.
I'm hearing that they plan to productize this model they've developed. Excited to see what they do with it because the photos are beautiful.
Further Study 📖
AI in architecture: Blurring the boundary between the physical and the virtual world through AI — read article
AI image editing: Meet Prompt-to-Prompt: An Artificial Intelligence AI Model That Brings Image Editing Capabilities to Text-to-Image Models — read article
AI in healthcare: AI Researchers At Mayo Clinic Introduce A Machine Learning-Based Method For Leveraging Diffusion Models To Construct A Multitask Brain Tumor Inpainting Algorithm — read article
arvizu_la, WeavingWithAI, Gothlytical Art, shaktidreams
This week's contest theme was isometric.
Created by WeavingWithAI
Created by Gothlytical Art
Created by shaktidreams
Glad to have you along for the journey, and thanks for helping us to grow to over 850 subscribers in 5 weeks. We're working on major changes to Lore.com to make it a more useful resource. More to come soon. :)
See you next week!
-Nathan Lands & Michele du Toit
Support the Newsletter