We did it! Over 2k subs in 3 weeks of this daily email. Thank you all so much 🥹. And welcome, to the 162 new folks joining the action since yesterday.

The weekend's here and I'm going to London to be a social butterfly.

Let's get to it. I'll see you Monday!

Prompt: List the topics in today’s email:

🙂 Pieter Levels new AI project📝 Overview of the AI writing landscape📖 A videobook🧃 Product marketing with AI

🫦 BEN'S BITES

• Tony Tassell (hey, similar to Tossell!), wrote a piece in the FT²³⁴ about the rise of AI-generated images - what is now possible and what is coming 🔮.

• Sony’s new camera¹⁴⁹ uses a dedicated AI processor to identify and track subjects. It’s only $3,800 😳.

• Pieter Levels is at it again, this time with an AI Avatar project³⁶³. Upload some photos of yourself, train a model on it and then generate 30 AI avatars of your own face in different styles. I don’t need to pay $30 to see my mugshot in different styles… but people are paying.

• A great overview of the AI writing landscape³⁷⁸. This post looks to the future and touches on; price, regulation and ethics.

• AI-generated audio summaries of podcasts²⁷³.

👋 Too many links?! I created a database for all links mentioned in these emails. Refer 1 friend using this link and I'll send over the link database.

• A new model for Structure-based Drug Design¹⁵⁰ with Equivariant Diffusion Models

• Yesterday we saw Spide-verse and Arcane-tuned models, and today there are; Disney¹⁸², Dark Academia¹²⁹, Archer¹⁵¹, Cyberpunk¹³³ and tons more being released.

• Broken Neural Scaling Laws¹⁵⁴. This research proposes a much more accurate approach.

• Gradio Crash Course²¹⁰ - Fastest way to build & share Machine Learning apps, by Assembly AI.

• Introducing The Stack¹⁵⁶ - a 3TB dataset of permissively licensed code in 30 programming languages. If you want your code excluded from the training, there is an opt-out form and data governance plan.

• Daniel Gross built a videobook. It generates images as it reads the story to you. Demo here²³² and code here¹²².

• Scale built a way for marketers to AI-generate UNLIMITED and INFINITELY CREATIVE images of products³¹⁸ for - ad creatives - brand campaigns - social media. It's in early access. And no, coconut is horrid - I'm a Pamplemousse kinda guy 😉

• Google acquires Twitter-backed AI avatar startup Alter for $100 million²³⁶.

• Generate and fine-tune music²⁹⁰ on Musika²⁹⁰.

• DALL-E released collections! a way to share your creations with the world and let others see what you made. Midjourney has had this for a while, good to see DALL-E finally release this. An example collection¹⁸¹.

• Yohein built a private mobile & web app²⁰⁷, with no code, that pulls in all school-related emails for kids and summarizes and shows action items. Gmail (filter) → Zapier (tags and gpt3 summarise) → Airtable → Softr.

• Eric Schmidt and his wife, Wendy, have announced their AI in Science Postdoctoral Fellowship to accelerate the next scientific revolution, with a cool $148M in funding. Imperial²²⁹ College and San Diego⁸⁴ are two of the universities chosen.

• The race is on to find the applications of generative AI that will make a mark on the world. “AI's New Creative Streak Sparks a Silicon Valley Gold Rush”²⁰⁴.

• Apple is now hiring for AI/ML residencies¹⁴⁰ for summer 2023. Job listings are here¹²⁸.

• A multiplayer experiment for Stable Diffusion¹⁷³, SD infinity - coming soon. Similar to the SD canvas I shared yesterday.

• This study extends Maestro, a speech-text joint pretraining framework for automatic speech recognition¹³⁷ (ASR), to speech-generation tasks.

• A large-scale Chinese text-to-image diffusion model, which progressively upgrades the quality of generated images - ERNIE-ViLG 2.0¹⁷⁶.

• Post Parrot²⁶³ - A free marketing tool that generates post titles that drive engagement on Reddit. These were the suggestions I got:

• A few pages from a Midjourney-produced printed manga¹²⁹, AbsXcess.

• Intel announces up to $10 billion in cost reductions¹¹⁸ through 2025, after poor performance.

• Here are some pain points while using AI for drug development¹⁰³; smaller datasets, complex data, complicated data labelling, data bias, lack of data standards, lack of biological data and regulation.

• Potential applications of generative art beyond the explosion of images²⁰⁶.

🖼 AI IMAGES OF THE DAY

"Salmon in a river" - well, technically, it's correct!

Tiramisaurus Rex

👋 SEE YA