Hey everyone, and welcome to the 121 new folk joining us since yesterday.

A few of you have asked, so I've created an AI job board³⁶⁵ + talent collective. If you're looking for AI job opportunities, sign up here²⁶⁶ (it's free) and I'll match you with companies hiring (you can remain anonymous too). If your company is hiring, you can post jobs and get access to the talent here²⁰⁹. It's early so expect this to grow over time. Everything is vetted by me.

Let's get into it!

p.s. I keep forgetting to actually link the Discord²⁴⁹ if you're interested.

Prompt: List the topics in today’s email:

🎬 AI-generated Marvel movies🎥 Process, understand and search video🤔 Simplifying AI terminology📊 Generated data charts

🫦 BEN'S BITES

We got the scoop! Sahil Lavingia (founder of Gumroad) has launched Ask My Book⁵³². It’s being launched later today but he gave me the all-clear to announce it here first. Sahil wrote a book ‘The Minimalist Entrepreneur’ and has built this tool so that you can ask it any question and it will answer it in real-time, with a voice cloned from Sahil’s. It’s a really impressive way to see how AI tools combined can create a really helpful experience. He’s hosting a webinar on Friday that explains how he built it²⁵⁷.

A demo of Game art³⁹⁷ made easy - Simple Stable Diffusion to Puppet: Stable diffusion -> concept art -> Automatic Puppet 2d Animation

Universal image segmentation, and how it can be used to improve performance across multiple image segmentation tasks⁴¹⁸. A truly universal framework should be trained only once and be able to achieve state-of-the-art performance. The OneFormer framework proposed in the article is a step towards making image segmentation more universal and accessible.

LumaAI is hiring for multiple roles⁴²⁰!

👋 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.

DreamArtist³⁹⁸: With just one training image it learns the content and style in it, generating diverse high-quality images with high controllability.

A good podcast with Alex Wang of Scale AI³⁴⁰ on the state of AI, startup building, AI in defence + ethics and learning to think, the AI moats and more.

A tutorial on how Corridor Crew put Tom Holland into The Spider-Verse³¹⁷.

The next Google search engine will be Generative AI⁴⁴⁰. The current design of search engines is based on outdated technology. AI could be used to create a new type of search engine that is more efficient. But, training a model is expensive.

A unified model for three motion and 3D perception tasks³⁰⁸.

Any Roam lovers in the house? Use machine learning and graph theory to gain insights into your notes²⁹⁷.

Sieve raised $4m and launched its AI API Beta. Sieve is a platform that makes it easy to process, understand, and search video⁴¹³, and it can be used to create interactive features without setting up a separate backend.

I wrote a little thread on simplifying AI terminology⁵⁶⁷.

Create your own AI-powered glossary⁵⁶⁷.

How fast can you speak English, without making mistakes? Take the Rap God Test²⁹⁷ (read a Wikipedia passage quickly) and get judged by OpenAI Whisper STT scoring your speech's intelligibility. Each error you make adds 1s to your time.

Create your own Avatar story with this Diffusion model.³²¹

Generating Images with OpenAI’s DALL-E^286† via Data Fetcher and Airtable

Podcast version²⁰⁷ of Mario’s “What to Watch in AI” article.

Generated music from the most iconic album covers²⁵³, using Img-to-Music.

Scott Belsky’s take on Generative AI³⁶⁷: what disruptions are overblown, the perils of building start-ups (or features) on others’ models, and what will (and will never) change.

The State of Multilingual AI²¹⁵. How multilingual are current models in NLP, computer vision, and speech? What are the main recent contributions in this area? What challenges remain and how can we address them?

How AI will disrupt productivity tools³⁷⁹. Admin structure, generative research, smart insights & suggestions, and task actions.

Does heavier use of AI assistance result in more vulnerable code²⁴¹? It seems so!

Chula is a tool that lets you type what your chart should show^634†, it’ll find the data and create it.

A new method for pretraining audio representations³⁵⁸ by combining audio data with natural language descriptions.

An end-to-end locomotion system⁴⁷⁴ that is capable of traversing stairs, curbs, stepping stones, and gaps. The system is trained in simulation and transfers to the real world without any fine-tuning.

A new system for decoding visual stimuli from brain recordings³⁴³, called MinD-Vis. The system is designed to improve upon existing methods by providing more faithful details and meaningful semantics in the reconstructed images.

A new text-to-image model called Paella³⁵⁶ requires less than 10 steps to generate high-fidelity images.

The use of score distillation to text-guide a NeRF model to generate a 3D object²⁴³.

How generative AI is changing creative work³⁰⁴.

An AI tool directory⁵⁵² - Futurepedia.

🖼 AI IMAGES OF THE DAY

👋 SEE YA