Hey! A few things from me:

• Introducing Ben’s Bites Hackathon! 12-19th December, up to $15k cash prizes up for grabs. More info + signup here^1,075.

• Want to sponsor Ben’s Bites? Here are a few ways³¹⁷ you can partner with me.

• Are you interested in hosting a meetup in your city? Sign up here²⁵⁵. We've had over 120 signups for the London meetup²¹⁹ in Jan.

On to the email!

🤌 Ben's Picks

• Ethan Mollick, a professor at Wharton, tested ChatGPT to see how much of his job could be done by the AI tool; teaching, research, outreach and public engagement. (link⁸¹⁴)

• Whisper v2 was (secretly?) launched via a GitHub commit that folks were quick to spot! (link⁵⁷⁸)

🛠️ Cool Tools

• CookupAI - No-code AI app builder. (link⁹⁵⁰)

• ShowGPT - A collection of ChatGPT prompts. (link^819†)

• Postedby.ai - Create personalised postcards using AI. (link⁵⁷⁶)

• Pitch Mini Yohei - a bot that collects basic data from founders. (link³⁶⁰)

• A browser extension that augments your ChatGPT prompts with web results. (link³⁷⁸)

• ChatGPT for Mac, living in your menubar. (link⁵⁹¹)

• A curated list of awesome tools, demos, docs for ChatGPT and GPT-3. (link⁵⁵⁰)

• ChatGPT directly within Google Docs as an Editor Add-on. (link⁴⁴¹)

• KrishnAI - a natural language interface built on top of GPT and an in-house context engine based on Puranas. (link³⁴¹)

• Perplexity - GPT web search. (link⁴³²)

• Neeva - revolutionising search with an ad-free, privacy-first model. (link⁴⁶⁴)

👋 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.

🎓 Learn

• Learn Prompting - A documentation site teaching how to use prompts. (link⁶³⁰)

🔬 Research

• Music2Vec - a self-supervised learning framework for learning useful representations of raw music waveforms. (link²⁸⁹)

• NeuralUDF - for reconstructing surfaces with arbitrary topologies from 2D images via volume rendering. (link¹⁹³)

• Spatial diffusion model (SDM) for image inpainting uses a few iterations to deliver informative pixels to the entire image. (link¹⁸⁸)

• Adaptive Diffusion for Image Reconstruction (ADIR) exploits the prior learned by diffusion models while retaining agreement with the observations. (link²⁶²)

• Universal Document Processing (UDOP), a foundation Document AI model that unifies text, image, and layout modalities with varied task formats. (link¹⁹⁷)

• M-VADER - a diffusion model for image generation that can be specified using arbitrary combinations of images and text. (link¹⁸³)

• How machine learning models can learn shortcuts to make correct predictions, and how input salience methods can be used to verify that models are not relying on such shortcuts. (link¹⁵⁴)

• RANA - photorealistic synthesis of humans under arbitrary viewpoints, body poses, and lighting. (link⁴²³)

• A self-supervised approach for correspondence estimation that learns from multiview consistency in short RGB-D video sequences, which allows for correspondences to be obtained over longer time frames, increasing both the diversity and difficulty of sampled pairs. (link²¹⁵)

• A new approach to video understanding using a self-supervised learning method called InternVideo. (link³⁵³)

• Reference-Based Non-Photorealistic Radiance Fields (Ref-NPR) is a controllable scene stylisation method that uses radiance fields and a stylised 2D view taken as reference to generate non-photorealistic and continuous novel view sequences with reasonable stylisation in occluded regions. (link²⁶⁴)

• MoFusion - a framework for unified motion synthesis, which uses a Transformer backbone to ease the inclusion of diverse control signals and pretrains the backbone as a diffusion model. (link³⁰³)

🤓 Everything else

• David Spinks wrote about AI in the context of community building. (link²⁶⁹)

• I was on Marketing Against the Grain podcast talking about AI. (link^163†)

• SellScale has raised $3.4m to give AI superpowers to sales teams. (link¹⁹⁰)

• ChatGPT is now on Hugging Face. (link⁴³³)

• ChatGPT writing a product roadmap for Instagram Stories. (link³³⁸)

• How AI agents can better communicate and cooperate in Diplomacy. (link¹⁶⁵)

• An article in the New York Times “The Brilliance and Weirdness of ChatGPT”. (link²⁴⁷)

• Using ChatGPT to simulate a React app. You type in a description of an "app" and it's rendered straight into the browser. (link³³⁷)

• “I Taught ChatGPT to Invent a Language”. (link²⁸²)

🧑‍💻 Who's hiring in AI

VEED.IO - Simple Online Video Editing. VEED is hiring AI / ML engineers¹²⁰ to level up its creative toolkit and make it more magical.

Buildspace - where builders, build! They're looking for an ML/AI instructor to build their new course²¹⁷.

Add your company here¹⁵⁴ | Join the Talent pool¹⁶⁸

🖼 AI IMAGES OF THE DAY

Photo of Super Mario singing on stage at Woodstock 99

👋 SEE YA