Ben's Bites
Posts
Can you find the cork?

Can you find the cork?

🦾 Robot hands 🌅 Shutterstock's partnership with OpenAI 🔎 Identifying objects in images 📝 Summarise video and audio with an API

Ben Tossell
October 26, 2022

Thanks for reading! Ben’s Bites is a roundup of what’s going on in AI. Not subscribed? Do that here.Link overload? Refer 1 friend with this link and get access to the filterable+searchable database.

Ding-dong! I'm more reliable than the Amazon delivery driver.

Welcome to the 131 new folks getting updated with the latest and greatest in AI.

Let's get into it.

Prompt: List the topics in today’s email:

🦾 Robot hands🌅 Shutterstock's partnership with OpenAI🔎 Identifying objects in images📝 Summarise video and audio with an API

🫦 BEN'S BITES

Six data and AI startups had valuation multiples of 87.5X revenue or more - The Information published their 2022 most promising startups. The AI/Data startups included are; Runway, Vic.ai, Hex, StarTree, Superconductive, Tabular and Onehouse.

Shutterstock will not accept AI art to be uploaded but will reward creators whose work was used to train them. It will also start selling AI-generated stock imagery with help from OpenAI.

Robotic hands can basically solve a Rubik’s cube. That makes one of us.

An argument for why VCs are making a big mistake investing in Generative AI. Will argues that the best implementation of AI is in the boring menial tasks everyone has to do, and AI is very good at. Rather than everyone becoming a digital Picasso.

Related: A thread on the speed of advancement in AI & how much of an impact that is going to have on many jobs, soon.

Identifying objects in images. Google has released a simple and scalable approach for open-vocabulary object detection and image-conditioned detection. Using the demo is pretty fucking cool, in my opinion. Although, it struggled to find the cork...

Is it possible to transform sensitive user data in a way that guarantees fairness while still being useful?

A non-intimidating guide to getting into AI research, and what skills you need to contribute.

👋 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.

AI chip startup Axelera lands $27M in capital to commercialize its hardware, to create a game-changing, user-friendly and scalable technology with superior performance and efficiency at a fraction of the cost.

Meta is making progress in using AI to compress audio files for quick and easy sharing.

DALL-E 2 fails to reliably capture common syntactic processes

Write meeting notes 20x faster with Supernormal. Here’s an example from an hour-long call. But we're getting rid of hour-long calls right 🥹

You can summarise videos, podcasts, and basically, any audio or video with a single API call.

Get 1 second cold boots for large ML models. WTF is a boot? Luckily Erik also wrote a short thread on this.

Using GPT-3 to help you write emails. Not this one though! (yet_

This sweater developed by the University of Maryland is an invisibility cloak against AI. It uses "adversarial patterns" to stop AI from recognising the person wearing it. [insert Harry Potter reference]

These brothers got rejected by AI Grant, and I think it’s a pretty cool tool. Nutrify - Upload a photo of food and find out its macronutrients. Fun fact: Ben’s Bites was what I called my funny food reviews on Instagram.

Compose AI - a free Chrome extension that cuts your writing time by 40% with AI-powered autocompletion.

Yesterday we learnt about NeRFs. Here’s another demo of creating "drone" shots from your phone footage, including the process too.

A book written by GPT-3, illustrated by DALL-E 2 and Stable Diffusion. Cool, now I can read it with AI and get it summarised with AI. An AI circle of life.

A demo showing how you can now generate ads from any URL in seconds with AI.

Stable Diffusion v1-5 Gradio Demo is out on Hugging Face Spaces by Runway

Mind-reading, from a distance! Scientists can now "decode" people's thoughts without even touching their heads.

A state-of-the-art planner for self-driving.

Who’s building a next-gen search engine using LLM/generative AI? Will Google move fast enough vs other incumbents or startups? The economic opportunity here may be bigger than everything else, combined.

Instantly summarise text.

What the Google Spam Update means for AI-generated articles.

Announcing The Farama Foundation - a new nonprofit organization designed in part to house major existing open source reinforcement learning (“RL”) libraries in a neutral nonprofit body.

For the first time and in near-atomic detail, scientists at Oregon Health & Science University (OHSU) have revealed the structure of the key part of the inner ear responsible for hearing. And it immediately suggests mechanisms by which one might be able to compensate for those deficits.