AI highlights from this week
The most interesting AI updates I read in the last week
Hi readers!
Here’s some of the most interesting AI updates that I read in the last week.
P.S. Don’t forget to hit subscribe if you’re new to AI and want to learn more about the space.
Highlights
1/ A great thread by Sridhar Ramaswamy on future AI market dynamics. Sridhar is the founder of Neeva, a search engine that has its own ChatGPT-like AI built into the search box. Sridhar previously headed advertising at Google.
1/ Yesterday, @vivek7ue and I outlined the companies best positioned to take advantage of AI's breaking dam.
— sridhar (@RamaswmySridhar) 4:27 PM ∙ Jan 12, 2023
Today, we're taking you through what the market looks like today + where it's headed this year.
🔖 Read on...
2/ Jason Calacanis got his hands on a new Chat-GPT iOS app that OpenAI are working on. Check out the demo below:
.@OpenAI is building a ChatGPT-powered iOS app with an iMessage-like interface
— This Week in Startups (@TWiStartups) 2:41 AM ∙ Jan 10, 2023
@Jason got an invite to the private beta, and he demoed it in on today's podcast!
check it out:
3/ This post by is a comprehensive roundup of everything that happened in AI in 2022.
4/ Microsoft plans to acquire a 49% stake in OpenAI worth $10 billion which is literally a Big Deal.
Lots of great, new details about the $10 billion Microsoft/OpenAI deal -- including the fact that $MSFT previously invested $3 billion (not just $1 billion) in OpenAI.
— Alexei Oreskovic (@lexnfx) 5:30 PM ∙ Jan 11, 2023
Inside the structure of OpenAI’s looming new investment from Microsoft and VCs
5/ And finally, ChatGPT jumps the shark thanks to Ryan Reynolds…
You knew it was just a matter of time until we did this (extend the @Mintmobile savings with @OpenAI, that is).
— Ryan Reynolds (@VancityReynolds) 2:20 PM ∙ Jan 10, 2023
Everything else…
Here’s a few other things I read this week:
⚡Yann LeCun (@ylecun) in 1993 demoing LeNet-1⚡
— Jousef Murad 🧠 💻 (@Jousefm2) 11:29 AM ∙ Jan 6, 2023
This was the world's first convolutional neural network for handwriting recognition. Today—29 years later—this same approach powers the majority of computer vision.
A historical moment for machine learning!
Google just released Muse: a text to image model achieving state of the art performance. It is significantly more efficient than diffusion models. Muse also allows for inpainting, outpainting, and mask-free editing without finetuning or inverting the model.
— bleedingedge.ai (@bleedingedgeai) 7:52 PM ∙ Jan 5, 2023
DALL-E generates pixels from text. Now meet its cousin, VALL-E, that generates audio from text @MSFTResearch!
— Jim Fan (@DrJimFan) 4:21 PM ∙ Jan 6, 2023
VALL-E’s resemblance to DALL-E v1 and Parti @GoogleAI is striking. Image and audio are both continuous signals, but they can be quantized into discrete tokens.
1/🧵
How is FSD evaluated at @Tesla?
— Andrea Lonza (@lonzaandrea) 9:44 AM ∙ Jan 6, 2023
@romiphadte & Gabriel Gheorghian talked about it in a recent talk at BazelCon.
There are two ways:
◉ Open-Loop (Fleet) evaluation
◉ Closed-Loop (Simulation) evaluation
The evaluation runs 2M+ simulations per week on 5k+ FSD computers
1/n
So I've been working on something.
— Alistair (@AlistairPullen) 2:48 PM ∙ Jan 5, 2023
Introducing Buildt: Google for your codebase! Our AI-powered search allows you to find code by searching for what it does, not just what it is.
Available now as a VS Code extension for JS and TS projects, with 15 more languages coming soon!
Some fun/obvious ideas for where chatGPT will go:
— andrew chen (@andrewchen) 8:35 PM ∙ Jan 6, 2023
- crawls internet, reads news
- talk to APIs for booking hotels and restaurants, mapping, rideshare, food delivery, etc
- can look at my data/email/etc and ask/answer questions
- personalizes answers to my preferences, geo, etc
what do we have here?
— scott belsky (@scottbelsky) 6:11 AM ∙ Jan 6, 2023
still digging in, but seems like a fascinating application of AI for education: start a chat with any historical figure to learn and converse based on their full history.
“Historical Figures” 🤯
Replacing Siri with ChatGPT via an iOS shortcut has been such a huge productivity upgrade. Highly recommended!
— John Coogan (@johncoogan) 6:28 PM ∙ Jan 8, 2023
Here’s the recipe to make Siri/Alexa 10x better:
— Jim Fan (@DrJimFan) 5:08 PM ∙ Jan 9, 2023
1. Whisper to convert speech to text. Best open-source speech model out there.
2. ChatGPT to generate smart home API calls and/or text response.
3. VALL-E to synthesize speech. It can mimic anyone’s voice sample!
Quick figure 1/3
The real AI winner: Microsoft
— Ben Tossell (@bentossell) 10:10 AM ∙ Jan 10, 2023
- $1bn in OpenAI (2019)
- Owns GitHub which released Copilot
- Dalle powers Bing Image Creator
- Valle which produces an AI voice from 3s of input
- GPT in Word, PowerPoint, Outlook etc
- GPT-powered Bing
- $10bn, 49% stake in OpenAI + 75% profit
Recently, I’ve read and overviewed publications for nearly 20 different large language models (LLMs) from GPT to ChatGPT. Here’s what I learned… 🧵 [1/10]
— Cameron R. Wolfe (@cwolferesearch) 6:56 PM ∙ Jan 10, 2023
My students created The Jam Machine, an AI fueled composer!
— Dr. Tristan Behrens (@DrTBehrens) 1:57 PM ∙ Jan 5, 2023
huggingface.co/spaces/JammyMa…
@huggingface
Thanks for reading The Hitchhikers Guide to AI! Subscribe for free to receive new posts and support my work.