Discover more from Machine Learnings
What will the defining AI interaction be?
With deep dives into OpenAI culture, ambition, and aims for AI alignment.
We’re back with another edition of Machine Learnings, brought to you by the folks at Heyday.
Heyday is an AI-powered memory assistant that resurfaces content you forgot about while you browse the web.
We have a future-oriented lens in just about everything we see. How will this invention change the world? What else may come from this leap?
But sometimes we have to look back and appreciate the advances we’ve already made. This year has been a great reminder of that, and it’s only September.
At Heyday, we’re building and writing deeply about coaching these days.
This week, we launched the first version of our advanced AI tooling, specifically aimed at the executive coaches of the world. If you have clients you deal with routinely, let’s talk.
What we're reading.
1/ [Long read] This is the most in-depth analysis of OpenAI’s foundations, challenges, progress, and current state that we’ve come across. Well done. Learn more at WIRED >
2/ My favorite design read in recent times. David Huang provides great historical context on the interaction changes that define our current interfaces. What will the massive interaction change from AI be? Who will design it? Learn more at Proof of Concept >
3/ Adept has launched the most powerful permissively-licensed language model with <10 billion parameters. that includes unused embeddings for multimodal extensions. Learn more at Adept >
4/ There’s so much good stuff from MIT’s publishing wing this week, that we’re going to link to a roundup. How to talk to kids about AI, Chinese chatbots and emotional support, and a host of other interesting links. Strap in. Learn more at MIT Technology Review >
5/ Before we get into the nastiness of election season ad campaigns, Google is making the call that political ads must disclose if they use AI. Let’s see if the effort to educate folks on deepfakes will work. Learn more at BBC >
6/ [Research] We’re based out of San Francisco, so the ever-changing weather during the day is something we’re always interested in. WeatherBench2 raises the bar again, and this work shares important historical progress we’ve made. Learn more at Google Research >
7/ We start with OpenAI. We close with OpenAI. Here’s a thorough interview with Jan Leike, head of OpenAI’s alignment research, on how they’re approaching the alignment problem. Learn more at IEEE Spectrum >
Research for this edition of Machine Learnings was enhanced by Heyday, the AI-powered memory assistant.
Thanks for reading Machine Learnings! Subscribe for free to receive new posts and support my work.