About Thinking AI
One of the most common sentiments I hear on the internet right now (or at least the corner of the internet I hang out on) is some version of “AI is moving too fast to keep up”
I’m writing this on March 29th, 2023 — here’s a quick survey of some (very large!) things that happened in the last week:
Cerebras Systems open-sourced a SOTA distilled LLM (and then Nomic AI did too)
A bunch of smart people called for a 6 month halt on training larger models
BlinkDL_AI trained an RNN competitive with transformer architectures
An AI-generated picture of the pope tricked a bunch of people
Keeping up really does feel like a full-time job. Luckily, it more or less is (part of) my full-time job. I get paid to know what is going on in AI.
The goal of this publication is to make it easier for more people to understand and reason about the latest news in AI.
Goals:
Daily: Publish an update ~4 days a week (M-Th).
Short-form: Updates should be consumable in ~60 seconds
Opinionated: I don’t want to be an aggregator, I want to provide context as to what matters, and why it matters
Fresh: As often as possible, post about recent events
Why?
There is a lot of content out there, so a few thoughts on why I’m doing this at all —
Writing is the best way to clarify thoughts: My first why is selfish — I find that my thoughts are much clearer after I’ve written about them. This gives me a reason to write more often.
I think I’m a good writer? I’ve written a decent amount as a professional and people tend to like what I write.
Spreading AI information is important: I believe AI is going to fundamentally change society, and that it is important that as many people as possible can weigh in on that fact.
So with that, welcome on this journey with me! Expect the first update to be tomorrow — March 30th, 2023.
