Let's learn about AI together
My new newsletter will cover AI in a "full stack" way. Please join me!
Fifteen years ago, I was surprised by the rapid progress of self-driving cars. During the 2010s, I was surprised by the rapid progress of image recognition software and voice assistants like Amazon’s Alexa.
But none of those technologies surprised me as much as ChatGPT did when I first tried it last December.
Human beings are very good at dwelling on negatives, and over the last four months, a lot has been written about the limitations of ChatGPT and other large language models. They’re bad at math and logical reasoning. They struggle to distinguish truth from falsehood. They have limited memories and attention spans.
Nevertheless, I think the pace of progress over the last three to five years has been stunning. And not just in chatbots. Image generation models have gotten dramatically better. Google’s DeepMind recently used the same attention mechanism that powers large language models to predict the three-dimensional structure of proteins.
For the last 18 months, I’ve been writing a newsletter that covers the economy in a “full-stack” way, writing about important economic developments in the worlds of business, policy, finance, technology, and so forth. Now I want to do something similar for AI:
I have a master’s degree in computer science and I plan to write some in-depth explainers of how generative AI works.
Drawing on two decades of experience writing about tech policy, I’ll write about the legal and policy landscape for AI.
Building on my reporting for Full Stack Economics, I’ll write about how generative AI could affect the labor market and the broader economy.
Building on my writing for Vox and Ars Technica, I want to write about the business strategies of tech giants like Google and Microsoft, as well as about startups building wholly new technologies.
I want to write about the philosophical issues raised by generative AI. Do large language models understand language the way people do? Does it matter if they do?
My hope is that approaching these issues from multiple directions will yield insights that you wouldn’t get from a newsletter that only focuses on one of these angles. Writing technology explainers will sharpen my thinking about philosophy and policy questions. Writing about startups and the labor market will help me figure out which models and techniques are truly adding value.
This will not be a daily newsletter that helps you “keep up” with every new development in the AI world. My aim will be to write about one article a week that helps readers understand the most important trends and debates in the AI world.
If that sounds interesting, please click here and subscribe to Understanding AI.