Tomorrow is Thanksgiving, and this year we’re grateful for a lot of things, including the community that reads the newsletter and sends us emails. However, I’ve been told to keep it terse since people are busy cooking stuffing.

Here are five good AI-related links:

“The Way We Measure Progress in AI is Terrible,” by Scott J. Mulligan in the MIT Technology Review. Requires a subscription. But I agree. It is terrible!

“A Revolution in How Robots Learn,” by James Somers in The New Yorker. I expect a lot of this next year as people figure out how to apply learning models to robots. Only good things will result.

“Transparent’s new speaker is an epic audio homage to brutalist architecture,” by Jonathan Bell in Wallpaper*. Does this relate to AI? Only in the sense that if a robot decided to make a nice pair of speakers, it would probably look like this.

“AI squirrel spotter deployed to protect endangered red squirrels,” by Mickey Carroll in Sky News. The headline underplays the Beatrix Potter-meets-Skynet aspects of the story, which explain how squirrels can be individually identified by their unique whisker patterns.

“Getting started with AI: Good enough prompting,” by Ehtan Mollick in One Useful Thing. Mollick asks us to update our thinking and consider AI less as a cheerful, weed-addled intern and more as a very forgetful coworker.

Happy Thanksgiving! See you next week!