Links for June 9, 2023
No book this week—I've miles to go with Anna Karenina, and I spent Thursday & Friday at The Advances in Social Genomics Conference. My dissertation (the proposal for which I'm currently writing) is aimed at studying how data-sharing works in social genomics, and how the human aspects of it—institutions, politics, economics, ethics—shape the field. So more thoughts on all that to come!
In "this is fine 🔥☕🐶🔥" news:
"Your Phone Wasn’t Built for the Apocalypse": Clear, chilling writing on the expectations embedded in technology—if I were teaching an intro class on Science and Technology Studies, this would be on the syllabus. "Color is always constructed in a picture, never simply reproduced." Ian Bogst / The Atlantic
You may know ethanol levels in gasoline are capped in the summer to control ozone (which is especially important when we're also, y'know, breathing in wildfire smoke). You may not know that ethanol biofuels… are actually just about as greenhouse-gas intensive as fossil fuels, and are a phenomenally stupid use of land, too. Michael Grunwald / NY Times
Marina's writing about machine learning, so here's a lot on that
By the way, can we stop calling it AI? The term's even less useful now than it was five years ago, when I first heard jokes about how "AI" was just what you called machine learning or neural networks when you wanted someone to give you millions of dollars.
“Microsoft just gave itself a full-screen ad in search results by faking an AI interaction” When you search for "Chrome" in new Bing, an ad pops up pretending to be an unbiased machine. Instead of what you're looking for, it promotes… Bing itself. Sean Hollister / The Verge
An AI reporter reviews the 2023 edition of the World Book encyclopedia Benj Edwards / Ars Technica:
"As I carefully pulled them out of the box one by one, I enjoyed feeling the weight of the information in my hands. It felt like stepping back onto dry land after a long boat ride. It's hard to put a name on that emotion."
Ban "counterfeit people", writes philospher Daniel Dennett / The Atlantic. More specifically, ban LLMs from using first-person pronouns, blogs Kevin Munger / Crooked Timber. More academically, here's a paper on all the other ways programmers intentionally and unintentionally anthropomorphize AI, and why they shouldn't Abercrombie & colleagues / arXiv .
Hayao Miyazaki, the founder of Studio Ghibli, upon seeing an artificial intelligence presentation: "I strongly feel that this is an insult to life itself" via Twitter
- I love the surprising way this snail crosses this chasm.
This is Some Preliminary Thoughts, Bennett McIntosh's blog. You can sign up for email updates here, or unsubscribe here.