Great column from @ShiraOvide, looking at differences between private messengers. Note Telegram isn't listed, because Telegram is not a private messenger.
And note that Signal is the gold standard, providing the foundation for the rest.❤️
https://t.co/nQiAwNmcE3
Fellow writers: There a magical residency called Dear Butte, from which I just returned. You can stay in this lovely little house in Butte, Montana for up to 10 days and work on your writing project, without having to clean or socialize or talk to anyone. Highly recommend. ✨✨✨
Anti-hype LLM Reading List
This is actually a really good list of papers and reading materials on LLMs. Love the curation by @vboykis.
https://t.co/XYQP1FcQnC
Meta confirms once again what I have been saying; LLMs continue to have trouble with factuality.
All those “factuality will be solved in 2025” claims are just fantasies, not projections that directly follow from data.
“Every time I’ve done this, I’ve seen something I didn’t expect to see, learned something I didn’t know,” @BKCHarvard's Dr. Rumman Chowdhury (@ruchowdh) tells @WillOremus of @washingtonpost
Must-read piece on red-teaming generative AI models
https://t.co/6Gu7wNHnPm
7/ The company will switch its legal basis to Consent, meaning that users will have to affirmatively opt into behavioral advertising, not merely *not* opt out. This will apply to all users in the "EU, EEA and Switzerland."
Yesterday, Meta announced that it will change the legal basis through which it processes certain data in the EU under the GDPR from Legitimate Interest to Consent. This is a substantive policy change that will have serious repercussions on digital advertising in the EU. (1/X)
Ezra Klein’s podcast this week is being guest hosted by his producer who sounds *exactly* like him, to an extent that I’m fairly sure there’s a kind of Talented Mr Ripley thing going on. Remains one of the best ongoing bits in media, just bananas. Just listen to him
Mexico’s cartels are taking the “Monster” truck concept to deadly new realms, modifying pickups with gun turrets, steel armoring, battering rams—and much more
w/ @emiliano_mega
https://t.co/w4mZ9iTOhP
Exposing your eyes to sunlight in the morning increases daytime energy & mood etc. & improves nighttime sleep, but it also triggers a cascade of short, medium & long-acting peptides & hormones that powerfully modify state of mind & body in other (positive) ways too. *Blink as needed of course; don’t harm your retina. Science can’t satisfactorily fix broken retinas yet…
And no, there is no true artificial replacement for sunlight that is as good, but if you can’t view sunlight in the morning then bright artificial light would be your next choice. And don’t forget that there is sunlight on cloudy days. It’s especially important to view that sunlight on cloudy days.
And while we're talking SF Chronicle, this series is the most astonishing work I've seen from the paper—or any local outlet—in the 12 years I've been living in SF. Incredible, courageous, essential reporting by @MeganCassidy and @gabriellelurie. https://t.co/zbIdELVYaq
Thinking in pictures, thinking in patterns, thinking in words—these are quite different experiences. But do thinkers themselves fall into such neat categories? https://t.co/7xTbjPMFhU
"There are statements in the document...that building a wall in fact will effectively create new threats by encouraging illicit air traffic & maritime traffic, and also by encouraging the creation of new tunnels & new threats at ports of entry.” #TrumpWall https://t.co/MtMzKEEarQ
"As the boat lay anchored in a nearby cove that evening, the brilliance of the Milky Way above was rivalled only by the glow of the blindingly bright lights at the salmon farm."
Read @jessimckenzi's richly illustrated report from the waters of Patagonia:
https://t.co/BbJgcfVTK8
Christopher Caldwell on Silvio Berlusconi, exclusively for @compactmag subscribers.
Get a full year for just $20 with code CALDWELL.
https://t.co/ToMgjU2Ops
wait, can i just gaslight ChatGPT into thinking it's incorrect and offering me an apology? (the first time was an accident where *i* actually screwed up but convinced it that it was wrong; the second time was on purpose.)