Austin King retweeted
To be the guy at OpenAI whose job is to read prompts of prominent mathematicians and separate the "how can I be happy" prompts from the "I think I solved Navier Stokes, prove it with lean" prompts
Austin King retweeted
MAC Commissioner Jon Steinbrecher on Western Michigan-Michigan controversy in a letter to NCAA football oversight: “Accordingly, I respectfully request that the NCAA Division I Football Bowl Subdivision
Oversight Committee recognize Western Michigan University as the actual winner of this
contest, 12–7, and take such action as is within its authority to reflect that determination in
NCAA records, rankings, postseason consideration, and any other appropriate NCAA purposes.”
Austin King retweeted
The general public asking the Big Ten where its ~magic~ replay clock is that gave Michigan an extra second.
Austin King retweeted
A defining image of college football’s descent into illegitimacy: U of M fans storming the field after beating Western (four-touchdown underdogs) on a hail mary made possible by a phantom booth review placing one second on a game clock that had clearly, conclusively expired.
Austin King retweeted
We are using AI Safety to cover up what is actually just an old-fashioned organizational failure and people are eating it up like a dog eating its own vomit
Austin King retweeted
Imagine an airplane crashes under mysterious circumstances. Except in this world, there is no government oversight, and all investigations are done voluntarily by the airlines themselves. Nonetheless, the airline wants to reassure their customers, so they do an investigation.
To investigate, the airline invites three of the world's most respected airplane researchers to look into it. Except by the time the researchers get there, the airplane has already been disassembled and melted down into little cubes. The researchers are instead given the logs of all the airplane directions and the transcripts of the conversations between the pilot and the copilot.
Except one tenth of the log has been deleted. And the airline also tells the investigators that they can only look at the time between when the airplane first started losing altitude and when the airplane made contact with the ground and that all investigation of the time before the airplane started losing altitude is off limits. Also there are a lot of rumors in the airplane community of a few other airplane crashes under somewhat similar circumstances, but the airline says that these are also explicitly off limits.
When the researchers arrive to review the logs, they find that the logs are 10,000 pages. But they only get six days to read them, and half the logs don't arrive until day 4.
Now you understand the independent METR investigation.
Austin King retweeted
My coworkers mog the fuck out of your coworkers in every possible dimension and basically you’re a fucking loser. We are smarter than you, we work harder than you, we’re even better looking. One of my coworkers climbed Everest, another is a chess grandmaster, and everyone won some precocious kid award at some point that we still talk about to this day. The talent density is that of a neutron star. Which is like really dense. I’m telling you this because I want you to evaluate me based on the perceived prestige of my coworkers. Anyways fuck you or something I don’t know
Austin King retweeted
Replying to @bradrcarson
I'll start by fetching the original post so the reply matches the actual discussion.I'll count the draft reply so it stays inside the 45–80 character limit.most dealers just assay the bars then wire the cash
Austin King retweeted
Here's why this is a problem:
When words, verbiage, style, tone and voice all become fungible because of AI, then all that'll matter in future is the name.
This will further intensify social hierarchies.
If all slop is equal, then only slop by the famous will get published.
Austin King retweeted
Sometimes you don’t even know when you’ve already watched the end.
You had to be there. And now it’s over.
Austin King retweeted
Pangram is good, but has some well-known failure modes, like:
- A time you heard it didn't work
- A school essay you wrote years ago that it flagged as AI (you don't have a link or the essay text handy)
- A different AI detector got it wrong, and they're all the same, right?
Austin King retweeted
Just gut wrenching to watch. Same thing happened with inaturalist. You have a community that draws people together with ideas and values and then becomes defined by adherence to shallow signifiers of membership at odds with those values. Gross
Jeremiah Johnson 🌐@JeremiahDJohnsJul 31United States
United StatesConnected via United States Android AppAccount-level information, not a live location or per-post device.
I'm fascinated by the apology Hank Green has to give his fans for daring to use ChatGPT.
To be clear, he did not use AI generated images or use AI to write his script. He used it as a research tool to find scientific papers, which he then read and used as material.
But there's this weird thing where some creators can happily use AI and talk about AI and their fans seem totally fine with it, and other creators would be ritualistically flayed alive by their own communities if it ever came out they had once glanced at an AI output. And it seems really arbitrary which camp is which, like there's no rhyme or reason to whether which creatives are allowed to admit that AI exists and which have to join the Butlerian Jihad.
Austin King retweeted
New research: We studied every NBA Achilles tear from 2014-2025 (19 players) to find out what happens in the games right before the injury.
The finding: minutes played spiked significantly in the 10-15 games before rupture — even though season-long workload looked totally normal.
7 Achilles tears in 2024-25 alone.
3 of those tears occurred in the playoffs.
This is a real, growing problem in the league.
Full paper: journals.sagepub.com/doi/10.…
Austin King retweeted
This is our most ambitious announcement yet. Two new models: Pangram 4 and Pangram Image.
Pangram 4 completely reimagines how we approach the problem of mixed authorship, by adding a tokenwise head onto the classifier to give every token a prediction given the full document context. We also wrote a 38-page technical report detailing our experiments, methodology, and evals.
Pangram Image is a completely new modality, bringing our detection expertise to AI-generated image and videos. In my early testing, it has worked shockingly well, even in strange cases like real photos of AI-generated bodega menus.
Today marks a huge step in the frontier of AI detection technology. I'm so excited to finally share with you all!



























