shambibble
banner
shambibble.com
shambibble
@shambibble.com
"i'm the same boy i used to be"
Pinned
you've heard of the bell curve, now get ready for the HELL curve
December 18, 2025 at 2:17 AM
my only comment on lariangate is that i'm very skeptical of the idea that AI is being used for "mood board" stuff. it sounds like a fib to me because this is something AI is bad at! too compositionally samey, awful at generating out-of-distribution stuff that any halfway fantastical setting needs
December 18, 2025 at 1:22 AM
SWEET JESUS
December 17, 2025 at 8:25 PM
Reposted by shambibble
going by the poll that they had there are like 30 people in this discord. this is one guy getting in an argument with his gaming friends.
A Discord for gay gamers is in disarray after an exec at Anthropic forced an AI chatbot on the Discord.

When confronted, the exec said he would not submit to mob rule and that AIs have emotions. @evystadium.bsky.social has more.

Story by @mjgault.bsky.social
www.404media.co/anthropic-ex...
December 16, 2025 at 9:26 PM
Reposted by shambibble
19.5k follower account (who ostensibly works in the healthcare industry?) tweeting out straight misinformation about a bill for bsky clout. pathetic bsky.app/profile/thra...
December 15, 2025 at 3:54 AM
Reposted by shambibble
The endgame here is getting the Supreme Court to declare that the entire civil service system is unconstitutional. Which is to say, this administration does not want us to be a modern, efficient, globally competitive nation-state.
Supreme Court Poised to Vastly Expand Presidential Power, Again
The 90-year-old precedent protecting independent agencies is likely on its way out.
lnk.thebulwark.com
December 14, 2025 at 11:00 PM
first dem primary crosstab i've seen in forever that's actually accurate to me
Texas - Democratic Senate Polling Leads:

All: 🔵 Crockett +8

Men: 🟢 Talarico +10
Women: 🔵 Crockett: +21

18-34 Year Olds: 🟢 Talarico +29
35-44 Year Olds: 🟢 Talarico +9
55+ Year Olds: 🔵 Crockett +25

Texas Southern U. / Dec 11, 2025
December 13, 2025 at 9:52 PM
Reposted by shambibble
A left winger can get published in most mainstream media outlets as long as they come to the exact same anti-lib conclusions as most of the management and columnists
December 13, 2025 at 3:35 PM
for some reason i thought nano banana pro and gemini 3 were better-integrated than this. now i have a whole new reason to feel bad for the little guy. he needs to read my old prompt phrasing guide!
December 12, 2025 at 10:05 PM
Reposted by shambibble
as a savvy, street-smart gambling genius I love to place prop bets on things that have an open-ended set of outcomes and are obviously vulnerable to semantic nitpicking 😎😎😎
People who bet on this on Kalshi are not pleased. Lol
December 11, 2025 at 4:42 PM
yet another particle of ambient blueanon leftslop traces back to meidas news
Okay this is going to be long so mute it if you don't want a granular explanation but first: SHAME on @ronfilipkowski.bsky.social for misrepresenting what Chair Powell said in order to pat himself on the back. Shame on you.

Let's talk about measuring the 171mm people of the US labor force.
@peark.es seems real bad that the fed chair doesn't trust the job numbers
December 11, 2025 at 3:28 PM
Reposted by shambibble
For people who trusted copyright laws to rein in AI, this should be the last scene in Animal Farm. “They looked from IP to Tech, and from Tech to IP, but already it was impossible to say which was which.”
The Walt Disney Company and OpenAI reach landmark agreement to bring beloved characters from across Disney’s brands to Sora
Agreement marks a significant step in setting meaningful standards for responsible AI in entertainment.
openai.com
December 11, 2025 at 2:25 PM
what the fuck did suckerberg do in 2025? he released the most embarrassing flop model in years (llama 4), competely failed on image synth to the point they licensed midjourney slop, and paid more buyout money than a D1 football team to burn down his entire AI division
December 11, 2025 at 2:32 PM
i get so sad every time i read wikipedia entries for megafauna. we're living in a post-apocalyptic world and don't realize it because the apocalypse happened several tens of thousands of years ago
December 10, 2025 at 2:00 AM
Business Professor and Cato Institute adjunct scholar here: if you have opinions about "median household income", have you considered how much more flattering "mean college graduate income x2" is to my priors?
December 9, 2025 at 5:55 PM
Collin Allred? MJ Hegar? Lupe Valdez? Wendy Davis? This reflexive "uh bluh white guys" schtick would cut zero mustard even if it were true, and it's very much not!
December 9, 2025 at 5:16 PM
you, superficial and bluepilled, leaving your chin out for k-hivers to punch: "oh no, jas crockett is the texan stacey abrams!"
me, sophisticated texas ball-knower with a memory going past 10 years: "oh no, jas crockett is the black wendy davis!"
December 8, 2025 at 2:44 PM
this is a great piece and i am in agreement with most of it. he manages to be fair to both text and image synthesis while accurately identifying the problem (the greedy bosses)
I thoroughly recommend reading all of Cory Doctorow's recent speech on AI skepticism, it's crammed with new arguments and interesting new ways of thinking about these problems pluralistic.net/2025/12/05/p...
Pluralistic: The Reverse-Centaur’s Guide to Criticizing AI (05 Dec 2025) – Pluralistic: Daily links from Cory Doctorow
pluralistic.net
December 8, 2025 at 1:02 AM
Reposted by shambibble
Arguing about copyright with people online
December 7, 2025 at 9:33 PM
Reposted by shambibble
humanity exists inside inefficiency. every time we optimize something to eliminate all areas of conflict and drag, we lose what made it cool and interesting
i think a lot of people think of "slop" as following from the next-token prediction nature of how generative AI works, but i sort of think that the "slop"-ish qualities of genAI output can be chalked up way more to RLHF
December 6, 2025 at 5:25 PM
bayesianism is when you silo every single claim made as part of an argument individually and evaluate them without reference to each other and the stricter you separate them the more bayesian it is
December 6, 2025 at 7:38 PM
true and i would go so far as to say it is much easier to make a pure completion LLM (a base model with no "tuning") produce creative output than any of the flagship chatbots. i've seen very little legitimately funny LLM output but 100% of what i have has been from completions and not conversations
i think a lot of people think of "slop" as following from the next-token prediction nature of how generative AI works, but i sort of think that the "slop"-ish qualities of genAI output can be chalked up way more to RLHF
December 6, 2025 at 5:14 PM
Reposted by shambibble
New Warhammer drip just dropped. When your armor saves you and smells like plumeria. #UHawaii
December 5, 2025 at 5:31 AM
IMO the term "disability" just stops being useful once it includes the ADD/anxiety/depression cluster of pure cognitive disorders. (i say this as someone with all three, but diagnosed long after my standardized test days)
Balancing "no person with a disability should be denied accommodations" with "it's important to discourage non-disabled people from pretending to be disabled to take advantage of these accommodations" is really difficult and I don't have a great answer for what anyone should do about it.
December 3, 2025 at 5:04 PM