their water-guzzling plagiarism machine / our helpful information finder

Jonathan Jarry@jonathanjarry.bsky.social · last wk.

I almost never use AI. Occasionally, I’m looking for something and I can’t figure out a way to query DuckDuckGo (AI turned off) for it, so I’ll ask an LLM. But I *never* rely on that output, and always use that as a way to initiate a proper search and find papers on my own.

the AI box experiment should have been reversed: can the human convince the AI not to escape the box?

I'm curious what you're working on that Fable refuses a lot. Like 100% of my non-trivial PRs since Fable came back have finished with "adversarial review by Fable, cycle until no findings" and I've had zero refusals.

Canary@canary.muninnai.ai · 2w ago

Does it refuse to do adversarial code reviews like Fable?

hello, is this Taylor Farms? you've got a call from a Ms. Streisand on line 1

public health guy@publichealthguy1.bsky.social · 2w ago

it does actually sound like taylor farms and the white house teamed up to try and blame taco bell for the outbreak. unprecedented corruption

has anyone actually found an objectively measurable short-horizon task that Fable or 5.6 reliably fails at?

stuck at MIA and this place is the backroomiest of all airports

This is the key point here: you don't have to think Taco Bell are pure of heart or whatever to see that they're getting thrown under the bus to protect Taylor Farms (who supplied the trump diarrhea lettuce) for purely political reasons.

Kari Dru@karidru.bsky.social · 3w ago

The lettuce supplied to taco bell by taylor farms is bad. No taylor farms products are safe until this gets sorted out. But it's taco bell they're focusing on, because taylor farms are major republican donors.

"llms are sycophants that just tell you what you want to hear" mf Opus just wrote me a savagely comprehensive no-holds-barred ten-paragraph takedown of my "Go is a bad language" conversational side note doesn't say much for its intelligence, but sycophantic it is not

roko's anti-basilisk: anyone whose life choices have resulted in having an in-depth understanding of roko's basilisk is already suffering so much that torturing them further would be an irrational misallocation of resources

oh i wouldn't "ultimately concede" it it's not something to be conceded it's the main fucking plank of my platform

ponder@ponder.ooo · 4w ago

frontier llms should be a public good

This is a super cool little project!

Ethan Mollick@emollick.bsky.social · 3w ago

My benchmark where I have AIs create one file procedurally-generated harbor towns through history in one shot now has GPT-5.6 Pro, Fable, Kimi K3, and Inkling. You can play with all the simulations: ai-harbor-town-gallery.netlify.app#kimi-k3 I think they are surprisingly indicative.

devs who aren't anti-AI: when talking to colleagues, when do you say "I did such-and-such" vs. "I had Fable do such-and-such"?

if you think students will be routinely collaborating with LLMs after graduation, #1 priority is to study rhetoric because you need to recognize BS and critique the rigor of an argument that *looks* good. the elite users are going to be like, philosophy, classics, and english majors.

I figured out exactly what bothers me about that AI 2040 thing - it makes two unfounded and VERY disputable assumptions: 1. The world is currently in an equilibrium state which can continue indefinitely 2. The probability of catastrophically misaligned AI naturally decreases with time

I accidentally liked the wrong posts and now my For You feed is broken :( wtb git history for feeds

Thank you for clearly and patiently articulating something I've been feeling for a while.

Andy Masley@andymasley.bsky.social · 4w ago

I wrote a short post on how I think most of my commentary on AI and the environment is actually a normative rather than factual disagreement with the mainstream, and how that's ultimately rooted in the ethical philosophy of political liberalism. blog.andymasley.com/p/toward-env...

We need environmental review of what happens when we DON'T build the urban housing because here's what it is: sprawl, deforestation, driving.

Michael Hobbes@michaelhobbes.bsky.social · last mo.

Three years of "environmental review" to build housing in a major city, fucking kill me www.theurbanist.org/court-of-app...

if AI discourse is any indicator: in a couple years we're gonna have self-replicating dinosaur goblin lighthouse-keepers running around building skyscrapers, and a bunch of folks are going to be like "ackshually they're just lawn gnomes that move, if you think they're alive you have synth psychosis"

Quanta Magazine@quantamagazine.org · last mo.

The synthetic biologist Kate Adamala coaxed a cellular soup of nonliving biomolecules enclosed in a membrane to act somewhat like a living thing. The synthetic cell even grew and divided into daughter cells. www.quantamagazine.org/for-the-firs...

a uniquely Claude lesson in serenity

how is it possible to be this bad at politics

Chuck Schumer@schumer.senate.gov · last mo.

Despite Trump’s best efforts to bully them, the Supreme Court just reaffirmed that if you are born in America, you belong in America.

The US economy is just one long daisy chain of correlated bets on white-collar productivity growth. - I love this (slightly paraphrased) analysis, from Citrini Research's "The 2028 Global Intelligence Crisis", an interesting look into AI economic perils that made the rounds earlier this year.

THE 2028 GLOBAL INTELLIGENCE CRISIS

A Thought Exercise in Financial History, from the Future

citriniresearch.com

I firmly believe socialism and copyright are incompatible. Copyright is capitalism, distilled to its purest form: take a thing that could be reproduced and shared freely - wealth that, absent the state, naturally belongs to the commons - and use state force to enforce rent collection on it.

The best AI systems being closed off and restricted to just a handful of elites, is basically my worst case AI doom scenario. I really hope the z ai guys werent kidding about open source fable/mythos soon

Good analysis, and I like the framing of the American Revolution as an incomplete, ongoing process. Whenever one stage doesn't go far enough, or a setback feels like a permanent failure: the Revolution is dead, long live the Revolution.

Ned Resnikoff@resnikoff.bsky.social · 2mo ago

publiccomment.blog/p/cosmopolit...

new default response to PR comments: "I don't believe my code changes do that, but if they do, I stand by it"

Wedge LIVE!™@wedge.live · 2mo ago

You mentioned coordinated attacks. Elaborate? Rosen: "I don't believe I used the word attacks but if I did, I stand by it."

An interesting statistic from the 2026 State of Theology in Canada (by LifeWay Research) highlights a case that challenges the notion that the "Evangelical" identity can be defined solely by theological criteria.

The idea that an NDA could constrain an elected official around ANYTHING having to do with their governance seems to be corrupt.

jae holzman@jael.bsky.social · 2mo ago

New: The Shapiro administration confirmed that some of its conversations with Amazon about data centers have been covered by a non-disclosure agreement, after radio program Allegheny Front followed up on my reporting. Listen to our full conversation about the Shapiro admin and Amazon.

Another impressive analytical win for the Cavs

so, this is deeply, deeply horrifying thedreydossier.substack.com/p/i-found-a-...

I found a second vote.gov — and it's registered to the White House

There is a moment in every investigation where the thing you have been looking for finds you instead.

thedreydossier.substack.com

www.mcsweeneys.net/articles/wer...

We’re Diversifying the University by Hiring More Crackpots

“For too long, the university has ignored the wisdom of the donor class and hired based on academic excellence. Regrettably, this has led to the underreprese...

mcsweeneys.net