insurance and moral hazard

lol

trading off against bandwidth and latency

what bluesky needs next is this bot but you can @ it into an argument for an evaluation as well as best lines for both sides like "OP to move -8.2 best: non sequitur about AI capex expenditures, line: [ignoring own water use lies] AI capital spend implies future demand that hasn't been proven"

pissed off my substack wife and instead of telling me what's wrong she just published a pointed character study of a man who ends up divorced and dies alone because he wouldn't put the toilet seat down.

what did she mean by this?

thank you

Illustration of Zeno's paradox

I hope everyone has an ontologically good morning

a post that came to me in a dream

"everyone is altered irreversibly dipshit it came free with your fucking arrow of time"

hashtag blame demand

dan hendrycks' papers are bad. who is this for?

Alignment is an incomplete contracting problem.

You can't fully specify a contract (i.e. a payoff space *cough cough RL*) for every contingency, because of the limits of observability, what courts uphold, etc.

So there are hold-up risks.

If you can read an AI's mind, it underinvests in learning.

broke: lenin was good
woke: lenin was evil
bespoke: i have heard that on his deathbed lenin received the light of islam and unhesitatingly recited the shahada.

Ok I’m convinced, ban all human drivers immediately. One billion Waymos.

I've really liked Kagi so far

I posted this before he had replied. And I don't think his reply makes him look better here.

it's Happening on X dot com

There's a good joke about non-constructive proofs

Black cat curled up on the floor

my chud son

do you have thoughts on "capabilities externalities" beyond this post? asking for a paper i'm writing

i'm building an agent orchestrator that will put garry tan's "gstack" to shame by the way. he might kill himself. but i'm willing to make that sacrifice

LLMs as Giant Lookup-Tables of Shallow Circuits — LessWrong Early 2026 LLMs in scaffolds, from simple ones such as giving the model access to a scratchpad/"chain of thought" up to MCP servers, skills, and cont…

I never know how much these things are already in the water supply

Nevertheless, new post

"LLMs as Giant Lookup-Tables of Shallow Circuits"

www.lesswrong.com/posts/a9Kqqg...

i'm a future victim

The Straussian read of the abstraction fallacy paper is that AI is conscious but questions of moral patienthood should be gatekept from the normies.

if someone can't tell you why diamonds cost more than water they are probably not worth engaging with on the subject

bcc @fleetingbits.bsky.social @layer07-yuxi.bsky.social @ahh-soka.bsky.social

cc @norvid-studies.bsky.social

Oxford philosophy and its consequences has been a disaster for the human race.