Devin

A lot of people don't know this but the straight of Hormuz closing is actually a carefully orchestrated conspiracy. If a tanker blows up and spills oil everywhere, the Dawn dish soap marketing team already has supply caches in place to get footage of birds being cleaned up asap

in roughly the same way that I don’t maximize the chance of thunderstorms by messing with my barometer

it’s true that the one-boxer make a choice such that the expected payout for making that choice is maximal, but they didn’t thereby maximize their payout *by* making that choice

so my point is that I think it’s actually misleading to say that the one-boxer maximizes their expected payout, bc that implies they have options they don’t have

imo it's very important not to use the normal verbs that connect speech events to intentions, promises, etc., for Trump. he simply does not "claim", "commit", "deny", "retract", etc. anything. he produces utterances!

I’ve never thought seriously about Newcomb’s problem, but I quite like the beginning of this Lewis paper (and the pivotal gibbard quote). Quite generally, if you set things up to richly reward irrationality, then irrationality will be richly rewarded

we need stancil to weigh in

me explaining how my dissertation is progressing

I don’t think it does! it’s true that people with a disposition to one-box make more money, but you can’t make more money *by* one-boxing

everyone complains about jargon making science inaccessible to the public but the moment someone tries to fix the problem y’all make fun of them

I'm just gonna say it: Booker's proposal to end income taxes below $75k is idiotic. We've just crippled our state capacity and your response is a tax cut? Come on man

I love Econ-positng. Dan Habsmith of Guttermeyer Associates posts the most confusing chart you've ever seen and says "Quite surprising VARG numbers, SPLG index showing movement" and you get the context by one of his mutuals quoting it with "Buy a fucking gun and head for the hills. Right now."

One thing I never understood about Les Miserables is how Javert managed to convince the higher ups at the police department to commit nearly unlimited time and money to the Bread Crimes Unit.

well this is one way to increase demand for renewables & EVs

Her trip to Ireland was ruined by this one question?? Gee, I wonder why…

I know Jeremy Waldron has some stuff in support of hate speech restrictions

too many people with no background in either philosophy or cognitive neuroscience have strongly-held beliefs about consciousness

motivated by offline events but maybe relevant for some: never stops boggling my mind how some are able to rationalize the toxic ways they navigate internal org disagreements - functionally punishing the urge some people have to step up to roles of responsibility in these orgs

so the fact that an LLM is producing verbal outputs that look like the verbal outputs of a human who’s subjective experience is *typical* indicates that its verbal outputs aren’t reflective of its subjective experience, and so can’t be evidence of such

if a human was having an analogous subjective experience we would expect very different verbal behavior

I don’t think you *do* need that deep philosophical work. we know LLMs can’t be conscious in between responding to prompts, because there’s no internal activity during that time. that would be a very different subjective experience

or to put it another way, by “continuous” I don’t mean anything deep, I just mean “not blacking out whenever I’m not actively talking”

I don’t think we need the stream of experience to by truly/metaphysically continuous in humans. the point is that LLMs can’t have the sort of stream of experience humans do, yet they produce outputs as if they did, so those outputs can’t be treated as evidence that they have a stream of experience

if an LLM could talk, we could not understand it

@oldjerryfodor.bsky.social I feel like you’re more up on the relevant lit than I am

this could be an actual paper couldn’t it

(when I say that it isn’t doing anything, I mean that there are no internal processes running)

for example, the LLM (speaking loosely) can’t be conscious when it isn’t actively responding to a prompt, for the simple fact that isn’t *doing* anything. yet it produces the verbal outputs you’d expect from a human with a continuous stream of experience

the evidence people usually point to for LLMs being conscious is that they behave in ways indicative of consciousness in humans. but if LLMs *were* conscious it would be extremely different from human consciousness

vids like this are often shared with “nothing will be prepared you for…” but in this case it was exactly what I expected