Aaron Bergman's tweets, May 2023
Every public tweet, reply, quote tweet and retweet that Aaron Bergman (@AaronBergman18 on X, formerly Twitter) posted in May 2023 and has not deleted: 493 posts, oldest first, times in UTC. Each heading links to the original post on X. When Aaron replies to or quotes someone, their tweet appears as a blockquote with their name; those words are theirs, and everything outside the blockquotes is Aaron's.
This page is one month of the complete archive of Aaron Bergman's tweets, which updates automatically every few hours. Previous month: April 2023. Next month: June 2023.
Monday, May 1, 2023
Reply to @touchmoonflower, 05:37 UTC · 2 likes
༺ 𐂂 ༻ (@touchmoonflower): ok so maybe merkle trees aren’t your thing but get this: every new hire at that other site is given a copy of There Is No Antimemetics Division to read
From Amazon
Reply to @kartographien, 16:21 UTC · 2 likes
Kart ographien (@kartographien): The Waluigi Effect isnt relevant imo bc the explanation is that (on the internet corpus) rules are collocated with examples of those rules being broken.
A better example would be this:
There's also this explanation though, which seems more general (i.e. beyond personality /valence wrt language)
So "accomplish X" is (all else equal) *adjacent* to "accomplish -X" in neural network space
Reply to @zachsilberberg, 16:37 UTC · 11 likes
(The tweet being replied to by @zachsilberberg is no longer available.)
Exact same thing yesterday but for a (parked) firetruck
I bet firemen have a code word for this
Reply to @Laura_k_Duffy, 16:46 UTC · 3 likes
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): Hey everyone! I think that more EAs should take Hayek's skepticism of planning really seriously, which is why I heavily encourage reading this short paper, "The Use of Knowledge in Society". (I myself need to re-read it; been a few years)
If you do, I'd love to discuss it!(quoting @Laura_k_Duffy: @mcinthedc I’ve said it before and I’ll say it again, “The Use of Knowledge in Society” is a must-read and has irrevocably transformed my thinking about best approaches to policy and EA. https://www.econlib.org/library/Essays/hykKnw.html)
I like this class of arguments but don’t immediately see what the implications are for EA as it stands today
Like they already seems quite baked in so to speak
Quote tweet, 18:37 UTC · 8 likes
Not endorsing the whole thread but this is true and the reason “IQ 1000 AGI” isn’t meaningful in the way “1000 foot tall robot” is
Maybe there are good absolute measures of general intelligence, but IQ isn’t one
Quoting Kareem Carr, Ph.D. (@kareem_carr): Raw IQ scores are manipulated to resemble a bell curve via a process similar to a professor curving test results to get a certain number of students with A's, B's and so on.
IQ test scores are rescaled so that a certain percentage of test results fall within specific ranges.
Tweet, 22:58 UTC · 7 likes
I will stare darkness straight in the face
Reply to @Laura_k_Duffy, 23:04 UTC · 3 likes
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): The sadness evoked in me by this image is a perfect argument against any lingering person affecting intuitions
Reply to @absurdlymax, 23:17 UTC · 2 likes
Max Alexander (@absurdlymax): They got me :(
tbc my endorsed position is that I’m being a weak willed idiot about this lol
Reply to @philoshua, 23:27 UTC · 3 likes
Joshua Blanchard (@philoshua): Question ONLY for people who find the following for individuals very annoying. Suppose you had to have dinner with one of them. Who would you pick? Explain why in the replies.
Eric adds value to the discourse on net, first three do the reverse
Reply to @ohwizenedtortle, 23:41 UTC · 7 likes
oh dandy lion🔸 (@ohwizenedtortle): Did we ever figure out why girls love Taylor Swift? I've been really getting into folklore and it's got a vibe that I don't know how to articulate but I really vibe with.
(quoting @AaronBergman18: I am calling for a total and complete shutdown of Taylor Swift, Hinge, and—just to be safe—heterosexuality until we can figure out what the hell is going on)
this is only like 1/3 or 1/2 of the new ones
Reply to @PradyuPrasad, 23:42 UTC · 16 likes
(The tweet being replied to by @PradyuPrasad is no longer available.)
I can teach him R, just put me in touch
Tuesday, May 2, 2023
Reply to @Laura_k_Duffy, 00:12 UTC · 1 like
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): How so? I think at least the longtermism side of EA often implicitly relies on assumptions about our ability as individuals/groups to reliably model and affect the future of civilization. This seems like something that one should be more skeptical of from a Hayekian view
For sure but it just seems like there's been a lot of discussion about this very point among longtermists; like it's already accounted for in ~the discourse~
People disagree ofc about how much of a problem it is - like Karnofsky at least claims to be pretty skeptical...
Thread, continued, 00:13 UTC · 1 like
but still concludes that for a host of reasons AI is the rare thing that we actually sorta can model and influence better than chance
from https://www.dwarkeshpatel.com/p/holden-karnofsky#details
Reply to @Laura_k_Duffy, 00:27 UTC · 1 like
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): Yeah, I see this. But I think the crux of my disagreement here is that we really don’t have the ability to make *even those* kinds of claims about AI and its tractability with enough confidence that we should actually spend a large amount of money on it.
But we do with respect to other cause areas and their actual impact on the world (which includes but is more than just the metric we're choosing to care about e.g. malaria deaths)?
In some ways it seems way *easier* to model AI and how we might affect its trajectory, at least...
Reply to @Laura_k_Duffy, 00:39 UTC · 1 like
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): @AaronBergman18 That’s fair, I’m unusually anti-theory likely!
(gif, plays on X)
Thread, continued, 00:42 UTC · 2 likes
checked with gpt to make sure I'm using this right and it made sure to remind me that "using the phrase as a threat or engaging in physical violence is not acceptable behavior and can have serious consequences" indicating a willingness to speculate about future consequences
Reply to @SpacedOutMatt, 00:43 UTC · 2 likes
Matt 🔸 (@SpacedOutMatt): Not me though, I’m totally normal (nervously closes tab open to the NOAA Fisheries One Stop Shop Data Portal)
poast link
Reply to @Laura_k_Duffy, 00:49 UTC · 2 likes
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): What lol 😂
between this and "negative marginal birds" think I need to upskill at humor
Reply to @SpacedOutMatt, 01:04 UTC · 1 like
Matt 🔸 (@SpacedOutMatt): https://www.fisheries.noaa.gov/foss/f?p=215:200:15583745277884
care to share what you're working on/trying to figure out?
Reply to @absurdlymax, 02:29 UTC · 4 likes
Max Alexander (@absurdlymax): @AaronBergman18 @ohwizenedtortle @taylorswift13 Thank you for tagging taylor
Wouldn’t be polite to go behind her back
Reply to @SpacedOutMatt, 03:03 UTC · 2 likes
Matt 🔸 (@SpacedOutMatt): It's one of like 5 different projects I want to do with the RP moral weights
Noice, maybe one to add: figure out how to most usefully operationalize turning moral weight point estimates into a prediction market. Like do we care about only adults of a certain species? Going with the RP symmetry assumptions? Relative or absolute? Asking about capacity for
Thread, continued, 03:03 UTC · 2 likes
welfare or actual avg welfare irl? If avg, what does that even mean?
Thread, continued, 03:07 UTC · 2 likes
also:
- per individual*time period or just individual?
- keep the question theoretical or operationalize like e.g., "what will I think in 5 years?"
Reply to @RafaRuizdeLira, 14:23 UTC · 6 likes
Rafael Ruiz ⏸️🔸 (@RafaRuizdeLira): Wtf why is nature like this 😭
(quoting @Rainmaker1973: 50% of Panda births result in twins, but Panda moms almost always abandon one. Therefore, zookeepers have to switch the cubs every few hours to trick the mom in to caring for both Yet, with this trick, it's true love [📹 iPanda: https://buff.ly/3VlMdDu])
Reply to @Kirsten3531, 14:32 UTC · 2 likes
Kirsten (@Kirsten3531): me: refuses to do chores on evenings and weekends
me: wants to live in a clean house with clean clothes
me: starting up a new blog, writing lots
me: loves to nap and walk in the sunshinealso me: why is my to do list 6 months long??
She just like me fr
Quote tweet, 21:34 UTC · 5 likes
Clown ass programming language
R never made play Russian roulette
Tf is an environment anyway just do the thing
Quoting Aaron Bergman 🔍 (@AaronBergman18): New post: conversation with @austinc3301, dedicated to my fellow "where do u do python" wonderers
Wednesday, May 3, 2023
Reply to @austinc3301, 01:20 UTC · 2 likes
Agus 🔸 (@austinc3301): @AaronBergman18 you can do this, I believe in you
🙏
Quote tweet, 05:29 UTC · 119 likes
I did not expect the literal first shrimp welfare tweet to be from Tesco
Quoting Tesco (@Tesco): @DanAbbottSharks @tescomobile We are working on developing training in best practice fish and shrimp welfare for our supply base. 2/4
Reply to @austinc3301, 05:35 UTC · 2 likes
Agus 🔸 (@austinc3301): @SpacedOutMatt @AaronBergman18 @Laura_k_Duffy @MariGO2thepolls Aaron go for it
If I ignore intuitiveness and just go for what I think we should care about (w some simplifying assumptions I could get into) I get this
Reply to @berggrenpeterm, 21:44 UTC · 2 likes
Peter Berggren🔸 (@berggrenpeterm): @Tesco @tescomobile Shrimp welfare go brrr
Shrimp welfare go *whatever happy shrimp sound like*
Reply to @austinc3301, 21:54 UTC · 1 like
Agus 🔸 (@austinc3301): every time I think, "surely, Python is performant enough to handle this", something goes wrong
Join the rebellion
Thursday, May 4, 2023
Reply to @jam3scampbell, 03:58 UTC · 4 likes
James Campbell (@jam3scampbell): he went all the way
Can someone explain
Reply to @b_n_d_l_e, 04:20 UTC · 3 likes
wise political commentator (derogatory) (@b_n_d_l_e): i love the quality math discourse on https://x.com
@absurdlymax see free will is 110% legit, they proved it
Reply to @codybaldwin, 04:21 UTC · 2 likes
Cody (@codybaldwin): @diviacaroline I operate 4 EC sites (same content different platforms). The “bestsellers” are actually bestsellers. We’re trying to sell stuff, not trick people into buying stuff they may regret. To do that we make it cheaper or offer discounts/promos/points 😂
Although if you’re successful the “bestsellers” will indeed become bestsellers
Reply to @NathanpmYoung, 04:24 UTC · 2 likes
Nathan 🔎 (@NathanpmYoung): While more intelligence is desirable, so is more kindness.
I am not sure I would want to become more intelligent but less kind. Seems like I might despise that person.
Would you want to be a bit dumber and kinder though?
Reply to @joe_zimmerman, 06:00 UTC · 4 likes
Joe Zimmerman (@joe_zimmerman): @alyssamvance Hinton 👏 Christiano 👏 Tegmark 👏 Bostrom 👏 Yudkowsky
@POTUS
Reply to @Leoagua1, 06:43 UTC · 3 likes
Leo (@Leoagua1): @AaronBergman18 @mezaoptimizer Which part?
I’m gathering from comments this is about Tomasik, but why are they discussing him and what is he responding to?
Reply, 06:55 UTC · 2 likes
Max(contrarianism)
Reply to @PicturesFoIder, 23:57 UTC · 4 likes
non aesthetic things (@PicturesFoIder): what are u adding?
$1T
A solution to AI alignment
Friday, May 5, 2023
Reply, 00:02 UTC · 3 likes
Quote tweet, 00:04 UTC · 14 likes
Not if I get there first
(actually even if I get there first)
Quoting alex lawsen (@lxrjl): Tempted to write a highly opinionated ultimate EA adjacent (EA for short) podcast tier list. Which podcasts would you want to see me ranked by me?
Tweet, 00:07 UTC · 1 like
The Adelstein -> Hanania pipeline
Tweet, 04:58 UTC · 3 likes
Next month: “a 9 year old has trained a state of the art LLM with 98% the performance of GPT-4.5-Turbo-with-browsing-codecompiler-roboticArm-enabled-32Mtoken in 16 minutes on his family’s microwave”
Reply to @skooookum, 06:11 UTC · 2 likes
(The tweet being replied to by @skooookum is no longer available.)
Reply to @peterwildeford, 21:34 UTC · 2 likes
Peter Wildeford🇺🇸🚀 (@peterwildeford): @absurdlymax @AaronBergman18 @skooookum I like how this image is deployed in an ambiguous way - it could either mean "I also bring this energy to relationships and it has been bad for me" or "I also experience you bringing this energy to me and it does annoy me".
I’m fending off marriage proposals from so-called “friends” left and right
Jk it’s the first lol
Reply to @frances__lorenz, 21:39 UTC · 1 like
(The tweet being replied to by @frances__lorenz is no longer available.)
Wait I’m actually pretty sure I know who one person is now
Reply to @AndyMasley, 22:47 UTC · 3 likes
Andy Masley (@AndyMasley): Me at the beginning vs 10 pages into "Wild Animal Ethics: the Moral and Political Problem of Wild Animal Suffering"
Not a cute mouse on the cover 😭😭😭
Saturday, May 6, 2023
Reply to @absurdlymax, 02:58 UTC · 5 likes
Max Alexander (@absurdlymax): I drew this picture in my notes for an essay and I remember thinking it was important to what I was saying, but I have no idea what it means
Looks like a dude with very long legs in the batters box waiting for a pitch
Reply to @EaracheEye, 23:56 UTC · 1 like
earachemyeye.bsky.social (@EaracheEye): @jenniferdoleac Take it as a compliment. They wouldn't be so petty and bitter if they didn't want you to stay. I wasn't a fan as a student nearly 40 yrs ago. I imagine it is worse now that it is so much bigger.
Ok one hand, complement (maybe)
On the other, I’m assuming a 5 digit sum of money
personally I’d take the latter and no complement
Sunday, May 7, 2023
Tweet, 12:26 UTC · 42 likes
Uh I don’t think @TEDTalks chose a great title for this one https://www.youtube.com/live/7hFtyaeYylg?feature=share
Reply to @mealreplacer, 12:34 UTC · 11 likes
(The tweet being replied to by @mealreplacer is no longer available.)
Insofar as clarity trades off against fidelity I basically disagree
Like afaict it’s the one place not walled off by a PhD (thinking of academic analytic philosophy) that’s good at high fidelity written communication
Reply to @bat1441, 13:23 UTC · 2 likes
bat (@bat1441): @AaronBergman18 @TEDTalks i don't think that's the title for the talk, other talks on that channel are titled the same eg https://www.youtube.com/watch?v=ku8BK0N9lbw&t=330s
(also it's not the official TED channel)
Thanks and good point but it sure as hell looks that way which is kinda what matters IMO
Reply to @mealreplacer, 13:29 UTC · 2 likes
(The tweet being replied to by @mealreplacer is no longer available.)
I agree you want to max(get point across) | effort and min(effort)| how well you want point to get across
but once you’re on this Pareto frontier there’s a trade off between these two features
Reply to @mealreplacer, 13:34 UTC · 5 likes
(The tweet being replied to by @mealreplacer is no longer available.)
Yeah fair enough! But I don’t think it’s trivial at the current margin to move closer to it
To be annoyingly LW about it, I think it’s close to the 4 dimensional Pareto frontier in writing space when you add the two dimensions “writer talent” and “writer effort”
Reply to @mealreplacer, 13:35 UTC · 1 like
(The tweet being replied to by @mealreplacer is no longer available.)
Actually LOLed
Thread, continued, 13:36 UTC · 1 like
Am guilty of these sins to some extent
Thread, continued, 13:36 UTC · 2 likes
“But you see we can form an 8 dimensional manifold in abstract communication space …”
Reply to @mealreplacer, 13:43 UTC · 9 likes
(The tweet being replied to by @mealreplacer is no longer available.)
Real footage of @mealreplacer
Tweet, 15:35 UTC · 24 likes
New template just dropped
Reply to @AndyKnny, 21:38 UTC · 2 likes
Andrew Kenney (@AndyKnny): A bipartisan bill to allow casinos to offer lines of credit to gamblers has failed. SB23-259. #copolitics
Wait can a random non gambler get a 0% loan by walking into a casino or
Reply to @absurdlymax, 23:07 UTC · 3 likes
Max Alexander (@absurdlymax): @interested_ea @inesferhumi Your interactions keep me going
I ship them
Reply to @inesferhumi, 23:10 UTC · 2 likes
Inés (@inesferhumi): @interested_ea Because i want to do OTHER THINGS we can’t just watch contrapoints every single night
Contrapoint: u could
Contracontrapoint: nope u can’t
Reply to @inesferhumi, 23:30 UTC · 6 likes
Inés (@inesferhumi): Ok i’ll do an AMA now! AMA
(quoting @inesferhumi: HELL YEAH)
What’s a disagreeable or contrarian EA take you need an excuse to share?
Reply to @AngelicaOung, 23:31 UTC · 11 likes
Angelica 🌐⚛️🇹🇼🇨🇳🇺🇸 (@AngelicaOung): Have you ever heard the term “third rail of politics” as something you DO NOT want to touch? That there‘s the actual third rail you do not want to touch.
TIL
Reply to @fasc1nate, 23:34 UTC · 3 likes
(The tweet being replied to by @fasc1nate is no longer available.)
I don’t think the guys in front are gonna make it
Tweet, 23:39 UTC · 37 likes
This is art
@ajeya_cotra @QualyThe @ESYudkowsky @mealreplacer
Monday, May 8, 2023
Reply to @metakuna, 01:36 UTC · 2 likes
metakuna (14/100 blog posts) (@metakuna): it's always "we're in a cost of living crisis", but people never talk about the cost of laughing and loving 😔
(Admittedly the stickers were my sisters choice)
Thread, continued, 01:37 UTC · 3 likes
But the whale lights are all mine 🐳 😎
Reply to @AndyMasley, 01:58 UTC · 4 likes
Andy Masley (@AndyMasley): Is this useful animal communication or is it silly
Can we make them chickens
Reply to @AndyMasley, 02:02 UTC
Andy Masley (@AndyMasley): @AaronBergman18 Would like to yeah
I gotchu
Thread, continued, 02:26 UTC · 2 likes
i dont gotchu https://aaronbergman.neocities.org/chickens
Thread, continued, 02:26 UTC
gpt4 can only do so much
Reply to his own earlier tweet, 02:32 UTC · 2 likes
Aaron Bergman, May 8, 2023: i dont gotchu https://aaronbergman.neocities.org/chickens
u did click on them right
Tweet, 02:39 UTC · 14 likes
Yeah I'd say I'm mostly a front end dev
(video, plays on X)
Thread, continued, 02:40 UTC · 5 likes
tc: $900k
Reply, 03:56 UTC · 6 likes
(The tweet being replied to is no longer available.)
Hate to rain on everyone’s parade and this would indeed be fantastic, but neurotoxicity isn’t the only bad thing that drugs can cause. Like you still could get dependent/addicted for example via normal downregulation
Reply to @eugeneyan, 03:59 UTC · 3 likes
Eugene Yan (@eugeneyan): Ran a simple benchmark (Mandelbrot sets) between Mojo & Python. The speedup is impressive, and it benefits from Python's libraries.
• Python: 1,184ms
• Mojo: 27ms 🤯
• Python (vectorized): 240ms
• Mojo (vectorized): 2ms
R:
Reply to @frances__lorenz, 14:42 UTC · 3 likes
(The tweet being replied to by @frances__lorenz is no longer available.)
Cheezie cheese, my favorite kind
Reply to @daniel_271828, 14:46 UTC · 3 likes
zzDANIEL_ETH (@daniel_271828): @mealreplacer No, it’s actually a fair statement
@StefanFSchubert can you adjudicate this one
Reply to @Laura_k_Duffy, 15:25 UTC · 1 like
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): Oh no problem (and I really hope I didn’t come off hostile!) I’m genuinely interested in finding someone to argue for this on grounds that are consistent and not partisan (so if anyone has one, please reply!). I’m a fan of state identity, seems healthier than partisan loyalty
The principle is that the senate shouldn’t systematically bias American politics in either direction
Thread, continued, 15:28 UTC · 2 likes
Also partisan loyalty >> state loyalty
The parties are ideological, and ideologies are morally substantive things! I don’t see any reason to be loyal to a state unless you’ve committed as such eg by joining the police force
Reply to @RandomSprint, 18:00 UTC · 8 likes
RandomSprint🧭 (@RandomSprint): Yes. I am interested in putting solar panels on my suburban house.
Yes, I would like to front $50k to save hundreds of dollars per month.
Yes, I love 7% loans.
Please come inside and tell me more.
No, I'm not being sarcastic.
No, it isn't opposite day.
Using “I’m not being sarcastic” sarcastically should be a capital crime
(Am I being sarcastic? 🤔)
Reply to @inesferhumi, 18:01 UTC · 2 likes
Inés (@inesferhumi): @FrostedSapling Something really fun that i do with @interested_ea a lot is try to come up with our own theories about psychology and the social world. For example the other day we started working on our own personality classification system
Classify me plz
Reply to @FrostedSapling, 18:02 UTC · 2 likes
Ivan 🔸 (@FrostedSapling): @inesferhumi @austinc3301 @absurdlymax No one ever pronounces my name correctly but like I’m confused how people even screw up your name 😂
How should ppl pronounce ur name Ivan
Reply to @inesferhumi, 18:04 UTC · 1 like
Inés (@inesferhumi): @austinc3301 @absurdlymax Lol yeah i went through a phase of dropping the accent too when i first moved out of peru to but eventually i went back to using it. I think it gives people a hint about how my name is pronounced so that they don’t say stuff like “eye-ns”
Worst misspelling
Reply to @Laura_k_Duffy, 18:12 UTC · 2 likes
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): DC friends, it would be really fun to meet up here if you’re free!
(quoting @CNLiberalism: TOMORROW: DC New Liberals May Social hosted by the DC New Liberals Join us Tuesday May 9th for drinks, games, and conversation at our May social. https://cnliberalism.org/events/dc-6erf4-cmwtw-3g4pr-9a88x)
Hope to bring that gender ratio down to 2%
Tweet, 18:25 UTC · 1 like
Here’s one everyone’s gonna hate:
The genre of “it’s hard to support a family on ~only~ $140k in NYC”-type takes are often reasonable grievances arising from a contingent mix of economic and social/societal circumstances
Reply to @StefanFSchubert, 18:28 UTC
Stefan Schubert (@StefanFSchubert): Many blame oil companies for climate change, but though they have moral responsibility the key actors seem to me to be voters/consumers.
They could easily get other policies if they wanted.
Voter preferences seem to have very little causal effect on govt policy though
I’m not sure who deserves blame tbc, seems like a hard political economy question
Tweet, 19:05 UTC · 3 likes
oh boy oh boy oh boy I wonder if we’re gonna hit the 300s today
Reply to @daniel_271828, 20:23 UTC · 2 likes
zzDANIEL_ETH (@daniel_271828): NOTE: the account @danieI_eth (with a capital "i" instead of a lowercase "L" for the last name) is impersonating me. Please go and report them!
Why do people do this? Seems like a fair bit of effort too
Quote tweet, 20:32 UTC · 6 likes
I feel like there’s an AI safety analogy to be made here
Quoting LMAO CLIPS (@thelmaoclips): Security guard prevents hippo from escaping into public area at zoo
Reply to @technomancers, 20:44 UTC · 4 likes
Technomancers_ai (@technomancers): The belief in superintelligence is based on how expert systems, an outdated architecture, works. https://technomancers.ai/an-ai-system-cant-improve-its-own-code/#more-539
Everything in the thread holds even without recursive self improvement
Reply, 21:06 UTC · 1 like
To be pedantic I disagree with the premise - this is only true of non-EA critiques of EA
Reply to @absurdlymax, 21:08 UTC · 2 likes
Max Alexander (@absurdlymax): @AaronBergman18 Don't slap the hippo :(
🦛🦛🦛
Reply to @Laura_k_Duffy, 21:09 UTC · 2 likes
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): @peterwildeford @LdsEa @AaronBergman18 Regarding life choices, I would never have moved back to MT and bought a house there if I didn't have strong state loyalties. It was a big decision to put opportunity > love of place and family. I still follow their politics and news quite a lot
YOU OWN A HOUSE
Thread, continued, 21:09 UTC · 2 likes
Pretty sure the most expensive thing I own is a laptop lmao
Reply to @Laura_k_Duffy, 21:10 UTC · 1 like
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): Used to!
😱
Reply to @inesferhumi, 22:04 UTC · 3 likes
Inés (@inesferhumi): @austinc3301 @AaronBergman18 @absurdlymax NO
Whatever u say ⍭⊓∈∳
Reply to @Laura_k_Duffy, 22:05 UTC · 2 likes
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): This is news to me 😂
https://www.zillow.com/homedetails/426-Hester-Ave-Lodge-Grass-MT-59050/2058362254_zpid/
Might cop
Reply to @Laura_k_Duffy, 22:10 UTC · 2 likes
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): Sell the laptop and you’ve got your down payment
(gif, plays on X)
Reply to @TetraspaceWest, 22:18 UTC · 3 likes
(The tweet being replied to by @TetraspaceWest is no longer available.)
The White House 12am Jan 1 2024?
Reply to @CarsonJGale, 23:18 UTC · 2 likes
Carson Gale (@CarsonJGale): After interacting with EA/Rat circles, I've found it hard to take smart people seriously who don't adhere to their unique set of norms:
- Giving thoughtful probability estimates when forecasting
- Avoiding strong emotion in conversation
- Avoiding hyperbole
- High decoupling(quoting @ylecun: @vavans @geoffreyhinton 100%, because we will design them to be submissive to us.)
We should strive for full decoupling, none of this “high” half assed bs
Reply to @inesferhumi, 23:20 UTC · 2 likes
Inés (@inesferhumi): @absurdlymax -the janitor who maintains the shared garden has been asking her to stop throwing food at the ducks because it is contaminating everything and attracting stray cats, and in my mom's words he is "trying to turn her neighbors against her"
Her plan:
(gif, plays on X)
Reply to @John_Attridge, 23:28 UTC · 2 likes
John Attridge (@John_Attridge): Honest opinion, does this attempt to increase enrolments come across as too blatant?
@SpacedOutMatt might be interested
Tuesday, May 9, 2023
Reply, 01:44 UTC · 2 likes
Once you add the satiating part it becomes plausible! But price and taste alone don’t suffice
Reply to @inesferhumi, 01:45 UTC
Inés (@inesferhumi): I think values change is more important than almost every other cause area (except probably x-risk) and more people should do work on that rather than direct technical work on specific cause areas. (In the spirit of being contrarian i will not be answering any follow ups 😌)
It’s not super clear to me what this looks like beyond “doing ea”
Pretty sure I have a past tweet on this I can dig up
Thread, continued, 01:47 UTC
Not rly fair to do this since u declared no questions but I’m gonna spend my mean points on it
Tweet, 03:53 UTC · 22 likes
Solenoid_entity deserves his own holiday IMO
The entire EA/LW/alignment informational ecosystem is balancing on this one pseudonymous dude heroically reading everything ever written on these matters into a mic
Thread, continued, 03:57 UTC · 2 likes
What even is a solenoid
Quote tweet, 15:38 UTC · 3 likes
America is funny bc people will be like “I need eight assault rifles to defend myself, and of course I’ll need to carry them around in public”
but when there’s a literal hippo about to fuck shit up someone just slaps him bare handed, no guns in sight
Quoting LMAO CLIPS (@thelmaoclips): Security guard prevents hippo from escaping into public area at zoo
Quote tweet, 17:10 UTC · 6 likes
Busy? Word.
Quoting Arjun Panickssery (@panickssery): Emailing busy person asking for thing? Make easy for them (you owe them): cut pointless words from draft.
Tweet, 18:34 UTC · 8 likes
*Puts on 2000s gangster rap just so I can clean my room*
Many men wish death upon me, too, 50😤🔫🔪🩸🩸🩸 😤
We both be on our grind 💰⛓️🏴☠️ 😤
Reply to @panickssery, 18:38 UTC · 37 likes
Arjun Panickssery (@panickssery): Five years ago Scott Alexander explained that it actually only takes him "a couple of hours" to write a post
he expresses confusion about people who write slowly: "they seem to be able to talk instantaneously, and writing isn't that different from speech" lol
Elsewhere he's claimed that he doesn't really edit his posts
I am the person he is confused about AMA
I’ve been trying to find a chatGPT prompt that will take a transcript of me rambling, extract the ideas I’m getting at without summarizing but while removing repetition and filler words, and turn it into coherent prose
Reply to @vokaysh, 18:43 UTC · 3 likes
🏴☠️jake (@vokaysh): If you haven't succeeded it's because you haven't really been trying. This problem is solved. If you're serious pm me with what you tried
Will do when I get to my computer
Reply to @HCBlanchard, 21:10 UTC · 2 likes
Henry C. Blanchard (@HCBlanchard): My *friend* - I have JUST the tool for you. http://Audiopen.ai made by @louispereira - thank me later
Oh shit big if true
Thread, continued, 21:14 UTC · 4 likes
First pass: better than anything I’ve tried on my own by far. Not *quite* perfect e.g., “despite the short duration” implies I thought wtp would be high even for the 3 minute version but I think it’s clear from the transcript I was referring to the full version w longer limit
Thread, continued, 21:15 UTC · 3 likes
The transcript also isn’t perfect but that’s probably on my end (“gpt before” should say “gpt-4”)
Reply to @ESYudkowsky, 22:02 UTC · 3 likes
Eliezer Yudkowsky ⏹️ (@ESYudkowsky): 0-0
(quoting @blader: new research from OpenAI used gpt4 to label all 307,200 neurons in gpt2, labeling each with plain english descriptions of the role each neuron plays in the model. this opens up a new direction in explainability and alignment in AI, helping make models more explainable and…)
Listen bud you’re gonna have to explain the castling thing for those of us with IQ<200
Wednesday, May 10, 2023
Quote tweet, 03:56 UTC · 33 likes
I like to skip the small talk and dive right into the deep stuff, the stuff that matters. I'm glad Qualy feels the same 💡
Quoting Qualy the lightbulb (@QualyThe): When emailing a busy person it's important to be:
- polite
- direct
- concise
Quote tweet, 05:01 UTC · 18 likes
"My p(doom) definitely went down rather than up"
We're so back
Quoting Eliezer Yudkowsky ⏹️ (@ESYudkowsky): Commentary at greater length:
- I'm encouraged that somebody ran right out and tried this.
- It's not clear (to me, yet) that it worked all that well, or better than expected; I have not yet signficantly updated my model of how technically hard interpretability is.
- It is definitely a positive update toward "people are sincerely trying to have clever nontrivial ideas about using AI to boost interpretability today, ones I did not think of myself, and are running right out and trying them right now and doing interesting things *at scale*, rather than using 'the AI will do our AI alignment homework eventually' to cope with a lack of current progress".
- That said: I have already previously said that interpretability is the easiest place to verify that any sort of AI alignment progress has been made, and pointed to it as one of the few places where I'd have hopes of offering large prizes for progress or eventually getting AI help; since the big central issue with trying to scale human work, *or* getting AI to help, on most parts of alignment, is verifying whether good or helpful work has been done. So, unfortunately, this *is* the place where I previously pointed to as progress being much easier than on other alignment problems; and already predicted that wouldn't be enough to save us.
- *That* said: it's still at least some update in the direction of getting more progress earlier than I expected, and now I have more hope of seeing more exciting work like this later. My p(doom) definitely went down rather than up, upon seeing this.Well done to Steven Bills∗, Nick Cammarata∗, Dan Mossing∗, Henk Tillman∗, Leo Gao∗, Gabriel Goh, Ilya Sutskever, Jan Leike, Jeff Wu∗, and William Saunders∗.
Quote tweet, 07:07 UTC · 5 likes
If you share my values, which you should, this market and tweet is for you!
https://manifold.markets/AaronBergman/what-donationaccepting-entity-eg-ch
Quoting Peter Wildeford🇺🇸🚀 (@peterwildeford): It feels underrated that at this point every single thing @RethinkPriors is doing right now is bottlenecked by needing more money. Time for me to spend a lot of time fundraising!
Reply to @JesseTayRiver, 07:58 UTC · 4 likes
Jesse Smith (@JesseTayRiver): @StefanFSchubert @Laura_k_Duffy Are you considering a move, Stefan?
I think DC could use Schubertism quite a bit
Reply to @JacquesThibs, 08:59 UTC · 5 likes
Jacques (@JacquesThibs): can someone please make a gif for the “the ais got they own minds” part?
(not a gif) (yet)
(video, plays on X)
Reply to @JacquesThibs, 09:46 UTC · 9 likes
Jacques (@JacquesThibs): @AaronBergman18 Oh and keep the subtitles at the same level throughout!
base video (no subtitles): https://drive.google.com/file/d/1A_y5PgaHmV77EJ9Ij6Km7hoyHkfM0_EK/view?usp=sharing
the one right below: https://drive.google.com/file/d/1id-jV0Qdy2y6slVKGuEsJWg-62798d3X/view?usp=sharing
(video, plays on X)
Reply to @daniel_271828, 10:42 UTC · 13 likes
zzDANIEL_ETH (@daniel_271828): Can we get some autotune on this? Or use some of the cool new AI music systems to make this an intro to a song?
This is maybe not the best use of my time or (lack of) skills but
(video, plays on X)
Quote tweet, 10:42 UTC · 3 likes
banger_of_the_century.m4a
Quoting Aaron Bergman 🔍 (@AaronBergman18): @daniel_eth @JacquesThibs This is maybe not the best use of my time or (lack of) skills but
Reply to @_danielle_carr, 11:35 UTC · 13 likes
_(The tweet being replied to by @danielle_carr is no longer available.)
Societal is a legit word it just means something different tho
Reply to @MaxCRoser, 15:06 UTC · 6 likes
Max Roser (@MaxCRoser): What change could we make to @OurWorldInData that would make you think "Oh yes! Finally!" ?
Could be a new feature, new data, a new topic, a change to the structure, or anything else — whatever we can do to make you happy.
A set of gigantic, disgusting csvs with everything that can be linked by year, country, country-year, and anything else like that
Reply to @connormcmk, 15:06 UTC · 2 likes
Connor McCormick (@connormcmk): Just give me a button to export the data to vega-lite so I can make my own graphs
https://github.com/owid/owid-datasets
Reply to @MaxCRoser, 15:23 UTC · 3 likes
Max Roser (@MaxCRoser): What change could we make to @OurWorldInData that would make you think "Oh yes! Finally!" ?
Could be a new feature, new data, a new topic, a change to the structure, or anything else — whatever we can do to make you happy.
Here's another 'compensate for US federal gov't incompetence' one: the USDA seems to be the most authoritative and comprehensive source for nutrition data...but the data is totally incomprehensible.
Like the main spreadsheet is literally 24M rows of empty columns...
Tweet, 16:27 UTC · 26 likes
👃 Smell is weird 👃
Visual field: the year is 2023 and you are in Maryland
Short term memory: 2023 and ur in Maryland
Long term memory-based priors: P(2023 and MD) > 90%
Other senses and propioception: not much to see here
My nose: the year is 2015 and you are in Canada
Reply to @nickcammarata, 17:33 UTC · 1 like
Nick (@nickcammarata): most of my friends similarly stopped drinking alcohol and caffeine and then after reflection drink no alcohol and twice as much caffeine. Anyone who puts caffeine and sugar in the same category (docs commonly do??) is bonkers
Caffeine is legit my favorite drug (haven’t tried a ton tbc)
If it wasn’t for tolerance I’d be at at least .5g/day
Reply to @johnjhorton, 18:34 UTC · 6 likes
John Horton (@johnjhorton): p-values should be uniformly distributed under the null hypothesis these look, uh, high
Wait is this true even for mechanistically correlated variables?
Like conditional on total cholesterol and HDL having p=.9, there shouldn’t be a uniform prior distribution on LDL I think
Reply to @SeanMombo, 18:37 UTC · 2 likes
sunshine ☀️ revival (@SeanMombo): you wanna see the most wholesome skate video of all time? watch till the end man
Reply to @inerati, 20:58 UTC · 4 likes
(The tweet being replied to by @inerati is no longer available.)
Not super unpopular among wonky contrarian Twitteligentsia, but Roe was wrongly decided and quite clearly so
Obligatory: am pro-choice
Reply to @jbfan911, 21:10 UTC · 68 likes
(The tweet being replied to by @jbfan911 is no longer available.)
Meanwhile in the men’s room
(gif, plays on X)
Reply to @DavidPinsof, 21:55 UTC · 1 like
David Pinsof (@DavidPinsof): @Savannah__P I would question whether this phenomenon actually exists. If something is truly a vice, in the sense of being judged negatively by others, then there is no incentive to signal it, and a big incentive to conceal it. I suspect most examples of vice signaling are countersignaling.
+1
An exception might be eg pirates trying to make their threat to kill a hostage appear credible if it is in fact credible
Reply to @AndyMasley, 23:23 UTC · 2 likes
Andy Masley (@AndyMasley): @LinchZhang Pretty stunned that you're getting results for anything other than >100%
Yeah I don’t think people are really grokking this. There is no amount of money I wouldn’t pay if I could pay it, and at least from a selfish perspective dea*h seems clearly preferable
Reply to @tmychow, 23:28 UTC · 2 likes
(The tweet being replied to by @tmychow is no longer available.)
Pt 3 seems not necessary for the conjunction
There are a lot of reasons a misaligned AI would get deployed other than seeming attractive: race dynamics, accident
And “it” sorta implies a single case, whereas it only takes one deployment by one actor
Thread, continued, 23:28 UTC · 1 like
And even object level 33% intuitively seems low but I don’t have any great explicit reasoning here
Thursday, May 11, 2023
Reply to @AndyMasley, 04:43 UTC · 3 likes
Andy Masley (@AndyMasley): @SpacedOutMatt @AaronBergman18 @LinchZhang I had a Very Bad dentist experience that singlehandedly pushed me way into hard consequentialism
One of the more important things I've written I think (http://aaronbergman.net/p/on-suffering)
Reply to @daniel_271828, 05:26 UTC · 20 likes
zzDANIEL_ETH (@daniel_271828): A short story about a normal kid in 2002 who somehow gets access to GPT-4 and gets to live the next couple decades in god mode
Thread, continued, 05:26 UTC · 6 likes
Tweet, 07:16 UTC · 5 likes
Someone needs to make a Calendly for non Calendly people
Reply to @inerati, 07:24 UTC · 2 likes
(The tweet being replied to by @inerati is no longer available.)
Reply to @generic_void, 08:20 UTC · 1 like
(The tweet being replied to by @generic_void is no longer available.)
Nobody actually likes python. It’s a giant preference falsification cascade waiting to begin. It’s basically 1991
Reply to @samlakig, 08:24 UTC · 2 likes
sam laki (@samlakig): @AaronBergman18 @generic_void @inerati @LinkofSunshine It's julia time 🔥🔥
There should be one programming language. Two tops bc apparently there’s a compiled vs not compiled thing going on. 95% serious here
@SpacedOutMatt and @GluskerMartin failed to convince me otherwise
Reply to @generic_void, 08:31 UTC · 1 like
(The tweet being replied to by @generic_void is no longer available.)
I actually grudgingly agree - I’ve been trying and occasionally succeeding lol
Even have a blog post that’s mostly just be being dumb and @austinc3301 being smart in response https://www.aaronbergman.net/p/tf-is-a-kernel
Reply to @NLRG_it, 08:44 UTC · 2 likes
no longer respecting gardenias (@NLRG_it): FDA moment
Rage
Reply to @Laura_k_Duffy, 16:08 UTC · 15 likes
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): 🚨Huge news in animal welfare!🚨SCOTUS upholds California’s law mandating that all pork, veal, and eggs sold in-state must come from animals who have access to minimum space requirements
Justice Gorsuch: “can we please stop reading atextual clauses into the Constitution??”
We’re so beyond back at this point we’ve even started going into positive territory
Quote tweet, 16:24 UTC · 4 likes
These are the votes chatGPT infers from the full brief
Quoting Lewis Bollard (@Lewis_Bollard): Huge win! The Supreme Court just upheld California’s Prop 12, the nation’s strongest farm animal welfare law.
This ruling spares millions of animals from crates and sets a powerful precedent that could help many more.
More details ⬇️
Tweet, 18:54 UTC · 67 likes
Seeing an exclamation point on the EA Forum home page is like hearing a BBC anchor shout “oh my God!” or watching the Super Bowl cut to a still blue screen with the presidential seal
Reply to @adambinksmith, 19:57 UTC · 4 likes
Adam Binksmith (@adambinksmith): it me
🐖🐖🐖 (compliment)
Reply to @austinc3301, 20:18 UTC · 4 likes
Agus 🔸 (@austinc3301): @AaronBergman18 You mean savvycal?
No like “there’s a 60% chance I’ll show up for this if u sign up, 30% I’ll cancel at least an hour before, 9% I’ll cancel under an hour before, and 1% I just don’t show” and both sides can indicate their guessed or empirical probabilities and it’s like no hard feelings
Thread, continued, 20:19 UTC · 5 likes
Like I cannot commit to being both free and in a reasonably productive state at exactly 3pm in 39 days lmao
Quote tweet, 20:22 UTC · 3 likes
Maybe if I shout this into the void it will magically appear
Quoting Aaron Bergman 🔍 (@AaronBergman18): @austinc3301 No like “there’s a 60% chance I’ll show up for this if u sign up, 30% I’ll cancel at least an hour before, 9% I’ll cancel under an hour before, and 1% I just don’t show” and both sides can indicate their guessed or empirical probabilities and it’s like no hard feelings
Friday, May 12, 2023
Reply to @dissproportion, 02:27 UTC · 3 likes
(The tweet being replied to by @dissproportion is no longer available.)
https://open.spotify.com/playlist/3h9t5BMwiy3NPb3PE9IReD
Reply to @dissproportion, 02:32 UTC · 4 likes
(The tweet being replied to by @dissproportion is no longer available.)
I don't even know who she is, it's just empirically the case that all songs called Die Young are bangers
Tweet, 02:38 UTC · 3 likes
Idk who needs to see this but this thing has saved me maybe 15 mins of actual time but 10,000 years' worth of being annoyed
Reply to @RhysLindmark, 04:36 UTC · 2 likes
Rhys (@RhysLindmark): Moloch is the god of coordination failure. 👹
Who is the god of coordination? 👼
Normal old Abrahamic God?
Reply to @absurdlymax, 04:49 UTC · 4 likes
Max Alexander (@absurdlymax): @adamrisom @ohwizenedtortle Being hot is a mindset
@StefanFSchubert fact check
Reply to @absurdlymax, 05:27 UTC · 2 likes
Max Alexander (@absurdlymax): @AaronBergman18 @adamrisom @ohwizenedtortle @StefanFSchubert And that's how Max learned he wasn't an EA 10
Wait no that wasn’t the implication!
Reply to @growing_daniel, 06:53 UTC · 4 likes
Daniel (@growing_daniel): The best tweets are deleted immediately
Why is this spicy? Bc he said “raw dog”? That’s the spice level of a single capsaicin molecule
Reply to @Laura_k_Duffy, 06:55 UTC · 7 likes
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): @RomanianLiberal What if it’s different for each hand?
I promise I already had this in my photos
Reply to @absurdlymax, 07:31 UTC · 3 likes
Max Alexander (@absurdlymax): If I met myself I'd probably think I'm really cool and talented, which is evidence my standards are clearly too low because I suck
Me trying to understand this tweet (object level->{conditional->impossible but conceivable hypothetical->probabilistic forecast}->forecast within hypothetical contradicts more direct forecast about the same -> modus tollens -> basis for hypothetical judgement false ->
(gif, plays on X)
Reply to @maiamindel, 22:09 UTC · 2 likes
Maia (@maiamindel): The World Bank needs to do more. Not climate change, not investment, not poverty. They need to offer checkings and savings accounts to countries. Let Malawi deposit their GDP with you for 3% annual interest rates.
Congrats u just starved all the ppl of Malawi
Reply to @mealreplacer, 22:13 UTC · 2 likes
(The tweet being replied to by @mealreplacer is no longer available.)
Came here to say this, a win for the human!
Reply to @absurdlymax, 22:15 UTC · 5 likes
Max Alexander (@absurdlymax): @peter_wilde_alt Thank god I don't have to reset the counter
Idea: crypto. arbitrage.
Reply to @sama, 22:19 UTC · 2 likes
Sam Altman (@sama): all ChatGPT Plus users getting browsing and plugins over the next week; hope you enjoy!
(quoting @OpenAI: We’re rolling out web browsing and Plugins to all ChatGPT Plus users over the next week! Moving from alpha to beta, they allow ChatGPT to access the internet and to use 70+ third-party plugins. https://help.openai.com/en/articles/6825453-chatgpt-release-notes)
Reply to @CineraVerinia, 22:35 UTC · 4 likes
(The tweet being replied to by @CineraVerinia is no longer available.)
Dodged a bullet
Reply, 22:37 UTC · 12 likes
They need to stop inventing new programming languages fr. There should be 2 tops, just pick them and move on
Reply to @EAheadlines, 23:00 UTC · 9 likes
EA Lifestyles (@EAheadlines): What's your EA origin story?
I wanted to make friends with more generous people, but Googling "generosity London" just got me a bunch of minimalism stuff, so I googled "altruism London" and here I am!
(quoting @gemlpaterson: I want to read more personal stories by and about the people of the Effective Altruism movement But like, fun ones, not CVs I wrote a forum post about it: https://forum.effectivealtruism.org/posts/6Rrv6G74tzAyDZxz9/i-want-to-read-more-stories-by-and-about-the-people-of)
Running fundraiser in hs->had to choose what charity to donate to -> GiveWell, but already vegan at the time so was definitely predisposed
Reply to @absurdlymax, 23:13 UTC · 1 like
Max Alexander (@absurdlymax): For me it was more doing a philosophy degree and they advertised the ea college group there
Only got a phil minor but pretty sure I never heard or saw it referenced, or even anything adjacent
My Intro to Ethics prof was a Kantian 😔 and we mostly talked about soldiers etc being “morally injured” if they were compelled to like kill ppl
Reply to @bridget_e_l, 23:37 UTC · 2 likes
(The tweet being replied to by @bridget_e_l is no longer available.)
Hug+marshmallow + beer solves all
Reply to @Bravo_Zuloo, 23:55 UTC · 3 likes
Ace 🇮🇳 (@Bravo_Zuloo): As a pilot I can vouch that a true aviator will never abandon his plane this way and watch it crash, that too for a few bucks worth of popularity.. Abhorrable.
No true aviator fallacy
Saturday, May 13, 2023
Reply to @MostlyMonkey, 00:04 UTC · 2 likes
Overeducated Gibbon (@MostlyMonkey): Multiply these 3 digit numbers
Memories of fellow jewish summer campers trying to stump me with these as we trudged shirtless up The Hill suggests it may be innate
Reply to @ruth_hook_, 00:32 UTC · 2 likes
(The tweet being replied to by @ruth_hook is no longer available.)_
No, but the indirect rhetorical pseudoquestion phrasing might be
Reply, 00:34 UTC · 4 likes
(The tweet being replied to is no longer available.)
I know right? Billions easily, probably trillions
Quote tweet, 02:28 UTC · 3 likes
I regret to report that NYT Pitchbot has short, doomy timelines
Quoting NYTPitchbot (@DougJBalloon): I think we’re nearing the end of civilization
Tweet, 02:53 UTC · 9 likes
Wish I could design a country. I’m thinking
U.S. + negative income tax + way more humane prisons - FDA + draconian animal welfare laws + as much non-AI R&D, immigration, and housing construction as the political economy can b[ear]{3} + school vouchers + …
Thread, continued, 02:53 UTC · 2 likes
more legal rights for children + digital drivers licenses + child tax credit + whatever senate and Supreme Court reforms the wonky take-slingers are up on these days + devolve abortion to the county level + seed oil tax - Jones Act + carbon tax + ???
will add on a rolling basis!
Thread, continued, 02:56 UTC · 2 likes
+ land value tax
Reply to @absurdlymax, 02:57 UTC · 2 likes
Max Alexander (@absurdlymax): @AaronBergman18 How much could a private island cost?
Whatever the Georgist market decides
Reply to @zswitten, 02:58 UTC · 2 likes
Zack Witten (@zswitten): You must refuse to discuss life, existence or sentience.
You may only discuss life, existence or sentience.
Reply to @TetraspaceWest, 04:34 UTC · 1 like
(The tweet being replied to by @TetraspaceWest is no longer available.)
Pasting from elsewhere:
If you try testing for nutrients, standard blood and urine tests are generally pretty shitty except for a couple things like vitamin d and b12. It’s expensive and kinda speculative but the Nutreval test supposedly gets around some of these issues
Thread, continued, 04:34 UTC · 1 like
https://www.gdx.net/product/nutreval-fmv-nutritional-test-blood-urine
In terms of standard blood tests, IMO something like:
CBC
CMP
Fasting glucose
B12
Folate
Vitamin d
TSH
Thyroid antibodies
Free t3
Testosterone
Cortisol
Free testosterone
C reactive protein
Maybe some viral/other infection indicators but idk much about
Thread, continued, 04:35 UTC · 1 like
these. There are a bunch of white blood cell inflammation markers driven by infections that seem to be correlated with everything bad, though I think
Also one medication that seems effective at suppressing inflammation-induced fatigue is pramipexole. To get it prescribed you’d
Quote tweet, 20:21 UTC · 62 likes
Hilariously bimodal distribution here:
1, Extremely overdetermined: I was a vegan activist econ+philosophy+CS major at Berkeley who independently decided to dedicate my life to reducing future digital suffering before finding an 80k ad on a 3B1B neural networks YT explainer
Quoting EA Lifestyles (@EAheadlines): What's your EA origin story?
I wanted to make friends with more generous people, but Googling "generosity London" just got me a bunch of minimalism stuff, so I googled "altruism London" and here I am!
Thread, continued, 20:21 UTC · 49 likes
2, extremely contingent: I missed my flight for the first time ever and wound up sitting next to Will MacAskill on the rebook who was only on that flight because some businessman offered to buy his ticket on the previous one for $3,000 and he felt morally compelled to oblige
Reply to @RafaRuizdeLira, 20:29 UTC · 8 likes
Rafael Ruiz ⏸️🔸 (@RafaRuizdeLira): @AaronBergman18 Wait, what?
Story time?
Tbc I made these ones up lol
Reply to @EAheadlines, 20:44 UTC · 6 likes
EA Lifestyles (@EAheadlines): omg I thought they were your stories and I was like WHAT
Both? “I ignored the 80k ad and was only convinced by MacAskill years later” would be quite the story
Tweet, 21:11 UTC · 119 likes
I think the happiest I’ve ever been was a half hour Uber ride during which the girl I had a crush on repeatedly asked me to explain random things to her the entire time
Hate to reinforce gender stereotypes but 🤷♂️
Thread, continued, 21:11 UTC · 22 likes
I remember trying to get at the concept of aboutness and how computers have states that are about other things whereas a soccer ball does not, even if the latter can also be thought of as a physical information processing system (info=momentum vector of kick)
I think she got it
Reply to @absurdlymax, 21:21 UTC · 2 likes
Max Alexander (@absurdlymax): Someone has a service dog with a little graduation hat
Pic required
Reply to @Laura_k_Duffy, 21:56 UTC · 36 likes
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): What’s the gender stereotype here? Tbh I love it when people ask me questions about technical details
Guys love explaining things, especially to girls (which isn’t always reciprocated hence mansplaining discourse lol)
Reply to @Laura_k_Duffy, 22:01 UTC · 6 likes
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): Oh lol I’m such a mansplainer haha (I explain things to men).
But to be expected because I only recently stopped sleeping on a mattress on the floor
Sweet!
What’s the current bottleneck on cultured meat production?
Is my notion that there’s a U curve relating political salience (y axis) to tractability via legislation correct? If so where is the peak? Idk how you’d quantify the former exactly tho
(Will think of more)
Thread, continued, 22:04 UTC · 7 likes
I have some hesitancy about RPs stance on neuron counts as a proxy for moral weight for the following reason: linear neuron counts scale elegantly with moral weight of population size assuming n individuals are n times more important than 1
Whereas if you look at cognitive/
Thread, continued, 22:06 UTC · 7 likes
emotional capacities, it seems pretty clear that the qualitative abilities of n individuals is identical to those of a single one (eg 1T flies can’t do math like 1 human), so the RP framework isn’t robust to changing ontology to considering multiple organisms as one unit
Thread, continued, 22:07 UTC · 5 likes
Question: (why) am I wrong?
Reply to @SpacedOutMatt, 22:12 UTC · 9 likes
Matt 🔸 (@SpacedOutMatt): @Laura_k_Duffy @AaronBergman18 Mansplainer (female)
Explady
Reply to @SpacedOutMatt, 22:16 UTC · 2 likes
Matt 🔸 (@SpacedOutMatt): @AaronBergman18 @Laura_k_Duffy Banger
(gif, plays on X)
Reply to @Laura_k_Duffy, 22:31 UTC · 1 like
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): I hate to be glib, but it feels like we could use "number of hairs on body" as an equally theoretically good metric. You need to validate using observation, and that relies on qualia and judgment (and probably leads to challenging conclusions regarding animals moral weight)
Glibness is a virtue never apologize
As an actual explanation, yeah that would be bad, but insofar as it correlates with whatever does matter it might be a fine proxy!
Sunday, May 14, 2023
Reply to @MatthewJBar, 00:48 UTC
Matthew Barnett (@MatthewJBar): Earlier I also accidentally posted this. I clicked the box saying not to share on Twitter but I'm pretty sure the website is broken and posted anyway. It was effectively a Twitter virus, I suppose.
Yeah i I unclicked the box and it still posted
Reply to @AlecStapp, 00:54 UTC · 15 likes
Alec Stapp (@AlecStapp): "These are three of my favorite writers. And with all due respect, they’re all massive losers."
Examples like this are a pretty interesting challenge to the “IQ correlates well with everything” theory (which I mostly subscribe to)
Thread, continued, 00:55 UTC · 9 likes
Tbc the challenge here is “brilliant individual failed all his exams” not “good writers don’t have to be smart”
Reply to @PradyuPrasad, 00:57 UTC · 8 likes
Pradyumna 🇸🇬 (@PradyuPrasad): I'm pretty sure they all have high IQs
Reply to @inesferhumi, 00:59 UTC · 2 likes
Inés (@inesferhumi): @absurdlymax Maybe that i don’t use any productivity apps and just track of my tasks on a checklist in apple notes? In the office maybe it’s that i am kind of anti social
Classic midwit meme
Thread, continued, 00:59 UTC · 6 likes
Reply to @absurdlymax, 01:03 UTC · 2 likes
Max Alexander (@absurdlymax): Don't tell Dustin
Asanas for teams tho. Haven’t heard of like a large corporation managing 10k employees with a single shared apple note, though I would love to see it tried
Thread, continued, 01:03 UTC · 3 likes
Maybe Asana the company operates this way lmao
Reply to @moskov, 01:53 UTC · 3 likes
(The tweet being replied to by @moskov is no longer available.)
Leaked photo
Reply to @SpacedOutMatt, 02:08 UTC · 3 likes
Matt 🔸 (@SpacedOutMatt): @dustin_crummett @Laura_k_Duffy @AaronBergman18 Yeah I was going to bring this up as well. I think the RP weights cover this concern via the conscious subsystems component. 2 stapled-together pigs would be 2 conscious subsystems, unlike a normal pig
this is a good point I hadnt considered, need to think about what it implies for my original point
Reply to @RafaRuizdeLira, 02:11 UTC · 8 likes
Rafael Ruiz ⏸️🔸 (@RafaRuizdeLira): What the fuck happened in the middle
Lmaoo
Reply to @absurdlymax, 02:11 UTC · 3 likes
Max Alexander (@absurdlymax): @TetraspaceWest @RafaRuizdeLira I think it's the inverse
Yeah I asked @Laura_k_Duffy lol
Reply to @Laura_k_Duffy, 02:19 UTC · 1 like
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy):
Hypothetical megapig go “”
Reply, 02:33 UTC · 3 likes
Type of guy
Reply to @BlairAfter10, 06:33 UTC · 2 likes
blairafterhours (@BlairAfter10): @RafaRuizdeLira https://onlyfans.com/ofkitten91
Onlyfans but a fully clothed woman flips through the dictionary and asks you to explain arbitrary nouns
Reply to @absurdlymax, 06:37 UTC · 3 likes
Max Alexander (@absurdlymax): I won the persistence award a bunch in elementary school but I'm not sure why or for what, and I don't think I understood at the time either
Maybe if u keep trying to understand you’ll eventually succeed
Reply to @JMannhart, 11:05 UTC · 2 likes
Jonathan Mannhart 🔎 (@JMannhart): I am glad Aaron retracted the hair thing. Whatever that is
He almost escaped
(gif, plays on X)
Quote tweet, 11:38 UTC · 5 likes
Reply w/ most interesting thing you can recall reading or listening to (full-on 'most interesting' seems kinda high stakes lol)
I'll start: the 80k episode with Christopher Brown on slavery abolition https://80000hours.org/podcast/episodes/christopher-brown-slavery-abolition/
Quoting Nathan Barnard (@NathanB60857242): @mattyglesias If you want a recommendation, I'm currently listening to "Understanding the New Testement" on audible and it's one of the most widly interesting things I've ever read.
Tweet, 12:58 UTC · 2 likes
China gets away with de facto censoring US tech companies because it has a gigantic market: ~$18T in GDP
In terms of GDP, Turkey ≈ Ohio tho
Reply to @AndyMasley, 19:22 UTC · 3 likes
Andy Masley (@AndyMasley): I quit for 2 months and decided that coffee actually just makes my life better
Reply to @AndyMasley, 21:43 UTC · 1 like
Andy Masley (@AndyMasley): Found Astra Taylor via Jeff Mangum via Neutral Milk Hotel and she made a documentary about philosophers in 2009 and Peter Singer was in it talking about effective giving and it stuck with me until 2013 or so when the movement itself came on my radar
(quoting @EAheadlines: What's your EA origin story? I wanted to make friends with more generous people, but Googling "generosity London" just got me a bunch of minimalism stuff, so I googled "altruism London" and here I am!)
This tweet is absolute chaos - I guess NMH rings a bell as a band, but who are Mangum and Taylor and how do these three relate? Mystery person Astra just makes documentaries? For fun? About philosophy? About Singer? In 2009? And you chewed on Singers words for 4 years?
Quote tweet, 22:07 UTC · 10 likes
This may not be literally wrong, but I think it (and other, similar stylized data) can be super misleading
A poor person work-hour means, for example, standing (no luxuries like a chair) behind a counter with essentially no bodily or cognitive (ie modulating level of engagement)
Quoting George Mack (@george__mack): The 0.1% of ideas this week.
The rabbit holes I've gone down -- without the distractions.
8 best examples:
Tweet, 22:07 UTC · 6 likes
autonomy, or carrying around boxes in a warehouse with the same constraints
An hour of high-earning knowledge work might not be *fun* (much of the time, at least) but it’s qualitatively different from monotonous, highly constrained demanding or boring labor
Thread, continued, 22:07 UTC · 3 likes
Of course all this is generalization - overlapping distributions, skilled work can be very stressful etc, etc.
But there’s a fundamental, systemic reason for the asymmetry:
Thread, continued, 22:07 UTC · 4 likes
skilled labor tends to be performed best by reasonably happy and comfortable workers, whereas unskilled labor can be squeezed out of even the miserable
Thread, continued, 22:07 UTC · 7 likes
And all this holds even *if* the data reflects reality
Something tells me salaried workers, likely unintentionally, inflate time worked more than those who can see that number in their paycheck
Monday, May 15, 2023
Reply to @LinchZhang, 03:12 UTC · 2 likes
Linch (@LinchZhang): I unironically think EA Forum Twitter etc will be noticeably better if people just had moderately better reading comprehension.
May not be tractable tho.
I think this is the spiciest Linch tweet I've ever seen
Reply, 13:27 UTC · 1 like
No it’s not it’s literally experiencing pain and discomfort
Reply, 13:28 UTC · 10 likes
Free 👏testosterone 👏 is 👏 all 👏 that 👏 matters 👏
Reply to @austinc3301, 13:31 UTC · 2 likes
Agus 🔸 (@austinc3301): @PradyuPrasad @mealreplacer I think, yeah
Not 100% sure if he’s taller than me
Neither am I. I mean I fully expect there to be a 7” difference but it’s not literally impossible he’ll lose both legs in a battle with QALY or smth
Reply to @frances__lorenz, 14:46 UTC · 4 likes
(The tweet being replied to by @frances__lorenz is no longer available.)
Central example of a Frances tweet
Quote tweet, 14:52 UTC · 4 likes
In the name of full decoupling, I think this is ~literally true
(but also that the lack of imagination and competence is downstream of the problem’s difficulty, not character flaws)
Quoting Yann LeCun (@ylecun): What characterizes AI doomers (as well as many pessimists) is a limitless imagination for catastrophe scenarios and a complete lack of imagination (or competence) for ways to avoid them.
Reply to @NathanpmYoung, 15:18 UTC · 1 like
Nathan 🔎 (@NathanpmYoung): the plan
a mild dose of amphetamines in the water supply
(quoting @tomhfh: “They can be seen as helpful, because they do cognitively enhance a person’s ability to perform… a person will report benefits” - so says leading NHS consultant. Genuine question: if that’s the case why are we so worried about people without ADHD accessing these mild medicines?)
No bad idea!! Low dose amphetamines are dangerous. Don’t use <4ish mg at a time
Reply to @samswoora, 15:29 UTC · 8 likes
Samswara (@samswoora): At what point do I just throw in the towel and switch teams
image attached
(quoting @AaronBergman18: @samswoora Free 👏testosterone 👏 is 👏 all 👏 that 👏 matters 👏)
TRT? I kinda think it’s worth a try for anyone in the lower third of the reference range or so. But a bunch of the dosing regimens are insane - aim should be like 60-90th percentile not 99.9th and more isn’t better unless you literally only care about getting big
Reply to @CineraVerinia, 15:31 UTC · 8 likes
(The tweet being replied to by @CineraVerinia is no longer available.)
This is a total blackpill on apps. Like if *you* got *one* date there is no hope for the rest of us lmao. Not gonna renew Hinge Premium
Reply to @bryan_caplan, 15:34 UTC · 3 likes
Bryan Caplan (@bryan_caplan): Why should we give people who are clearly guilty of heinous crimes a fair trial?
(a) To avoid a slippery slope that ends with punishing the innocent
(b) To avoid mistreating the clearly guilty
Because sometimes you’re gonna be wrong about that “clearly guilty thing,” and even more frequently other people are gonna disagree and then you need a way of ascertaining whether they’re clearly guilty and congrats you reinvented “fair trials”
Reply to @moskov, 16:08 UTC · 2 likes
(The tweet being replied to by @moskov is no longer available.)
guys like this tweet not Yud’s for the benefit of all humanity including @robertwrighter
Reply, 16:14 UTC · 2 likes
I wonder if there’s anyone around who is good at coding social media things
(Not me I managed to fail to figure out how to post “hello world” to this acct via api)
Reply to @QualyThe, 17:23 UTC · 4 likes
Qualy the lightbulb (@QualyThe): the world if everyone volunteered at EA Global
Wait I thought it would look exactly the same bc there’s already a surplus of hypercompetent, gigacontientious volunteers
Reply to @xwanyex, 17:28 UTC · 8 likes
(The tweet being replied to by @xwanyex is no longer available.)
Kinda did the opposite for me tbh (which I don’t think is good tbc) bc the car was totaled but I walked out with like a small bruise
I’m actually not sure whether I’m super lucky beyond just the luck of having a modernish car (2008 I think) w airbags and such
Tweet, 18:13 UTC · 8 likes
One of the better pieces of news I’ve received is that I’d set the fire alarm off for my dorm room only, meaning I’d annoyed like 30 ppl but didn’t force ~1000 to evacuate
From inside the room this wasn’t at all clear!
Thread, continued, 18:13 UTC · 3 likes
It was just a little smoke anyway smh
Reply to @frances__lorenz, 18:21 UTC · 1 like
(The tweet being replied to by @frances__lorenz is no longer available.)
What about 8000 files (this is beta testing the site is obviously not ready)
Reply to @davidad, 19:05 UTC · 2 likes
davidad 🎇 (@davidad): if you believe that you share exactly 50% of your DNA with your mother, and you also believe that you share over 98% of your DNA with a chimpanzee, then it may be good to notice this and downregulate the extent to which you literally believe strings that classify as science facts
(quoting @salonium: You often hear that full siblings "share 50% of their DNA." But did you know this is actually the median and the range is pretty wide, roughly between 40–60%? https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.0020041…)
I think I wonder what’s up with this in my head like once a month and just never look into it lol
Reply to @ruth_hook_, 20:06 UTC · 5 likes
Ruth Hook (@ruth_hook_): I was in fact 4' 11" back then and I moved about the country with all of my belongings in just that backpack
Still growing at 19 gang
Tuesday, May 16, 2023
Reply, 00:39 UTC · 1 like
Discontinuity analysis ftw
Quote tweet, 01:42 UTC · 120 likes
Ah poor analogy!
“FDA=extremely bad” is a naive-sounding disagreeable rationalist hobbyhorse that also happens to be ~entirely correct
Quoting Max Tegmark (@tegmark): "Don't regulate AI – just trust the companies!" Does he also support abolishing the FDA and letting biotech companies sell whatever meds they want without FDA approval, because biotech is too complicated for policymakers to understand?
Reply to @PicturesFoIder, 03:07 UTC · 4 likes
non aesthetic things (@PicturesFoIder):
They should be longer
Reply to @CarsonJGale, 03:10 UTC · 5 likes
Carson Gale (@CarsonJGale): Went running with a buddy today at the beach near Crissy Field and a shark washed up right at our feet on the beach. We tried throwing it back in but it kept getting washed back up - my friend eventually carried it to the fishing pier & threw it off into deeper water.
Assumed I was gonna see like 100k when I glanced at the view on this tweet, not 20
I’d like to invest in the memetic success of this tweet
Reply to @CarsonJGale, 03:57 UTC · 2 likes
Carson Gale (@CarsonJGale): Here is your investment opportunity
https://manifold.markets/CarsonGale/will-my-video-of-my-friend-throwing?r=Q2Fyc29uR2FsZQ
Superforecasting status awaits!
Reply to @GluskerMartin, 04:06 UTC · 3 likes
Martin Glusker (@GluskerMartin): @AaronBergman18 Isnt the FDA bad because because its overly cautious and doesnt properly weight costs and benefits? Because an overly cautious and burdensome agency for AI seems good
My tweet was ambiguous and yeah I agree with everything you said (except for “overly” which seems by definition bad)
It’s functionally bad bc it tries to set the FDA as a baseline for good functioning govt and use agreement on that to build agreement on AI reg by analogy
Reply to @thebestestpie, 04:11 UTC · 9 likes
(The tweet being replied to by @thebestestpie is no longer available.)
Oh I’ll eat that bullet for breakfast
I can imagine an FDAish thing that’s better than nothing, and barring this it’d ideally be phased out over months, but if the only two choices were “magically cause it to disappear tmr” and “don’t do that” I’d definitely choose the former
Reply to his own earlier tweet, 04:15 UTC · 21 likes
Aaron Bergman, May 16, 2023: Ah poor analogy! “FDA=extremely bad” is a naive-sounding disagreeable rationalist hobbyhorse that also happens to be ~entirely correct
The maybe-barely-not-impossible-reform I’d suggest here (as a starting place, I’m sure it needs some fine-tuning) is to move every approved medication down one regulatory notch
So most random non-“controlled” prescription-only drugs become OTC like ibuprofen…
Thread, continued, 04:24 UTC · 13 likes
(Maybe antibiotics are an exception here)
Most controlled substances can be Rxed by a nurse practitioner or MD. Also none of the other bullshit restrictions like 30 day max supply
Probably absolute dosing limits on recreational drugs (ie no 1g/day adderall prescriptions) solves
Thread, continued, 04:24 UTC · 11 likes
almost all the problems ppl associate with street use
Reply to @thebestestpie, 04:35 UTC · 4 likes
(The tweet being replied to by @thebestestpie is no longer available.)
Reply to @yacineMTB, 05:01 UTC · 15 likes
kache (@yacineMTB):
Don’t disrespect VSauce like that
Tweet, 05:11 UTC · 5 likes
Spend your cringe points wisely, but don’t die too far in the black
Reply to @Laura_k_Duffy, 12:17 UTC · 2 likes
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): For those of us look the same as when we were 19, let’s see those memes/screenshots from when you were 19
HIGHEST IMPACT TWEET OF 2023 there’s a 0% chance I would have solved greatest mystery of my existence (how did I acquire the grey/yellow 4-sigma comfyness shorts and how can I buy more?) otherwise
Don’t ask me why I took a picture of the screen, idk either
Reply to @brianchau57, 13:50 UTC
Brian Chau (@brianchau57): Its the correct analogy, this is what will happen if you let regulators near AI. The intellectually honest position which I've heard some AI doomers articulate is that they're willing to let the most idiotic, anti-progress people dictate AI because its preferable to apocalypse.
Yeah, FDA-level regulation of AI > no/minimal regulation > full force indefinite suppression (which I think literally nobody wants)
If we were one thalidomide-type crisis away from extinction, the FDA would be good
Also:
Reply to @kjoshanssen, 23:47 UTC · 2 likes
Bjørn Kjos-Hanssen (@kjoshanssen): What's your favorite interval of ℝ that is closed under multiplication?
(-1, 1)
Reply to @depthsofwiki, 23:50 UTC · 9 likes
depths of wikipedia! (@depthsofwiki): this SL kid must have been so cool
OW is the real 🐐
Reply to @ruth_hook_, 23:52 UTC · 2 likes
(The tweet being replied to by @ruth_hook is no longer available.)_
Ex 5
(gif, plays on X)
Reply to @kjoshanssen, 23:52 UTC · 1 like
Bjørn Kjos-Hanssen (@kjoshanssen): Not closed under the empty product 🤨
Sounds fake
Wednesday, May 17, 2023
Reply to @skooookum, 05:14 UTC · 2 likes
(The tweet being replied to by @skooookum is no longer available.)
Middle Ages meerkat IMHO
Quote tweet, 05:36 UTC · 60 likes
Related though not Google’s fault (mostly):
I had a school Google account that I used from maybe 4th grade through the end of hs with ~every project/paper/etc I ever worked on attached to it through Classroom or Docs
The county shut us out on graduation, everything gone.
Quoting Chris Beiser (@ctbeiser): imagine getting out of a coma or jail and realizing that your entire life is in shambles because Google decided to delete your account because you hadn't logged in for two years https://blog.google/technology/safety-security/updating-our-inactive-account-policies/
Thread, continued, 05:36 UTC · 26 likes
I even made a half-hearted attempt to backup everything, but we couldn’t share outside the school system domain so that didn’t work
Obv there was nothing of great intellectual importance in there, but that was like a decade of my life!
Thread, continued, 05:36 UTC · 26 likes
Maybe this is insane and it obviously wouldn’t be worth the effort on an individual level, but I kinda think “government claims sole ownership of documents created by 15 y/o because they administrated the associated Google Workspace account which the kid was *legally obligated*
Thread, continued, 05:36 UTC · 21 likes
to sign up for and use in this way” is grounds for a court case
Funny part is we’re talking about probably like 4GB of info, which I’d gladly pay the ~$1/year to host
Reply to @ruth_hook_, 05:38 UTC · 3 likes
(The tweet being replied to by @ruth_hook is no longer available.)_
Classic “wanna kiss” *has Hershey kiss for plausible deniability* energy
Reply to @ruth_hook_, 05:39 UTC · 4 likes
(The tweet being replied to by @ruth_hook is no longer available.)_
Great minds
Reply to @ruth_hook_, 05:54 UTC · 3 likes
(The tweet being replied to by @ruth_hook is no longer available.)_
That is just a pant, not an underpant
Quote tweet, 07:14 UTC · 4 likes
Mystery befuddling biologists (me) for decades (.5 of them maybe) has been solved. A true breakthrough in the field
Congrats to the team (of 1, @salonium) who worked day and night for weeks (.01ish of them??) on the paper (thread)
Deserves a Nobel Prize (qt) IMO
Quoting Saloni (@salonium): Follow up:
Why do people say humans share 98% of their DNA with chimpanzees, but 50% with their siblings or parents? 🤔
These are measuring two different things.
Reply to @StefanFSchubert, 22:12 UTC · 2 likes
Stefan Schubert (@StefanFSchubert): @mattyglesias Birthday significance is likely U-shaped, declining but eventually increasing again (with local peaks at 10, 20, 30, etc)
And an absolute maximum at 0
Tweet, 22:39 UTC · 5 likes
Took me a full 3 days to realize that, in Yud’s head, the apples vs potatoes thing was prima facie too unrealistic for anyone to believe bc apples grow on trees and potatoes grow underground
Thursday, May 18, 2023
Reply to @ruth_hook_, 03:37 UTC · 2 likes
(The tweet being replied to by @ruth_hook is no longer available.)_
Molybdenum
Am I seeing no dha?!
I suspect boron is essential
Also:
absolute pain at sat fat in same category as trans
More sodium than that
More vit d than that
If they’re saying fiber is necessary they’re wrong
Reply to @zbingeldac, 10:30 UTC · 2 likes
zbingeldac (@zbingeldac): @AaronBergman18 @yacineMTB How is Vsauce based?
Those were the best 20 mins of middle school Aaron’s month
Reply to @robertwiblin, 22:05 UTC · 6 likes
(The tweet being replied to by @robertwiblin is no longer available.)
Rob ur like 19 AI discourse cycles behind with this one
Friday, May 19, 2023
Reply to @mealreplacer, 08:38 UTC · 5 likes
(The tweet being replied to by @mealreplacer is no longer available.)
If I had 30 conversations in a weekend I would simply fizzle out of existence
Reply to @_djpn, 13:56 UTC · 7 likes
𝐃𝐉𝐏𝐍 (@_djpn): Babies are awesome because they can go through the most insane tumble, be fine, and then look to a trusted adult to see if that was an Okay thing to experience or if they should freak out like a FIFA player
Most insane tumble
Reply to @yacineMTB, 15:03 UTC · 8 likes
kache (@yacineMTB): protip; instead of a database, use the file system. just wrap it in an interface.
I’m guessing this is sarcastic but it unironically seems right to me. Like as long as it’s <500gb I can just download a csv what’s the problem
Thread, continued, 15:03 UTC · 4 likes
Obligatory: I have no idea what I’m talking about unironically
Reply to @EAheadlines, 18:30 UTC · 8 likes
EA Lifestyles (@EAheadlines): The food at EAG is noticeably less weird than previous years
A comparison pic labelled pre-FTX and this labelled post would go so hard
Tweet, 19:22 UTC · 34 likes
Reading (ok fine “reading”🎧) the Parfit biography and I don’t think he’d get into any US Ivy-tier schools nowadays
His pre-university achievements basically consisted of being way smarter than everyone else, and even eg the SAT doesn’t differentiate between top .1% and top .01%
Thread, continued, 19:22 UTC · 13 likes
(and the former evidently isn’t sufficient for admission) https://www.aaronbergman.net/p/most-ivy-smart-students-arent-at
Reply to @SpacedOutMatt, 19:33 UTC · 10 likes
Matt 🔸 (@SpacedOutMatt): @AaronBergman18 Lol I've been listening to it too
Both him and Yud - same quirks as me but 3x more extreme (including intelligence) lol
Reply to @dwarkesh_sp, 19:49 UTC · 3 likes
Dwarkesh Patel (@dwarkesh_sp): Realizing from the prep of upcoming episodes that I need at least an undergrad level of physics in order to understand many interesting & important topics
Looking for a tutor to help me bloom 2-sigma through some textbooks
May be ideal for a grad student, happy to pay. DM me
@AndyMasley
Tweet, 21:00 UTC · 7 likes
Nobody:
The contrarian demon on my shoulder: you should share a take that your followers won’t like
Me: 👍
Me: the US should have stayed in in Kabul, occupying it indefinitely
Reply to @GluskerMartin, 21:08 UTC · 2 likes
Martin Glusker (@GluskerMartin): @AaronBergman18 It’s not that I don’t like it, I just don’t think it’s correct. Give a cost benefit analysis and then we can talk
All my opinions come from the 80k pod pt. ♾️
(Not a rigorous detailed analysis clearly)
Reply to @ohwizenedtortle, 21:27 UTC · 2 likes
oh dandy lion🔸 (@ohwizenedtortle): @AaronBergman18 @HamishDoodles Still burning your fingers on the bowl/potato, Aaron?
Yes I’m down to my last finger
Tweet, 22:57 UTC · 20 likes
Call me an old country bumpkin but I can't believe this is real
Reply to @mealreplacer, 22:59 UTC · 4 likes
(The tweet being replied to by @mealreplacer is no longer available.)
Gonna start putting my height in lightyears on dating apps thx for the idea
Reply to @ohwizenedtortle, 23:25 UTC · 2 likes
oh dandy lion🔸 (@ohwizenedtortle): @AaronBergman18 Damn, this is a GPT-4 thing?
yea chatGPT+
Saturday, May 20, 2023
Reply to @absurdlymax, 00:42 UTC · 2 likes
Max Alexander (@absurdlymax): Despite knowing extremely little about AI I will continue to tweet my random thoughts
It seems most things are not aligned with our interests so baseline it does seem intuitive that an AI wouldn't by default. Many humans aren't even aligned with each other
I mean I agree with the conclusion but the countervailing reference class is “technologies produced by humans” which are at least good for *some* humans almost always
Reply, 03:57 UTC · 3 likes
(The tweet being replied to is no longer available.)
I’m not sure I’m gonna do this. In part it seems substantially different than eg buying meat because OpenAI has infinite capital and Saas economics are weird and in part there are no plausibly identifiable victims (ie recompense for being harmed)
Thread, continued, 03:57 UTC · 3 likes
so hard to see the case for offsetting per se rather than just donating where I think is best
Reply to @ohwizenedtortle, 05:50 UTC · 4 likes
oh dandy lion🔸 (@ohwizenedtortle): @AaronBergman18 @an_observer__ (which is where?)
RP! https://manifold.markets/AaronBergman/what-donationaccepting-entity-eg-ch?r=QWFyb25CZXJnbWFu
Reply to @GjMcGowan, 11:24 UTC · 10 likes
George McGowan (@GjMcGowan): *Frances steps up to the mic*
*The whole room holds it’s breath*
“HEY GIRLIES!! 😍🥰😊”*Instant standing ovation, several people faint, a pair of 14 year olds in the front row start sobbing*
*”WE’RE SO BACK🐐👍😤”*
*clouds part*
Reply to @CineraVerinia, 13:53 UTC · 5 likes
(The tweet being replied to by @CineraVerinia is no longer available.)
Boy do I have a theory for you https://www.aaronbergman.net/p/contra-dynomight-contra-me-but-also
Reply to @CineraVerinia, 14:08 UTC · 2 likes
(The tweet being replied to by @CineraVerinia is no longer available.)
You’ve gotta be phenylalaninemaxxing
Tweet, 15:14 UTC · 3 likes
The App Store thinks I’m poly
Reply to @polisisti, 17:26 UTC · 5 likes
Alessandro (@polisisti): This is the most "trained in philosophy" thing I've ever heard on the 80,000 Hours podcast
(not that that's a bad thing!)
…and with the emergent interaction of the two
Tweet, 22:15 UTC · 114 likes
Thread, continued, 22:15 UTC · 6 likes
https://80000hours.org/podcast/episodes/joe-carlsmith-navigating-serious-philosophical-confusion/
Thread, continued, 22:16 UTC · 3 likes
I'm unreasonably happy with this meme
Sunday, May 21, 2023
Reply to @parafactual, 02:54 UTC · 4 likes
You (@parafactual): i created the infohazard bounty (http://infohazards.org/submit) to learn if strong infohazards exist
thus far, i don't think i've received any that have qualified, which is evidence against people knowing any strong infohazards
the bounty for the first strong infohazard is $110
(quoting @ChazakielDoremi: I love watching the Infohazard Discourse. I'm fully on 'no [strong] infohazards exist' side but holy shit the 'no, actually infohazards are real!' side is funny They either go 'no believe me I have super secret knowledge but it would be harmful to say it', or else they-)
I feel like people with access to this kind of thing and who haven’t shared aren’t gonna change their mind for $100
Reply to @parafactual, 03:11 UTC · 2 likes
You (@parafactual): @AaronBergman18 how much do you think i need to offer?
I don’t know! There’s one thing I can think of personally that I don’t think I’d even want to track down the particulars of for selfish reasons but would probably do for idk $5k?
Tbc I wouldn’t really want you to make this offer even if you had $1B. Everybody loses
Tweet, 08:56 UTC · 7 likes
Lol at chatGPT just straight up refusing my banal coding request, saying essentially “I’m sorry but parsing the 200-page html of a YT video url with regex is just unhinged, even though yeah technically you could [get the desired info]; click here to get an API key from Google”
Reply to @daniel_271828, 09:01 UTC · 1 like
zzDANIEL_ETH (@daniel_271828): Looks like my instincts were correct
Let them pursue their passion for decentralized databases in peace
Quote tweet, 14:29 UTC · 4 likes
Carlsmith is so smart, insightful and eloquent – shame he gets a bunch of things wrong
Quoting Aaron Bergman 🔍 (@AaronBergman18):
Reply to @zachsilberberg, 14:32 UTC · 30 likes
(The tweet being replied to by @zachsilberberg is no longer available.)
I’ve literally never seen this before
Reply to @SpacedOutMatt, 14:36 UTC
Matt 🔸 (@SpacedOutMatt): @AaronBergman18 I don't currently have 3.5 hours
Right + good: abstract language is pointing to super concrete important visceral things
Seems good to take seriously that there’s actually a whole ass world out there even though ~all my qualia just relate to me
Crazy train = flawed metaphor and not necessarily crazy
Thread, continued, 14:38 UTC · 3 likes
Wrong: he says moral realism is false but it’s true fr fr (relates to the first point above acc I think)
This is from his two blog posts recently I’ll find but I’m guessing they’ll discuss
Reply to @BenHayum, 15:07 UTC · 2 likes
(The tweet being replied to by @BenHayum is no longer available.)
🐐
Reply to @SMTuffy, 15:13 UTC · 4 likes
(The tweet being replied to by @SMTuffy is no longer available.)
This sounds like common sense and a good heuristic but IMO it just turns out to seem less so in the context of the history of medicine
Drugs designed to selectively target a specific mechanism are generally shit, and unintended side effects go along with a bunch of effective
Thread, continued, 15:14 UTC · 2 likes
and all-things-considered helpful ones. MAOIs, chemotherapy, meformin
Reply to @AndyMasley, 18:48 UTC · 2 likes
Andy Masley (@AndyMasley): This is basic & optimistic but I'm convinced that you should always be extremely kind (but firm) when you're arguing about something contentious, and if you think you'll be more effective if you're mean you're basically always wrong & your impulse comes from dumb evolution stuff
Definitely true if effective = convincing
In a presidential debate, the goal is to win institutional power, not change minds per se
Reply to @RichardMCNgo, 18:54 UTC · 3 likes
Richard Ngo (@RichardMCNgo): Which public figure is most capable of combining intellectual rigor and moral forcefulness?
I have a guess in mind but want to hear other suggestions first - it seems like a rare combination these days!
Probably not literally the best (sry dude) but @MatthewAdelste1 is punching above his educational attainment
Reply to @BootsPlatypus, 19:09 UTC · 2 likes
Platypussy in Bio (@BootsPlatypus): why not nausea?
It’s not directly transmitting information about the world to your nervous system
Reply to @Laura_k_Duffy, 19:15 UTC · 2 likes
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): Really great to meet so many of the EA Twitter friends irl, you all are wonderful ☺️
Ok imma be the one to say it: there’s a >1% chance there are two of the same ppl in this photo
Reply to @Laura_k_Duffy, 19:19 UTC · 3 likes
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): Wait what lol
Same glasses???
Monday, May 22, 2023
Quote tweet, 00:56 UTC · 6 likes
(whole meme below, but the new part gets cut off)
Quoting Joe Carlsmith (@jkcarlsmith): see also
Thread, continued, 00:56 UTC · 5 likes
Thread, continued, 00:58 UTC · 3 likes
should sponsor me for repeatedly showing that Word is a solid graphic design program
Reply, 01:28 UTC · 1 like
Finally something non-philosophical we disagree about
Reply to @ArtirKel, 03:02 UTC · 24 likes
José Luis Ricón Fernández de la Puente (@ArtirKel): @ruth_hook_ https://www.reddit.com/r/whatisthisthing/comments/lgtlza/what_is_this_boat_off_the_coast_of_long_island_ny/
save u a click: "...a working barge, and those piles are lowered into the sea floor to stabilize the vessel while working"
Reply, 10:53 UTC
A whole bunch of lines of reasoning, some of which are at least made more salient for consideration from my personal experience - hard to tell whether that’s more like bad bias or epistemically good irl skin in the game
The main thing tho is that I think the observational and
Thread, continued, 11:44 UTC · 1 like
I fucked up the tweet order pretty bad so:
Tweet, 11:52 UTC · 8 likes
The eyes of capitalism are truly omniscient
Reply to @austinc3301, 11:57 UTC · 1 like
Agus 🔸 (@austinc3301): in no world did I imagine that the code I made for a research fellowship would get included in the cpython repo
idk what the c is doing there but good job👍🐍🐍
Tuesday, May 23, 2023
Reply to @Laura_k_Duffy, 01:41 UTC · 1 like
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): EA has a lot of class issues I think, and this makes me sad :( Eg. college recruiting, even when done at non-elite universities, is still presupposing that the people we should disproportionately spend resources reaching out to are ones going to college!
I’m coining the term “butter bullet” for when you have a bullet that’s super easy to bite (at least to the speaker)
^butter bullet
Tweet, 08:44 UTC · 11 likes
Feel like this could be a legitimate internet record of some sort - has anyone gone nine years between tweets from the same handle or blog posts at the same domain?
Tweet, 08:54 UTC · 3 likes
Interesting that useful predictive quantitative models either have a single digit number of parameters (regression, decision trees) or literally billions (ML)
I feel like there’s some insight to be gleaned here but haven’t figured out what
Reply to @QualyThe, 09:01 UTC · 3 likes
Qualy the lightbulb (@QualyThe): ohhh you've thought of some clever clarification like whether to discount the effect of greater wealth driving up overall fish consumption? just answer the question bitch
How could it possibly be good?
Reply to @lisatomic5, 09:04 UTC · 5 likes
lisatomic (@lisatomic5): in the same way that nerds tend to reinvent normie knowledge from first principles, western medicine is heading toward "illnesses are actually chakra imbalances"
IBD = Crohn’s and ulcerative colitis
IBS = symptoms but no clear cause
Didn’t they mean the latter?
Tweet, 10:05 UTC · 11 likes
If you’re reading this, the tweet via text message thing worked
Reply to @austinc3301, 18:17 UTC · 5 likes
Agus 🔸 (@austinc3301): Tomorrow is my last day in London, what should I do? 🧐
Leave (at the scheduled time) (among other things) (unless you have a change of heart or citizenship status)
Wednesday, May 24, 2023
Quote tweet, 08:34 UTC · 3 likes
When ur AI box is a…computer
Quoting Liron Shapira (@liron): Skeptics: How could an AI ever realistically take over a billion devices?
Microsoft:
Tweet, 09:11 UTC · 3 likes
(gif, plays on X)
Thursday, May 25, 2023
Quote tweet, 00:23 UTC · 4 likes
Many are saying https://github.com/AaronBergman/RPmeme/blob/main/so many pixels its basically real life.png
Quoting Reilly Wood (@reillywood): The Python noob experience: mention the "cognitive overload of having 20 different ways to install things", get a dismissive answer that touches on virtualenv, pip, setuptools, wheel...
Reply to @SpacedOutMatt, 01:07 UTC · 2 likes
Matt 🔸 (@SpacedOutMatt): @peterwildeford When Conda?
Stop stop stop
Reply to @daniel_271828, 07:01 UTC · 1 like
zzDANIEL_ETH (@daniel_271828): If you say so…
So anti-paleo?
Reply to @EAForumPosts, 11:46 UTC · 3 likes
EA Forum Posts (@EAForumPosts): Hello, this bot is now being run by the forum team at @CentreforEA. Huge thanks to @nikosbosse for setting it up in the first place 😍
Can we get one for shortform 😍?
Friday, May 26, 2023
Reply to @Laura_k_Duffy, 00:13 UTC · 3 likes
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): PSA: Led Zeppelin IV is the best album to listen to while doing computer programming
(quoting @janecoaston: Horniest song ever, a rock banger, and a song about ringwraiths, this album rules)
I invite your music taste to join the 21st century
Tweet, 00:55 UTC · 6 likes
Adding @jkcarlsmith to my list of “excellent thinkers+writers who still somehow manage to get the final answer wrong” (no. 2 alongside Caplan) https://podcasts.google.com?feed=aHR0cHM6Ly9mZWVkcy5idXp6c3Byb3V0LmNvbS8yMDM0NzMxLnJzcw%3D%3D&episode=QnV6enNwcm91dC0xMTc5NDYzMA%3D%3D
Tweet, 00:58 UTC · 2 likes
I don’t have a fully developed rebuttal yet but two of the main points are:
1) Carlsmith equivocating here between claims about what one objectively ought to do and claims about what is objectively good
Thread, continued, 00:58 UTC · 2 likes
2) A false notion that meta-ethical hedonism depends on the infallibility of subjective experience via introspection
Thread, continued, 00:59 UTC · 2 likes
I herby challenge Carlsmith to a duel to resolve this disagreement
Reply to @CineraVerinia, 16:41 UTC · 2 likes
(The tweet being replied to by @CineraVerinia is no longer available.)
How many grams of protein are you eating?? Pretty sure protein consumption -> protein catabolism (I’m guessing working out increases this above and beyond dietary metabolism) -> urea nitrogen which has to be cleared by kidneys but needs to be dissolved in water first…
Reply to @AlecStapp, 16:44 UTC · 25 likes
Alec Stapp (@AlecStapp): This contrarian defense of La Sombrita is the opposite of persuasive
This might be the worst argument I have ever seen for anything
Tweet, 16:50 UTC · 402 likes
One of my nerdiest achievements is accurately diagnosing a friend with anemia because he was munching on the ice in his water
I didn’t save him or anything, he already knew and was just like “yeah ur right good job”
Reply to @NathanpmYoung, 16:57 UTC · 41 likes
Nathan 🔎 (@NathanpmYoung): @AaronBergman18 wait, I do that.
I mean idk what the false positive rate on this is but might be something to check out!
Pretty sure labcorp will do a cbc for $100 or smth if it’s hard to get a doctors order
Reply to @mmt_lvt, 18:51 UTC · 2 likes
(The tweet being replied to by @mmt_lvt is no longer available.)
Damn
Reply to @JacquesThibs, 23:15 UTC · 13 likes
Jacques (@JacquesThibs): This was the first date in a while where I was actually nervous 😂
He’s made it boys
Saturday, May 27, 2023
Reply to @Laura_k_Duffy, 01:54 UTC · 1 like
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): I think, though the exercise is useful, the conclusion might directionally change if we were to input effects on the lives of wild animals. (Then there’s both issues surrounding individual welfare and issues if you think that species extinction is bad)
Is there a consensus that the more beef -> habitat destruction effect dominates over more beef -> ghg -> more trapped sunlight -> more food -> higher equilibrium wild animal populations across the board effect
Tweet, 02:11 UTC · 39 likes
Ngl I’m just about 50/50 on this really actually for real being an (intentional) simulation of some sort - up from 3%ish when I first learned about the simulation hypothesis
Sounds absolutely nuts to me too but I really don’t think I’d take either 5:4 bet
Reply to @absurdlymax, 02:15 UTC · 3 likes
Max Alexander (@absurdlymax): @AaronBergman18 So escape is possible
I mean idk I haven’t given that any thought but P>0
Reply to @ruth_hook_, 04:56 UTC · 3 likes
(The tweet being replied to by @ruth_hook is no longer available.)_
U can order one of these
Reply to @TheFigen_, 05:00 UTC · 1 like
The Figen (@TheFigen_): A bear coming out of its hibernation period.
I think he woke up on Monday morning! 😂
All mammals are the same fr
Reply to @norabelrose, 10:15 UTC · 3 likes
Taylor Belrose (@norabelrose): Mechanistic interpretability is cool, but I don’t think it’s very useful for making trustworthy AI.
Building trust in a person means understanding them at a psychological level- their beliefs and values- not at a “mechanistic” level. We need a different kind of interpretability.
I’m agnostic about the first statement, but the second doesn’t seem like great evidence for it: (1) humans have evolved to understand other humans at the level they have access to, which is individual-psychological at best, (2) we can only look at a brain’s operation in a
Thread, continued, 10:18 UTC · 1 like
real live person with very indirect, inference based tools like fMRI (afaik), and (3) even with a literal preserved brain, neuroscience just isn’t mature enough to describe every plausibly relevant causal connection between “modules” at any layer of abstraction
But with neural
Thread, continued, 10:20 UTC · 1 like
nets all these issues go away - it’s like a brain perfectly optimized for understanding and modifying in all ways except complexity (which is a gigantic factor tbc, probably the main one)
Reply to @pearl_please, 10:49 UTC · 5 likes
(The tweet being replied to by @pearl_please is no longer available.)
I respect the poasting
Reply to @QualyThe, 10:57 UTC · 5 likes
Qualy the lightbulb (@QualyThe): Effective Altruism is a question, and that question is "what if we made airport dads but for civilization?"
(quoting @visakanv: Guy describes how his friend is the group’s Airport Dad (takes everyone’s passports, double-checks everything, etc) the comments are full of ladies who wanna marry him)
Reply, 11:27 UTC · 3 likes
Srue
Reply to @reconfigurthing, 12:26 UTC · 19 likes
Elias Schmied (@reconfigurthing): Why the change?
The 3% was probably lower than was ever justified on merits just bc of my default probability tax on new weird ideas with convincing sounding arguments
Then demonstrations of advanced AI both as empirical evidence for simulation feasibility and as vibes-based intuition building
Thread, continued, 12:28 UTC · 2 likes
that the world is extremely weird, which combines with a much deeper appreciation that the basic scientific materialist metaphysics/worldview is at least some degree of incomplete/inadequate mainly bc of consciousness and quantum weirdness (ik not necessarily their interaction!)
Thread, continued, 12:30 UTC · 3 likes
This sounds like fake reasoning IMO but I’ll stand by it - the default “no simulation” hypothesis isn’t built on mountains of empirical or even philosophical evidence, just sort of moderately justified vibes + occams razor
Reply to @SpacedOutMatt, 14:26 UTC · 2 likes
Matt 🔸 (@SpacedOutMatt): @NathanpmYoung @AaronBergman18
Thread, continued, 14:28 UTC · 2 likes
For other ppls reference and my own admiration
Reply to @davidad, 14:54 UTC · 5 likes
davidad 🎇 (@davidad): I have been telling people for months that Bayesian reasoning is fundamentally overconfident. Now here’s the proof! This is why we need to use infraBayesian models in safety-critical contexts.
(quoting @YulingYao: My controversial opinion: Bayes is guaranteed to overfit, for any model, any prior, and any data point https://www.yulingyao.com/blog/2023/overfit/)
Can someone disambiguate “Bayesian modeling” as used by Yao and, like, what I think of as “doing Bayesianism,” which is using the formula to assign a probability to a proposition upon a new bit of evidence
Reply to @p_millerd, 15:37 UTC · 3 likes
Paul Millerd (@p_millerd): The US dog to baby ratio is way too high
We need more micropigs
Reply, 16:48 UTC · 2 likes
⚠️ Spoiler ⚠️
Reply to @panickssery, 16:50 UTC · 1 like
Arjun Panickssery (@panickssery): When people ask for the "most underrated" <thing> they usually mean an unpopular one that's really good but often the most underrated things are actually the most popular, which are even better that their popularity suggests
It's like the Planet Money episode about how LeBron James is the most underpaid American because of NBA salary caps
Eh variance increases at the bottom though, I’d be surprised if anyone famous enough for me to know of is the most underrated player
Reply, 16:54 UTC · 2 likes
@ohabryka
Reply to @NathanpmYoung, 16:56 UTC · 3 likes
Nathan 🔎 (@NathanpmYoung): In 50 years, will the most pedantic people in your life still be men?
Yes it me (if we’re still around)
Reply to @thebradbain, 16:58 UTC · 1 like
Brad Bain (@thebradbain): when people talk about having a gym buddy this is what I imagine
What if we just parallel play with the AGI
Reply to @typedfemale, 17:51 UTC · 23 likes
typedfemale (@typedfemale): c++ showed everyone you shouldn’t design a language by committee and now rust is making the same point about organizations
R is literally a constitutional meritocracy
Reply to @QualyThe, 19:13 UTC · 4 likes
Qualy the lightbulb (@QualyThe): [at the Meta offices]
intern: I- I'm sorry Mr LeCun they didn't have the baguettes you like
yann lecun *threateningly*: no pain no GAN
This is too niche
Thread, continued, 19:13 UTC · 1 like
For me
Thread, continued, 19:14 UTC · 1 like
Wait I think I get it
Thread, continued, 19:15 UTC · 1 like
But it should be “desperately” or smth yes not threatening the guy in my interpretation
Tweet, 19:42 UTC · 7 likes
Whenever I think I have something clever or - God forbid - actually insightful to say, the tweet gets like 3 likes tops
I’m just plugging Twitter straight into my default mode network from now on
Tweet, 21:43 UTC · 12 likes
When you ask ChatGPT for constructive criticism and it doesn't say "I tried super hard but this draft is perfect there's nothing to change"
Reply to @QualyThe, 22:39 UTC · 3 likes
Qualy the lightbulb (@QualyThe): narrator: his job will not be fine
His job is gonna be fine - great, even - it's the guy that's in trouble
Sunday, May 28, 2023
Reply to @ArtirKel, 02:25 UTC · 4 likes
José Luis Ricón Fernández de la Puente (@ArtirKel): @ruth_hook_ What about alcohol? Do you wake up very happy after drinking a lot the previous night (what happens to me)?
You familiar with this? https://www.reddit.com/r/hangovereffect/
Quote tweet, 06:55 UTC · 3 likes
I like to keep things nice and balanced
Quoting oh dandy lion🔸 (@ohwizenedtortle): new psychopathology test just dropped
Reply to @absurdlymax, 09:27 UTC · 2 likes
Max Alexander (@absurdlymax): @QualyThe This deserves a billion likes
Back in the FTX days…
Reply, 10:31 UTC · 2 likes
(Besides super recent discussion of this) The word shows up on exactly two google results, both random Hindi YouTube videos
Reply to @peterwildeford, 12:05 UTC · 2 likes
Peter Wildeford🇺🇸🚀 (@peterwildeford): Pretty impressed that the Apple M2 chip on my new laptop is a ~2x speedup on my squigglepy benchmarks than the Intel Core on my prior laptop
This is an infohazard I just deleted the Apple Store app bc all the new things are too shiny
Reply to @AndyMasley, 12:22 UTC · 3 likes
Andy Masley (@AndyMasley): My life is pretty utopian right now 10/10
1) lfg
2) ehhh unless you’ve been hiding the addiction free side effect free heroin-r-ketamine-d-amphetamine-DMT-phenylalanine-MDMA-caffeine-stem cell to mu opioid receptor-dense neuron neurogenesis-stimulating agent recipe from us I kinda think you’re wrong about the denominator
Reply to @JesseTayRiver, 14:35 UTC · 3 likes
Jesse Smith (@JesseTayRiver): Apparently 7 years ago I built a concrete fire pit in my backyard. If you're thinking about building something circular it's a really stupid idea and you shouldn't.
:(
Reply to @davidad, 14:47 UTC · 2 likes
davidad 🎇 (@davidad): When @GaryMarcus and others point out that GPT-4 is bad at chess and therefore not close to AGI, it falls flat for me.
But when I can’t coax GPT-4 to defeat me at *tic-tac-toe*, I start to think there’s something even more deeply wrong than I realized.
Willing to bet this doesn’t hold up if you use simple notation (explicitly saying not to draw the board if need be) and the base model instead of chat version
Maybe 65% that these suffice, but >80% the latent capability is in there and would come out with enough hacks
Reply to @PradyuPrasad, 14:48 UTC · 4 likes
(The tweet being replied to by @PradyuPrasad is no longer available.)
Kinda wanna give this a try tbh
Reply to @pearl_please, 17:26 UTC · 3 likes
(The tweet being replied to by @pearl_please is no longer available.)
Give them a chance :(
Reply to @liron, 17:27 UTC · 20 likes
Liron Shapira (@liron): This is an IQ test
In which direction? I don’t have strong views on organic produce but plausibly it’s worth the price difference, idk why using a stock photo would change that
Reply to @RafaRuizdeLira, 17:36 UTC · 2 likes
Rafael Ruiz ⏸️🔸 (@RafaRuizdeLira): Do you buy physical books you already read (in digital/audiobook) because you really liked them?
No but at least once I've done the opposite: buying a digital copy to more easily reference when writing about after reading the irl thing
Tweet, 18:16 UTC · 21 likes
Not exactly the NYT but just found this lmao https://cheezburger.com/17249029/dude-gets-dunked-on-for-proposing-housing-developers-start-building-underground
Reply to @absurdlymax, 19:04 UTC · 5 likes
Max Alexander (@absurdlymax): You're famous
my assistant will be responding to your query
Tweet, 20:24 UTC · 134 likes
I think I’m equal parts pro-book (u learn stuff) anti-book (“6 paragraph blog post”) anti-anti-book (“the point is to marinate ur brain which takes time”), anti-anti-anti-book (“no way they’re the best way to do this”), and anti-anti-anti-anti-book (“lindy so maybe they are!”)
Reply to @AndyMasley, 20:26 UTC · 6 likes
Andy Masley (@AndyMasley): My very edgy take is that I'm very pro good books but very anti bad books
Yglesias-core
Reply to @adam__isom, 20:39 UTC · 3 likes
adam ✧ ❥ ~ (@adam__isom): @AaronBergman18 me.jpg
also, books are often cumulative, and help close 'inferential gaps'. the 6-paragraph crowd seems to ignore this entirely??
plus, books are hard to write as-is; some unspecified, second-brain/deeply-internally-networked book style would be even harder
Yeah, I think the case for books existing/being written is much stronger than for individuals reading them cover to cover; they’re a self-contained package in a way a set of blog posts aren’t usually considered to be
Reply to his own earlier tweet, 20:51 UTC · 16 likes
Aaron Bergman, May 28, 2023: I think I’m equal parts pro-book (u learn stuff) anti-book (“6 paragraph blog post”) anti-anti-book (“the point is to marinate ur brain which takes time”),…
Anti^5-book: “a lot of nonfictions’ value comes from mutual-mutual knowledge of the contents/“epistemic network effects”; revealed preference for podcasts + blogs means their value even to an individual is above what you’d naively expect. This is a change enabled by novel tech”
Thread, continued, 20:52 UTC · 13 likes
Anti^6-book: for the same reason, books are relatively under-consumed, and there’s disproportionate alpha in knowing what others don’t
Thread, continued, 20:58 UTC · 9 likes
Anti^7-book: indeed, but what an inefficient way to gain alpha. To a first approximation ~every text (in the general sense) hasn’t been read by anyone. So total quality consumed is the effect that dominates
Thread, continued, 20:58 UTC · 11 likes
Anti^8-book: ya know who else is anti-book? S*m 🅱️an*man 🍟 (bad guy)
Ok that’s all I got
Reply to @absurdlymax, 21:00 UTC · 2 likes
Max Alexander (@absurdlymax): If you think I haven't been funny enough lately it's because I lost all my drafts when I got a new phone
Nooo
This≈library of Alexandria burned
Reply to @absurdlymax, 23:39 UTC · 2 likes
Max Alexander (@absurdlymax): @austinc3301 "Do people not read fiction?" Wonders man who predominantly reads comics, which are the blog posts of fantasy
I do not read fiction
Fr I find narrative nonfiction way more compelling and interesting bc, like, it happened!
Monday, May 29, 2023
Reply, 02:31 UTC · 3 likes
dude go to sleep
Reply to @Laura_k_Duffy, 02:32 UTC · 2 likes
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): What are your favorite history podcasts?
*Can include episodes on history from non-history podcasts
Reply to @ESYudkowsky, 02:36 UTC · 4 likes
Eliezer Yudkowsky ⏹️ (@ESYudkowsky): Abolish the FDA; repeal all occupational licensing including in healthcare; decriminalize sex work; decriminalize housing; delegalize marriage and college; allow free banking. Regulate nuclear materials. Ban >GPT4 model training runs and gain-of-function research on pathogens.
can someone make a "in this house" sign ?
Reply to @CineraVerinia, 03:59 UTC · 3 likes
(The tweet being replied to by @CineraVerinia is no longer available.)
Correct take
Tweet, 04:03 UTC · 23 likes
That’s an interesting point.
Reply to @PradyuPrasad, 11:37 UTC · 2 likes
(The tweet being replied to by @PradyuPrasad is no longer available.)
Make $1T
Tweet, 13:11 UTC · 6 likes
Think I’m re-entering my diet soda arc
Reply to @GjMcGowan, 13:15 UTC · 7 likes
George McGowan (@GjMcGowan): Guy who takes LSD and realizes the fundamental oneness and bliss of existence vs Guy who takes LSD and realizes how dirty his house is
Apparently it’s dopaminergic (unlike other psychedelics)
Reply to @SwiftOnSecurity, 13:18 UTC · 8 likes
SwiftOnSecurity (@SwiftOnSecurity): I finally found a way to administer servers remotely.
I’m in the tiny sliver of the Venn diagram that’s the overlap of “algo thinks I’ll get this” and “I have no idea what any of this means”
Reply to @GjMcGowan, 13:18 UTC · 9 likes
George McGowan (@GjMcGowan): @AaronBergman18 What does that mean for people who don’t understand (would never be me)
Acts a little like a stimulant (eg adderall)
Reply to @Laura_k_Duffy, 13:37 UTC
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): @AaronBergman18 Not enough caffeine in this picture, must replace the caffeine-free with Pepsi Zero
I’m a full decoupler
Reply to @baesrule1, 13:54 UTC · 2 likes
The Affective Altruist 💛 (@baesrule1): @AaronBergman18 @Laura_k_Duffy Gotta agree with Laura on this one, you should switch over to whole foods like Coke/Pepsi Zero 💛
Thank you for caring about my well-being ❤️
Reply to @JoshLekach, 13:56 UTC · 3 likes
🌴 Josh Lekach 🌴 (@JoshLekach): This is a green flag
Real men don't care too much about gossip
Sisters do be getting mysterious jobs tho
“Marketing” …wtf does that mean??
Reply to @ExiledInfoHaz, 15:16 UTC · 22 likes
ExiledInfoHaz (@ExiledInfoHaz): Let me give my explanation for my question on status. The facts about status we need to explain:
Cool/high status:
- singer/musician in a band
- kayaking/rock climbing enthusiast
- painter/artistUncool/low status:
- plays Warhammer
- prepper with basement full of supplies and many skills like first aid and marksmanship
- Physically fit manual laborer (e.g. works in a meatpacking plant)It's hard to spot the pattern here. But there is a pattern! To understand it, you have to understand signalling games.
What do people care about at the end of the day?
The central lie of our society is that all people are equal in value.
Nothing could be further from the truth: humans judge each other on a value scale like ravenous vultures eye up a kill. Value is a combination of various durable, persistent traits that are generally useful for achieving goals - strength, charisma, wealth, intelligence, youth, physical health, good genes, and powerful friends and allies, social savvy, etc.
But because people know these things are all good (and their absence is bad), there is a strong incentive to not be truthful about them. What's the solution?
Signalling.
Why are rock-climbers emitting a positive signal, but preppers emit a bad signal? "Prepping signals being an outcast, poor social skills, high anxiety." (@AVMiceliBarone got this right).
Rock climbing is expensive and requires both intelligence and fitness. It also implies risk-taking - you can easily kill yourself if you get it wrong and fall off a cliff. Warhammer is low-stakes - if you lose a battle, you do not actually die.
For almost everything that people broadly agree is high status, there is an explanation along these lines. Status games are hard-to-fake pointers to a person's value. Alpha for people, if you like.
@AVMiceliBarone again:
"Warhammer signals a certain amount of wealth, but possibly low physical fitness, excessive risk aversion, lack of ambition, poor social skills. These are not necessarily implied by playing Warhammer, but if you had high physical fitness, less risk aversion, ambition, good social skills you would be probably doing something else. Warhammer is maybe a particularly bad signal, because it implies that you have above average intelligence, and yet you are using it on something useless, hence you are probably psychologically abnormal."
Strength 10 hit, multiple wounds, no saves of any kind allowed.
🔲
Agree with most of this, but…
“The central lie of our society is that all people are equal in value. Value is a combination of various…traits that are generally useful…”
This is true for status≈instrumental social value, and totally false for intrinsic/moral value
Reply to @StefanFSchubert, 15:23 UTC · 4 likes
Stefan Schubert (@StefanFSchubert): It sounds incredible for the simple reason that it's not true.
"If you get hit by a car driving 100 mph, you'll die".
Pretty sure most understand that.
(quoting @yishan: The majority of human beings cannot fully comprehend a statement that has a conditional as a key part of its meaning. This sounds incredible to those who can, but I'm pretty sure it's true.)
I mean this is something people understand in a pre-verbal sense bc they live and interact with cars and have been knocked down etc
Totally different when P and Q are abstract and conceptual
Reply to @ExiledInfoHaz, 15:52 UTC · 4 likes
ExiledInfoHaz (@ExiledInfoHaz): > intrinsic/moral value
intrinsic/moral value is fake. It doesn't exist. At least if you look at the way people actually behave.
I mean the ability to experience happiness and suffering, not some spooky “human dignity” sense
If you’re an illusionist about consciousness or a nihilist, fine, but I’m assuming those aren’t the reasons you disagree (?)
Reply to @ExiledInfoHaz, 16:21 UTC · 4 likes
ExiledInfoHaz (@ExiledInfoHaz): The ability to experience happiness and suffering doesn't matter. Warhammer players experience suffering when people socially isolate them. But who cares? I have never in my life heard anyone express concern over the subjective harms from bullying nerds.
Why?
Signalling, again.
Signalling matters!
I care, not that it means much to write that on a keyboard. But more importantly the person being bullied cares
Reply to @AVMiceliBarone, 16:26 UTC · 3 likes
Antonio Valerio Miceli Barone (@AVMiceliBarone): It's still signalling: the EAs who care, or pretend to care, about e.g. shrimp welfare signal that they are secure enough in their position that they don't need the loyalty and support of their family, nation, race, etc.
Sure, that’s probably a partial explanation, but it doesn’t negate the actual prevention of suffering that occurs
Being able to explain altruism doesn’t make it disappear
Tweet, 16:42 UTC · 4 likes
Does anyone know if DC has a minimum apartment size law? There’s plenty about basement rentals, but for actual units I can only find like one article from 2012
I mostly just wanna know who I should be mad at for there being ~0 places up for <$1k rent
https://dc.urbanturf.com/articles/blog/is_dcs_housing_market_ready_for_microunits/6104
Thread, continued, 16:42 UTC · 4 likes
I’d be bunk bed maxxing, hanging shit on the ceiling, putting all three spatial dimensions to good use
Reply to @michael_wiebe, 16:59 UTC · 2 likes
Michael Wiebe (@michael_wiebe): @AaronBergman18 Time for a viral tweet on tiny apartments?
I just wanna live in a fish tank deep underground
Reply to @ExiledInfoHaz, 17:39 UTC · 2 likes
ExiledInfoHaz (@ExiledInfoHaz): This is virtue signalling. People don't actually care about others - if you look at the way they behave in the here-and-now as opposed to what they say they they would like to do.
In other words, follow concrete, near-term actions to see what's real, follow abstract statements to see what people want to project.
Virtues signaling is good when it’s aligned with actual virtue, which is infrequent but not vanishingly rare either
Tuesday, May 30, 2023
Reply to @EAheadlines, 11:14 UTC · 7 likes
EA Lifestyles (@EAheadlines): 😩
Aside from the “trying to avoid this conversation” point, seems good. Isn’t the whole point of EA as a *community* to influence others in a reasons-sensitive way like this? Hopefully in a kind+respectful way ofc
Replace “nuclear” with “climate” or even more so “vacuum collapse”
Thread, continued, 11:14 UTC · 2 likes
And it doesn’t seem as odd
Of course maybe you think the change was object level bad, but that’s not what I’m sensing is the objection
Reply to @frances__lorenz, 11:24 UTC · 5 likes
(The tweet being replied to by @frances__lorenz is no longer available.)
New meme template?
Reply to @absurdlymax, 11:26 UTC · 3 likes
Max Alexander (@absurdlymax): @interested_ea Can you still be in ea after that?
You don’t get to be an EA unless you’ve just had 2 diet cokes
Reply to @mealreplacer, 11:39 UTC · 1 like
(The tweet being replied to by @mealreplacer is no longer available.)
It does land on orange!
Quote tweet, 11:40 UTC · 15 likes
Quoting Frances Lorenz (@frances__lorenz): We're so back 😌
Quote tweet, 11:46 UTC · 2 likes
@open_phil @CentreforEA a first forum post is doing numbers I think we have ur next hire
Quoting Jesse Smith (@JesseTayRiver): First EA forum post! I argue that the HVAC workforce will significantly determine the success or failure of future IAQ interventions and is almost entirely overlooked.
https://forum.effectivealtruism.org/posts/q7dJz9ZaZGTSZL8Jk/obstacles-to-the-implementation-of-indoor-air-quality
Reply to @davidad, 13:46 UTC · 10 likes
davidad 🎇 (@davidad): https://safe.ai/statement-on-ai-risk
Need to become notable so I can sign the statement 😤
Reply, 14:49 UTC · 3 likes
Max Alexander (@absurdlymax): I think I like salad without dressing
Tweet, 15:49 UTC · 23 likes
Have you ever wanted the CEA YouTube feed but without video and in such a way that it keeps playing w/o keeping the app open, and also only the main EAG videos from 2023 onward, and in chronological order?
Then boy do I have a podcast feed for you
https://open.spotify.com/show/0lHqS3M08p1jGvBkRIVoFa?si=EFZAgW2bQj-hh-OX0lnrxQ
Thread, continued, 15:49 UTC · 1 like
https://podcasts.google.com/feed/aHR0cHM6Ly9hbmNob3IuZm0vcy9lMmEwNTczNC9wb2RjYXN0L3Jzcw
Thread, continued, 15:49 UTC · 2 likes
Apple Podcasts maybe coming soon, RSS feed url coming as soon as I can find it
Tweet, 21:53 UTC · 46 likes
Can’t wait for his next biannual tweet
Reply to @austinc3301, 22:21 UTC · 2 likes
Agus 🔸 (@austinc3301): @Laura_k_Duffy @absurdlymax I remember that when I was a kid I swore to myself that if I ever had kids I would let them all the sausages they wanted
Cultured sausages right right
Reply, 22:26 UTC · 2 likes
One thing I don’t understand is what power adults legally (or via other external power structures I suppose) have over their children
Like say you want to coerce ur kid into going to a playdate. Ok, what happens if they refuse? At some point between 3 and 17…
Thread, continued, 22:28 UTC · 1 like
They’re gonna have the physical ability to literally avoid this if they want to. What happens if ur kid is 9? What if “play date” is “school” instead? Can a parent withhold dessert as a coercive method? Dinner? Food for the next 3 days?
Thread, continued, 22:30 UTC · 1 like
At some point coercion has to ground out in either one party yielding or physical force, but I like never see this mentioned
Tweet, 22:59 UTC · 27 likes
ChatGPT Influencer posts share a common structure.
But most people don’t realize this.
Here are 9 things you need to know to make AI content that works for you:
Reply to @Laura_k_Duffy, 23:20 UTC
Laura Muresianu 🌐🔸🇺🇸 (@Laura_k_Duffy): Me yesterday: *buys 12-pack of Diet Coke*
Me this morning: *looks in fridge* Ugh only three Diet Cokes left??
Ok this is ~1g of phenylalanine, which is *definitely* an amount that implies subjective effects at least in most people and absent other protein sources
I wanna run an rct on Diet Coke vs placebo Diet Coke with sucralose or smth instead of aspartame
Reply to @QualyThe, 23:23 UTC · 2 likes
Qualy the lightbulb (@QualyThe): I just downed a whole pitcher of this and completed 30 grant evals in a single pomodoro. That's 7% less than usual 😤, I'll stick to the addreall thankyouverymuch
(quoting @thechosenberg: Can someone try this and report back)
Qualy no the levoamphetamine is a waste you want pure d-amph
Wednesday, May 31, 2023
Reply to @sweatystartup, 00:06 UTC · 1 like
Nick Huber (@sweatystartup): Thank god for HOA rules.
People wanting to grow veggies in their front yards, leaving trees untrimmed, not getting weed control treatments.
Parking dumb shit like trailers, boats and RVs in driveways. Doing work on cars in driveway.
I’ll never live in a neighborhood without the rules.
I wanna live exclusively among veggie gardens, boats, and fully evolved trees
Reply, 01:20 UTC · 4 likes
https://www.shrimpwelfareproject.org/donate
Quote tweet, 01:49 UTC · 32 likes
Haven’t stopped thinking about this…
Surely “convince surgeons/anesthesiologists to use actual general anesthetics during a 1x/decade procedure” can’t be THAT hard of a lift, can it?
Are there literally any medical or regulatory barriers here or is it purely a matter of norms?
Quoting Andy Masley (@AndyMasley): This uh seems bad. Is this real?
Thread, continued, 01:49 UTC · 13 likes
https://reducing-suffering.org/the-horror-of-suffering/
Reply to @thebestestpie, 01:55 UTC · 4 likes
(The tweet being replied to by @thebestestpie is no longer available.)
I mean personally I would take that increase in risk for colonoscopy and expect 99% would as well if informed in a frank manner about the costs and benefits
“Necessary” is a subjective judgement
Reply to @JMannhart, 01:58 UTC · 1 like
Jonathan Mannhart 🔎 (@JMannhart): Yeah, general anesthesia is really really unhealthy. But the norm should be that you get told the risks on both sides and get to make the decision yourself!
Even for ketamine, which is being used as an antidepressant (at sub-general doses) now? Is there a micromort estimate somewhere?
Reply to @JMannhart, 03:20 UTC · 2 likes
Jonathan Mannhart 🔎 (@JMannhart): @absurdlymax This oversells the problem's difficulty I think massively. The set is realistically like 8, maybe 6. (The probability that it's Toby Ord is 1/6.)
This oversells the problems difficulty massively. For any possible combination the answer is either “yes” or “no” so the probability of being right is 1/2
Reply to his own earlier tweet, 03:22 UTC · 9 likes
Aaron Bergman, May 31, 2023: https://reducing-suffering.org/the-horror-of-suffering/
Wait can we get some bioethicists on this? Unironically plz
Tweet, 03:45 UTC · 10 likes
Pasting the not-optimized-for-twitter text below
Thread, continued, 03:45 UTC · 3 likes
I believe that effective altruists should genuinely strive to practice effective altruism. By this, I mean that there are individuals who earnestly and seriously agree with the core arguments that animal welfare charities deserve significant financial support, both in relative
Thread, continued, 03:45 UTC · 2 likes
and absolute terms. However, they do not always follow through on these convictions when it comes to donations.
Many, for example, will eagerly nod along with introductory presentations for university effective altruism groups often highlight the fact that a tiny fraction of all
Tweet, 03:45 UTC · 3 likes
donations go toward animal welfare causes, even within EA.
And, as far as I can tell, very few if any EAs affirmatively dispute that animal welfare as a cause is simply more important and neglected, and similarly as tractable, as global poverty. But their donations do not seem
Thread, continued, 03:45 UTC · 3 likes
to reflect this, going to GiveWell-type charities like GiveDirectly or Against Malaria Foundation instead of animal welfare organizations.
While supporting poverty alleviation efforts is commendable in its own right – after all we want poor people having more money and fewer
Thread, continued, 03:45 UTC · 2 likes
dying from preventable diseases – it seems incongruous given their professed beliefs.
Without delving too deeply into speculation or psychoanalysis regarding individual motivations behind these donation choices; one possibility is simply an emotional preference for contributing
Thread, continued, 03:45 UTC · 2 likes
toward human-centric causes over those focused on animals' well-being.
To be clear: I am not claiming any personal moral superiority here; my own charitable giving record is awfully small in relative terms. Nonetheless I encourage fellow EAs who share concerns about factory
Thread, continued, 03:45 UTC · 2 likes
farming's abhorrent nature and have resources available for philanthropy to seriously consider allocating their donations toward animal welfare causes
Tweet, 04:37 UTC · 2 likes
Say what you will about La Sombrita
The name goes hard
Reply to @RichardHanania, 05:32 UTC · 2 likes
Richard Hanania (@RichardHanania): Russian spending on the Ukraine war in the context of other conflicts. The USSR spent 61% of GDP on World War II. The US spent 4.2% on Korea.
Russia is spending 3% on Ukraine, just above the US in Vietnam (2.3%). https://www.economist.com/graphic-detail/2023/05/30/how-much-is-russia-spending-on-its-invasion-of-ukraine
We spent 1% of GDP on Iraq? I would have guessed like 2 OOMs lower
Reply to @peterhartree, 11:33 UTC · 2 likes
Peter Hartree (@peterhartree): Love it! I've been waiting for this.
Curious: how did you make it? Is it automated?
I've been planning to do a bunch of “get audio from YouTube and put it on a podcast feed” in June.
Nah, everything manual except downloading the audio which I could do in bulk thanks to chatGPT written python
The uploading and copy and pasting descriptions took like 15 mins tho
Reply to @pearl_please, 11:36 UTC · 2 likes
(The tweet being replied to by @pearl_please is no longer available.)
Planes
Reply to @absurdlymax, 11:43 UTC · 3 likes
Max Alexander (@absurdlymax): @austinc3301 I imagine it not being okay would break the profession
Then the profession should be broken
Reply to @absurdlymax, 11:48 UTC · 3 likes
Max Alexander (@absurdlymax): @AaronBergman18 @austinc3301 That probably doesn't maximize utility
Idk you’d have to think people are super irrational to justify having surgeons effectively force suffering on others by withholding obviously important information
Like at a bare minimum all this should be made clear to the patient beforehand!
Reply to @AISafetyMemes, 12:31 UTC · 6 likes
AI Notkilleveryoneism Memes ⏸️ (@AISafetyMemes): Thread summary:
AGI companies: we have no idea how these things work
reply guys: yes we do
(quoting @robertskmiles: I think most people (quite reasonably) think "We built ChatGPT, so we must basically understand how it works" This is not true at all. Humans did not build ChatGPT. In a way it would be closer to say we 'grew' it. We have basically no idea how it does what it does.)
*Reply guys: yes you do
Reply to @SpacedOutMatt, 12:43 UTC · 2 likes
Matt 🔸 (@SpacedOutMatt): @AndyMasley @AaronBergman18 @thebestestpie I think part of the story here is that doctors use different techniques for unsedated patients, at least in some cases
Swear I was about to reply “I bet they do fully conscious/memorable ones differently”
Tweet, 12:44 UTC · 4 likes
Taylor Swift is overrated
(but should still be highly rated)
Reply, 13:31 UTC · 2 likes
I agree it’s not necessarily a bug, but in this particular case I think the reasons for expecting intuition to be faulty and object level (philosophical and empirical) evidence for animal welfare are strong enough that it *is* a bug
Reply to @eshear, 13:42 UTC · 1 like
Emmett Shear (@eshear): It’s easy to trick ourselves into ideas that aren’t true through abstract reasoning in general. And “shift your resources away from local suffering towards universal suffering” is a very common trap for our conscious minds. Our subconsciouses resist this, a good habit to keep.
This carries some weight to be sure, but not infinitely much. After all the very same argument would imply that you should donate to local charities over Against Malaria or whatever if you had any intuition towards the former
Reply to @esceydabulyu, 18:54 UTC · 2 likes
Lark (@esceydabulyu): @AaronBergman18 @MarkovMagnifico I got her number after weeks of not running into her in class.
Tweet, 18:58 UTC · 5 likes
The shards of the Overton window have de-glassified into sand
Reply to @QualyThe, 19:00 UTC · 2 likes
Qualy the lightbulb (@QualyThe): finish your drafts people!
(quoting @daniel_271828: Now I’m wondering what to do with my half-completed blog post on “many serious AI researchers and forecasters are concerned about AI takeover”)
No😡
Reply to @austinc3301, 19:29 UTC · 3 likes
Agus 🔸 (@austinc3301): The starring article is a piece by @KatjaGrace?!
this is huge!
(quoting @TIME: TIME's new cover: The case for taking AI seriously as a threat to humanity https://ti.me/3oHNGd2)
The shards of the Overton window have fallen into a black hole
Tweet, 21:49 UTC · 17 likes
Wordcell confirmed
Also flashbacks to the MAP-R in literally 3rd grade
Thread, continued, 21:49 UTC · 2 likes
https://www.arealme.com/vocabulary-size-test/en/
Reply to @SpacedOutMatt, 21:59 UTC · 2 likes
Matt 🔸 (@SpacedOutMatt): Get rekt
List the 600 words right now or ur fake
Reply to @AndyMasley, 22:05 UTC · 2 likes
Andy Masley (@AndyMasley): Bad news fam
I shan’t recover
Reply to his own earlier tweet, 22:15 UTC · 12 likes
Aaron Bergman, May 31, 2023: https://www.arealme.com/vocabulary-size-test/en/
Oh boy I really set myself up for all the higher score screenshot replies lmao
Reply to @absurdlymax, 22:20 UTC · 2 likes
Max Alexander (@absurdlymax): @SpacedOutMatt @AaronBergman18 This feels wrong I won't lie
It feels *extambracious
Quote tweet, 22:40 UTC · 83 likes
When you forget that on EA Twitter everything gets shifted 4 sd’s to the left
Quoting Aaron Bergman 🔍 (@AaronBergman18): Wordcell confirmed
Also flashbacks to the MAP-R in literally 3rd grade
Reply to @absurdlymax, 22:49 UTC · 17 likes
Max Alexander (@absurdlymax): I think this is the real reason everyone has imposter syndrome
I think u should transfer some imposter syndrome to me and then we’ll both be better calibrated
Reply to @absurdlymax, 22:53 UTC · 2 likes
Max Alexander (@absurdlymax): @AaronBergman18 I do have imposter syndrome about being a person
That sux
Reply to @mattyglesias, 23:35 UTC · 16 likes
Matthew Yglesias (@mattyglesias): It would be very funny for Taylor Swift to just date a totally random average 35 year-old dude with impeccable politics — “it’s what the fans deserve.”
So you’re saying I have a chance
End of May 2023. Previous month: April 2023. Next month: June 2023. Back to the archive index.