@alth0u neocortex seems to be a bunch of independent models and then a coordination layer. at least according to the pop science. That’s ok I’ll still keep reading that garbage
@alth0u I wonder if LTH is actually viable in the brain
seems less likely due to extreme sparsity but it probably makes up for it in sheer neuron volume
@alth0u yeah some say we do a lot of synaptic pruning during sleep so that’s something
thus leading to “dreams = regularization / domain randomization” thesis
but it’s hard to know what to prune of course
the covid killed more than all the wars thing isn’t very compelling because it’s clear most human value systems consider violent death at the hands of others worse than “natural” death
@TheTyIerDurden yeah people act like QALYs are some kind of heartless modernism despite the fact that nearly every society would prefer its elders dying to its youths dying
@eigenrobot@cauchyfriend i think they should just get rid of the GS payscales and make hiring good talent easier
sucks but i dont see a future in which CDC or FDA can be privatized
@sgodofsk@eigenrobot@cauchyfriend yeah and its the higher ups who we count on to be good capital allocators / set wages for lower positions etc
pay them more and then give them more discretion / fewer veto points
@lolitsnotreal_@cauchyfriend@eigenrobot like if you literally just scroll down a market cap weighted list of industries to disrupt
the top would be like "software" then maybe finance (of all kinds), healthcare, and next oil & gas
@lolitsnotreal_@cauchyfriend@eigenrobot software is obviously on top of the software game, finance was among the first disrupted (eg paypal back in 1999 onwards to stuff like high frequency trading), healthcare is notorious hard to innovate in (but we do it anyway) but i don't know of any major O&G techcos
yall pray for my friend, he's living in a license-permit raj by the name of "California" that requires a city inspector to come by after installing a new garage opener
on the contrary becoming larger has given me exposure to some really excellent people and i don’t see the benefits of scale stopping anytime soon https://t.co/oFzGJO7BOp
@eigenrobot idk I think it’s at least reasonable considering they will take the flak in the end for any “dangerous” outputs
an organization that believes itself to have an important mission should be self defensive
@eigenrobot it’s not really about AI safety so much as it is about, it’s not worth risking our mission and funding streams for the sake of “Samantha” or whatever else unimportant chat bot usecase
@eigenrobot consider that these project December people were literally trying to raise the dead via AI necromancy
not hard to imagine what kind of psychological impact that could have on its users or what might happen if it tells some fragile person to kill themselves
@liminal_warmth@eigenrobot how is it unethical to make cheap ad copy? if anything I think it’s super low risk and not very fraught
i think we’re in agreement on their motives, just disagreeing on whether self defense is worthy goal or not
@liminal_warmth@eigenrobot but it’s very much not a normal business tbh
you can see that because they added this API on as an afterthought to their R&D program, probably hired zero enterprise salespeople and don’t really want to deal with the drudgery of making money at all
@_Sowellsista_@terfdavidicke too complicated is a euphemism for “not worth it” lol
germania was impossibly complicated but the Romans kept sending their vitality there to die because they wanted it so bad
@telmudic one of the worst psyops led by the enemies of tech is that “personal data” is of any value at all
nobody really cares about your genome or your dumb convo with your friend, pls
@boop that’s good to hear! honestly it made it more fun for me
so cool to be interested in something, ask the timeline, and immediately the relevant experts come and tell me about it
@lolitsnotreal_@LudditeHacker@cauchyfriend@eigenrobot I think you literally have to just start a new company in this area to break into it, which will require (1) tech branded ceo to get venture funding, (2) deep industry knowledge about oil and gas (3) high initial capital investment
it’s clear why it hasn’t happened yet
@lolitsnotreal_@LudditeHacker@cauchyfriend@eigenrobot the boomers that run these cos don’t care enough to optimize, and tech people are ideologically disgusted by oil and gas - so not surprising that this is the state it’s in
a lot of my friends have a criticism of flexport that it’s basically a “set of dashboards”
as ML folks they’re not impressed by the algorithms or lack thereof
but I guess it’s funny that that’s not really the value add is
@notlinear1 yeah
to be fair, the dashboard brain level introduces crazy step change of effiency
then later the stage where you can apply advanced statistics introduces a new level
the latter is more satisfying and "easier" than the first
@MegaBasedChad people hate salesforce because the product is basically violent to its users, not for lack of advanced tech (they have a crazy good ML division)
@alth0u@kelp_feeder i think the thing is that someone doing 100,000s of TEUs can already access pretty good logistics
but the SMBs see the greatest value add from flexport, etc, so they pay the most for the product
there’s a limited amount of compute substrate in the world
say AI is 1000x better than humans at portfolio modeling, but only 10x better at elder care. Then it makes no sense for AI to do elder care, and nursing home workers will be making infinite money through profitable trade
basic Ricardian trade theory premise:
if we achieved super intelligent AI that’s better than us at everything, there would still be opportunities for humans to work due to comparative advantage https://x.com/nominalthoughts/status/1421928620000432128
@YosarianTwo from all viewpoints, whether you're looking at it from the perspective of tech productivity growth, or ricardian trade, it should increase the income of humans not decrease
people didn’t even read the new regs before they started clowning them lol
they include such unreasonable statutes as “pigs should have enough space to turn around in their enclosure” https://x.com/Noahpinion/status/1421993817096425473
@metorotem i mean, of course hog farmers will say that
I have no idea whether it’s true or not but them claiming it makes no difference
worst case pork prices in CA skyrocket and new facilities are built to meet demand
@sanssomatic@aquariusacquah I’ve been hoarding Queen Elizabeth funko pops since prince Philip died. they will be worth billions in worthless fiat “GBP”
@Rationalbot i don’t really think it escapes notice at all. it’s incredibly politically fraught to be like “I’m gonna increase the price of hamburgers” or gas or whatever
it’s actually a sign of high state capacity when you can manage it (to meaningful ends)
@halvorz is any of it real? clearly my revealed preference is “no” because I haven’t forced my parents onto this stuff. but was hopeful about metformin
@Rationalbot oh I thought we were discussing lab meat vs boutique luxury farmed pork
but yeah I don’t think a 10% increase in breeding sow living space will result in the poors resorting to porridge diets, especially not in california
@Rationalbot consider how much standards have increased in factory farming even in the last 50 years, seems doubtful
as societies get richer they can afford less cruelty. I wouldn’t try to apply the same standards in India
@judahwantssocks maybe but these games are competitive and therefore may take an unbounded amount of compute
separate ASIs trying to beat each other at the portfolio game
@storebrandguy@lisatomic5 it’s probably a cowardly atomized way to view community but avoid/ get rid of them seems like the only option after “increasingly aggressive ribbing”
@akbirthko@halvorz from what i can tell the mouse thing is real but i dont know of any more complicated model organisms they tried it in
sinclair is like the 10th author on that paper
@terfdavidicke yeah agreed but its also clear that boys have a harder time sitting still for hours and focusing on average
resulting in a large number of them getting dosed into submission by pharma companies so they can compete
not sure if thats a good outcome
@terfdavidicke theres probably better ways for them to learn whereas im not sure you could shrink the asian-hispanic education gap by teaching them slightly differently
@theashelina yeah i agree with drunkroon. one of those extremely fake sounding truisms, im sure plenty of great writers are doing it with an audience in mind
@nickcammarata seems like there very much are degrees of required delusion and therefore phenotypes of founders
takes much bigger balls to be like “we will colonize Mars” than “we will make a food delivery app”
@Cererean yeah it’s offpatent, but
1) biologics don’t get generics, they get “biosimilars” which are way more expensive to certify
2) everybody wants the latest formulation of insulin because they’re slightly better
@orb_net@jeff82874662 Messiah is somewhat hard to read
the pacing sucks and the tone dramatically shifts but it’s actually way more meaningful to herberts theme
@monsterofcook correct, the themes are pretty reactionary
social darwinism + “hard times create strong men” + accelerationism
but it’s like an ironic meditation on all this because nobody can really say if the jihad was worth it
@halvorz i dunno though sometimes I’d rather have notable accomplishments even if no legible “deep thoughts”
surely the intricacies of thought it takes to build one of these things is more important than being able to sound smart and in touch on twitter
@Willyintheworld@halvorz no doubt about “societally useful” but i guess more pertinent if you’d willingly choose that lifestyle at great cost to yourself
@eigenrobot@R1Jack@GaltJhonnyRides@sgodofsk@ded_ruckus@bufordsharkley even relative nonparticipants like facebook or GM wait to roll out new product launches or announcements during the weeks before an election
all things considered i think delaying the vaccine announcement by 2 weeks can be forgiven after delivering miracle cures
@R1Jack@eigenrobot i think the latter
honestly it was an absurd result. it didnt even have to take into account brain volume or shape. literally just the pattern of vasculature is enough to distinguish self reported categories with 99% accuracy
@BroductManager@antoniogm creating a marketplace is always preferable to reselling/flipping
which makes me think someone probably already tried the marketplace idea and didn't succeed (or there's some regulatory barrier)
Every great jihad happens only once. The next Khalid Sheikh Mohammed won't fly a plane into the World Trade Center. If you are copying these guys, you aren’t learning from them.
@jane_gatsby in the computer assisted proof category, the Four color theorem is a super famous one (and I think also the first one)
as far as AI goes, maybe @rickyflows would know?
actually surprised at the naivety of
“this civilization peaked hundreds of years ago and so has pretty classical monuments and is therefore morally superior to my civilization which instead peaked 30 years ago and continues to build things in newer styles”
@powerbottomdad1 anyway i think its good advice to make despondent young people not lose hope but very bad advice to pour water on someones fire "theres always tomorrow haha"
compounding is real and the the best start early
i guess the real thing is that it feels like people enabling or encouraging procrastination
the energy isnt like "you're young, try new things and fail its fine" its often like "you're young, just go ahead and waste your time being self destructive its fine"
@samson_hu@SukhaniShri same my last year of college was like this lol. I was like I gotta party like I never have before bc it’s gonna get harder soon. And I was right
@MegaBasedChad@jeff82874662@Hellachans i think you mean it’s not specifically a financial asset because if you have a winery you’re probably gonna report your casks of wine on the balance sheet
@PrinceVogel well ill fully admit im status seeking (its right there in my bio! “in search of legacy”)
but im also genuinely high decoupling
do we have different meanings of that phrase?
@PrinceVogel there’s a few people I really dislike on here but it’s more for their vibe than their ideas
plenty of friends who I think are ridiculously dangerously wrong but I enjoy anyway
@karatalaamalaka hmm i would argue its way easier to bullshit software than other engineering disciplines
in fact the industry norm is to ship broken stuff and fix it in prod
@ctjlewis yep. options is an cash cow + the cultural caché among younger people means it can grow in a million directions
i give it 25% chance that i'm banking on robinhood in 5 years
is interesting but i dont understand doing a bio of highly dynamic people while they're still alive and have many productive years left
even jobs only agreed to talk to isaacson after he was sure he was dying https://x.com/elonmusk/status/1423104512479334402
@writingbree@StevenWillcox2 yeah fr he rehearsed speeches, had a signature look
whereas its clear elon gets prepped on his demos like 10 minutes before the demo
all tech journos would be out of business if it werent for their full time career comparing random things to Juicero.
so really it was one of the most effective investments of $120m ever made
when im feeling blackpilled on tech i just need to go find someone talking about juicero. makes me revert to the most hypertechno-optimist bubblegum pop retrofuture outlook. its therapeutic
for a long time the reputation of tesla engineering was bad bc the software side kinda sucked (relative to other players) and that’s what tech people are trained to look at. but clearly the other engineering divisions were running on high gear
@bayes_baes yeah i mean working you to the bone is orthogonal to good engineering right? and side note I don’t think any of them are complaining after their stock appreciation lol
click on links for 10 years of your life and you end up like me. a ghost of a man, training models on one screen and clicking terrible links on the other
@eigenrobot sales is very specifically a job that they know nobody cares about. there is nothing fun about it, just doggedly chasing people down who mostly hate you. thats why all the energy is in the commissions / money-making. mini wolf of wall street in every sales dept
@eigenrobot engineering on the other hand has lots of people geeking out over cool new toys. i get where you're coming from but i really dont feel right when i think my job's not interesting or impactful in some way
@SatyreContraire dont know if its bad engineering to have some quality issues when you perform an impossible ramp up
if they had a slow ramp up and also quality issues then i might call that bad eng
@ollyrobot im not trying to put him in or out of any bucket tbh I don’t care
i just think the fact that he read some blogs doesn’t make the blog responsible for everything he does especially when the blog might have endorsed the opposite
@ollyrobot@LudditeHacker@effendleton maybe. but you could probably categorize this guy in several ways, the least legible of which to the average Brit would be “rationalist”. even interpreting more favorably to you he would still be one politician among tens of thousands
@ollyrobot@LudditeHacker@effendleton long story short, better things to fixate on at least politically
the energy on here is more like “zealotry of recent convert”
@LudditeHacker@effendleton@ollyrobot I would argue all this is completely unimportant relative to the influence of rationalist thought on the AI community especially OpenAI
@LudditeHacker@effendleton@ollyrobot yeah i think the direction of technological progress is far more historically important than the vagaries of individual politicians. but i digress
@krishnanrohit@LudditeHacker@effendleton@ollyrobot disagree, I think it’s actually far from the broader AI push — you will rarely find Facebook or google people talking about “agi xrisk”. They don’t even take the concept seriously. Rather they are concerned about introducing model bias
@krishnanrohit@LudditeHacker@effendleton@ollyrobot well I am detracting from my own argument at the top eg that rationalist community is irrelevant. agreed on these two points. Re:xrisk they popularized it
my favorite period of history is the Finno-Korean hyperwar. not very much writing on this subject even though it claimed 10^8 times more lives than WWII. sad!
theory
being away from the center of the empire is good for heterodox thought / having different intellectual fashions
India is "crown jewel" so the colonial elites are making enough money to fund their side ventures https://x.com/TenreiroDaniel/status/1423390362366849027
wonder to what degree access to the arxiv firehose is making everyone hang out with brilliant people constantly
when i was researching i acutely felt the urge to google the vast trove of literature to solve various components of my problem instead of thinking through it myself https://x.com/erf_of_why/status/1423399763416784896
@wannabegroncho being a journo, even a crooked one, is a rare and hard skill lol
- not many people with that level of moral "flexibility"
- have to be really good at twitter bait
- have to really stick to a point and drive it home over and over
@halvorz but anyway i guess my point is that the school doesn't matter as much anymore considering digitization of research, constant firehose of papers in your general area
@halvorz there are no "cultural islands" anymore, even at <creationist school>
i'm not exactly sure how research worked in the past but i'm guessing it relied more heavily on physical journal subscriptions