--- Log opened Tue Apr 25 00:00:42 2023
00:05 < hprmbridge> nmz787> I'm pretty sure a professor of mine talked about some sort of subtractive DNA diff, something like immobilize one genome on a solid surface, or on beads, then just add the second DNA set and warm and cool to encourage hybridization, then wash the surface or remove the beads
00:05 < hprmbridge> nmz787> Phenotypic diff seems like the stuff of fairy tales
00:53 -!- flooded [~flooded@146.70.202.142] has joined #hplusroadmap
00:57 -!- test__ [flooded@gateway/vpn/protonvpn/flood/x-43489060] has quit [Ping timeout: 246 seconds]
00:58 < fenn> .t https://twitter.com/YaBoyFathoM/status/1649103596930187290
00:58 < EmmyNoether> Yudkowsky abandons alignment research 🙀
01:58 < alethkit> https://atomicsemi.com/careers/?ashby_jid=c991fb4d-e634-42f2-b392-798117326b88
01:58 < alethkit> markets are truly something else
02:04 -!- Netsplit *.net <-> *.split quits: caveman, pharonix71
04:08 -!- yashgaroth [~ffffffff@c-73-147-55-120.hsd1.va.comcast.net] has joined #hplusroadmap
04:35 < kanzure> re: chemical diff; we need a general library of DNA manipulation tools like string manipulation functions in programming. append, prepend, slice, delete, insert, join, split, index, find, findall, diff, foreach, etc.
04:35 < kanzure> eventually we will have these tools to manipulate individual DNA molecules.
04:41 < hprmbridge> msnewgooty> What’s the way you think they’ll manifest? Through biological implementation or something more mems like
04:42 < kanzure> hopefully something with a nanopore like a casette player but given how things have gone so far, probably separate bulk scale reactions
04:44 < kanzure> there have been some interesting papers in the past on physically manipulating DNA molecules
04:44 < kanzure> like stretching them out, using physical force with an AFM tip to unwind the double helix, and another one was something like a nuclease linked to a metal bead and the bead was moved to different regions of a stretched DNA molecule to cut it.
04:50 < kanzure> msnewgooty: i want something like ed boyden's expansion microscopy but for DNA enlargement (an XNA with large nucleobases and backbones that can be more easily physically manipulated but also chemically converted back to regular DNA)
04:51 < kanzure> probably a chemistry whiz could do this, don't think anyone has tried
05:03 < hprmbridge> msnewgooty> There was a company trying to use an XNA like that for nanopore
05:03 < hprmbridge> msnewgooty> It would be easier to read
05:23 < hprmbridge> msnewgooty> I think it was a church co? Maybe round 2012
06:30 -!- caveman [~caveman@gateway/tor-sasl/caveman] has joined #hplusroadmap
06:32 -!- pharonix71 [~pharonix7@user/pharonix71] has joined #hplusroadmap
07:02 < hprmbridge> jkhales> They captured their personalities so well
07:05 < hprmbridge> jkhales> Are there any recent updates on this floating around?
07:05 < hprmbridge> jkhales>
07:05 < hprmbridge> jkhales> I don't understand why this is so underfunded.
07:05 < hprmbridge> kanzure> thymus rejuvenation?
07:20 -!- caveman [~caveman@gateway/tor-sasl/caveman] has quit [Remote host closed the connection]
07:20 -!- caveman [~caveman@gateway/tor-sasl/caveman] has joined #hplusroadmap
07:32 < hprmbridge> yashgaroth> I don't get why fahy is trying to frame it as thymus-mediated rejuvenation, most of the results just seem like they're from growth hormone acting elsewhere in the body. Co-opting the thymus (or engineering thymic organoids) would be great for enabling a lot of therapies if we could use them to induce immune modulation, but that's separate from what he's doing
07:33 < hprmbridge> Perry> You really can't come up with any theory on why he might frame it that way?
07:34 < hprmbridge> Perry> None whatsoever? Not even the slightest idea?
07:34 < hprmbridge> yashgaroth> hahahaha ok fair
07:36 < hprmbridge> kanzure> lots of words (even too many, not sure I like this page- I don't think you can argue people into liking progress and innovation with words?)  https://www.tenentrepreneurs.org/operation-innovation-1#4
07:41 < hprmbridge> docl> I wonder if you can take a planar structure (maybe not graphene but something molecular thickness) and add a bitmap like arrangement of bases. QR code might be the better analogy. Then I wonder if these would seek out complementary structures in a solution. Not sure what the advantage is, but maybe useful for positioning things deterministically
07:41 < hprmbridge> Perry> Have you ever taken organic chemistry?
07:42 < hprmbridge> kanzure> there were ways of organizing reactants on DNA origami scaffolds.
07:42 < hprmbridge> docl> No, I'm reading a molecular biology text now though.
07:42 < hprmbridge> Perry> Take an organic chemistry class.
07:42 < hprmbridge> docl> I definitely plan to.
07:54 < L29Ah> 16:05:35]<hprmbridge> jkhales>
07:54 < L29Ah> i wonder what's this supposed to mean
07:59 < docl> discord shows message could not be loaded. perhaps replying to something that has fallen out of scope
08:03 < L29Ah> apparently kanzure and yashgaroth could load it
08:04  * L29Ah feels like discord is switching to "extend, extinguish" mode
08:07 < hprmbridge> msnewgooty> DNA origami...what promise it had. I was super excited about it maybe a decade ago. We could make molecularly precise metamaterials and do super-resolution positioning. Kind of weird that nothing much has come from it? Maybe Dees will get his nanopores to production quality.
08:08 < yashgaroth> I think it did that because jk's discord message had a newline in it, you didn't miss anything
08:41 < kanzure> there was "DNA nanoscopy" https://pubmed.ncbi.nlm.nih.gov/28947733/ but maybe can't be claimed as a DNA origami technique.
08:53 < hprmbridge> msnewgooty> Not widely used, for a couple reasons
09:31 -!- cthlolo [~lorogue@77.33.23.154] has joined #hplusroadmap
10:02 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has left #hplusroadmap []
10:21 -!- test_ [~flooded@89.45.4.3] has joined #hplusroadmap
10:24 -!- flooded [~flooded@146.70.202.142] has quit [Ping timeout: 255 seconds]
10:25 -!- test_ is now known as _flood
10:28 < caveman> what's this channel all about?
10:30 < caveman> someone from ##physics told me to join this place with a smile ':)'.  not sure what did he really mean.
10:31 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
10:33 < docl> transhumanism, and all the science we need to get there
10:35 < caveman> what is the generally advocated thesis in this channel towards transhumanism?
10:37 < docl> we generally want to improve the human condition via technology. too much philosophical pontificating is frowned on, but if you have a project or idea for one we'd love to hear it
10:38 < caveman> what does 'technology' mean?
10:41 < docl> biotech, nanotech, whatever it takes
10:41 < caveman> is it bio/nano-tech to put seed in soil, then pour water on it?
10:41 < docl> space stuff is relevant too, as is self replicating machines of all sizes.
10:42 < docl> eh, from a technical standpoint yeah but not the focus here
10:42 < caveman> is the ultimate goal maximising survival of humans?
10:44 < docl> talk of maximizing anything not really in vibe. life extension is common sense though. AI safety talk probably better suited elsewhere. (I'm a yudkowsky fan, but here isn't the place for it, majority are probably anti-fans)
10:46 < caveman> ok.  will check him on youtube.  but 1st i have to see why is jordan peterson crying again (just saw a video thumbnail).
10:49 < docl> have you considered adderall? just switched to focalin myself
10:50 < caveman> what does it do?
10:50 < caveman> as far as stuff that i swallow in my mouth go, i only want to insert: water, 9 amino acids, 2 fatty acids, 13 vitamins, 15 minerals (16 if you count cobalt; but we already did in b12), choline, creatine, fibre.
10:58 < docl> I'll take any edge I can get if it's low risk enough. the basics / nutrition are important though. focalin is dexmethylphenidate, so the right handed molecule from ritalin. I was on adderall a month ago, but it was keeping me up late. before that was vyvanse but the $300 copay didn't sit right with me.
11:00 < caveman> what does it do?
11:00 < hprmbridge> jaisel> curious about this; how do you guys structure your time to do research outside of your daily jobs/work?
11:00 < caveman> for a healthy human.
11:01 < caveman> jaisel:  it's difficult, but i have an hour of time allocated per week to plan my time, then i force myself into it.  it's difficult to follow/stick-to.  basically, you have to become a robot.  it's a nice journey.  i'm still undertaking it.
11:02 < hprmbridge> jaisel> gotcha, are you getting 6-8 hours of sleep still or do you sacrifice sleep for time to do work?
11:03 < caveman> i don't think that i get 8 hours.  maybe 6 or 5.  but my sleep is screwed up.
11:03 < kanzure> none of this is really research... research requires work.
11:04 < caveman> docl: what do those drugs do?
11:05 < docl> stimulants. are you camilliar with coffee?
11:06 < caveman> yes.  but not sure about its impact.  i don't notice anything from it.
11:06 < caveman> maybe a mild sense of being waken up?  not sure if it is significant enough to require to consume it.  maybe the /real/ solution is for me to man-up and just work irrespective of feeling tired.
11:08 < caveman> i don't mind consuming stuff, but i find it that, almost always, the simplest solution is to just do the right thing, and ignore the lazy feelings.  this happens by having the right kind of thought in the head.
11:09 < caveman> the right thought, in the right head, at the right time, can make all the difference  :3.
11:10 < docl> hmm. so these are drugs for treating adhd. everyone's brain works a little differently, but personally it's hard to pick what to focus on (although I can and do become absurdly focused by most people's standards). compulsory high focus tasks like paperwork are weirdly hard. they get easier on commonly prescribed adhd drugs.
11:11 < caveman> have you tried to just do them, without a drug, and accept to suffer pain?
11:12 < docl> sure. hard to stay motivated though.
11:13 < caveman> true.  motivation is a problem.  here, i found religion to help.
11:17 < docl> what, like threaten yourself with eternal suffering if you don't fill out the tax form
11:18 < docl> way easier to just do the low risk commonly prescribed drugs IMO
11:19 < caveman> lel!  no.  when it comes to filling out tax forms, it's the opposite:  threaten yourself with eternal suffering if you don't do your best to end taxation, artificially inflating fiats, and all other forms of theft.
11:19 < kanzure> this is boring me
11:19 < kanzure> why are you here again?
11:21 < hprmbridge> msnewgooty> Fortunately, or unfortunately, this is my full time job
11:21 < caveman> you work as a thief?
11:22 < hprmbridge> jaisel> haha thats awesome tbh. Im stll in school so balancing time is quite hard sometimes because i want to learn everything i can
11:23 < kanzure> caveman: please consider lurking for a while and see what hplusroadmap personnel get up to
11:23 < kanzure> or i can mute you i guess
11:23 < hprmbridge> msnewgooty> If there are opportunities for research in labs, I would go that route and seek them out. Mentors are essential
11:24 < caveman> kanzure: please suck a bag of dicks, and shove your 'mute' up your sorry ass.
11:24 -!- mode/#hplusroadmap [+o kanzure] by ChanServ
11:24 -!- mode/#hplusroadmap [+q caveman!*@*] by kanzure
11:25 < hprmbridge> glenn> caveman you are a looser
11:25 <@kanzure> alright maybe caveman will get a better sense for what we talk about after a few days of mute
11:25 < hprmbridge> kanzure> The bot isn't relaying the mute action, apparently.
11:26 -!- mode/#hplusroadmap [-o kanzure] by kanzure
11:26 -!- caveman [~caveman@gateway/tor-sasl/caveman] has left #hplusroadmap []
11:26 < hprmbridge> jaisel> yep! I’ll try and lurk and also find some good mentors to gain insight into how to move forward 🙂
11:28 -!- mode/#hplusroadmap [+o kanzure] by ChanServ
11:28 -!- mode/#hplusroadmap [-q caveman!*@*] by kanzure
11:29 -!- mode/#hplusroadmap [-o kanzure] by kanzure
11:29 -!- cthlolo [~lorogue@77.33.23.154] has quit [Read error: Connection reset by peer]
11:42 -!- Chiester [~Chiester@user/Chiester] has quit [Read error: Connection reset by peer]
11:45 -!- Chiester [~Chiester@user/Chiester] has joined #hplusroadmap
12:11 < kanzure> tom knight seems to think DNAScript does something more than 80mers? https://twitter.com/TomKnightSynBio/status/1650903738448502785
12:26 < hprmbridge> docl> would that be the syntex process in this video? https://www.dnascript.com/technology/
12:32 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has left #hplusroadmap []
12:43 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
12:46 -!- Malvolio [~Malvolio@idlerpg/player/Malvolio] has joined #hplusroadmap
13:17 < fenn> L29Ah, docl: the blank message from jkhales was literally a blank line. the previous two messages were in reply to the links i posted about the lex fridman/yudkowsky lampoon and fahy thymus rejuvenation. i'm not sure what is the best way to translate reply-to-message context; matrix includes a large chunk of the message you're replying to and it gets spammy real quick
13:18 < fenn> the matrix bridge i mean
13:20 < L29Ah> fenn: ideally one would insert a >quote when the quoted message is far away in the log, otherwise nothing
13:20 < L29Ah> thanks
13:23 -!- pharonix71 [~pharonix7@user/pharonix71] has quit [Remote host closed the connection]
13:23 -!- _flood [~flooded@89.45.4.3] has quit [Ping timeout: 265 seconds]
13:26 < fenn> re "I don't think you can argue people into liking progress and innovation with words?" ... people used to like progress and innovation, and now they don't. this is probably a direct result of soviet ideological warfare, the consequences of which most people are unaware and wouldn't believe even if shown. my own experience with changing my worldview seems to be mediated half by peer pressure and
13:26 < fenn> half by biochemistry perhaps due to diet (mental state i.e. depression/hypothymia/irritability/hypomania)
13:28 < fenn> i remember distinctly realizing that maybe the world wasn't a giant pile of conspiracies and oppression, and maybe there really were just a few people actually working really hard in universities and startups to try to make things better, but nobody was joining them in the effort because we're all so distracted
13:29 < fenn> after having gone on a raw vegan diet with lots of nutritional yeast for a few weeks
13:32 < fenn> this was in december 2006
13:37 < fenn> perhaps off topic, but this is where my "ideological warfare" meme comes from: http://esr.ibiblio.org/?p=260
13:42 < L29Ah> > this is probably a direct result of soviet ideological warfare
13:42 < L29Ah> as a soviet-born i must say that USSR ideology is a lot about techno-optimism; not sure what kind of memes it projected outwards tho
13:44 < fenn> yes the contrast is shocking
13:44 < fenn> now "westerners" are mocked for believing in crazy ideas
13:45 < fenn> and everybody just forgot
13:46 < L29Ah> FUD seems like a great strategy to stop humans from doing anything, at all times
13:46 < fenn> https://www.youtube.com/watch?v=9QgIdnFi8qc
14:04 < muurkha> soviet ideology was very much in favor of progress and innovation, wasn't it?
14:06 < fenn> internally
14:06 < muurkha> right, I see
14:07 < docl> same as it's approach to expansionism?
14:08 < muurkha> I'm not sure these memes can be entirely attributed to Comintern :)
14:09 < fenn> i'm tempted to throw up my hands and say it's lost to the mists of time and the fog of war and therefore unfalsifiable, but actually there are records in KGB archives of exactly what they did, it's just a lot of work to dig it all up and put the pieces together
14:09 < muurkha> I mean, Comintern tried to promote an awful lot of memes, and was pretty unsuccessful at most of them, especially the ones that contradicted internal Soviet messaging
14:10 < muurkha> Neat, is there a detailed unpacking of this somewhere?
14:10 < fenn> i don't know
14:10 -!- Malvolio [~Malvolio@idlerpg/player/Malvolio] has quit []
14:11 < muurkha> Ideas like "For a virtuous person, violence and war are never justified. It is always better to be a victim than to fight, or even to defend oneself. But ‘oppressed’ people are allowed to use violence anyway; they are merely reflecting the evil of their oppressors" are not just not Communist innovations but actually millennia old
14:11 < fenn> from where
14:12 < muurkha> well, the first part is a core belief of Buddhism and also shows up in the Gospels
14:13 < fenn> pardon me if i don't just believe you. this sounds nothing like roman or medieval understanding of morality, and i can't reconcile buddhism advocating violence even by the oppressed
14:13 < muurkha> I agree that it sounds nothing like roman or medieval understanding of morality,
14:13 < muurkha> and I did say 'the first poart'
14:14 < muurkha> except without the o
14:16 < muurkha> and the second part is just the right to self-defense, which has been a central part of at least European understandings of violence for at least 800 years, and is implicit in the Gospel as the commonly accepted point of view Jesus was said to reject with his "turn the other cheek" stuff
14:17 < fenn> you're missing the point, which is that it's a double standard
14:17 < fenn> you can't chop it into halves and say "this half looks good, and this half looks good, what's the problem?" because the problem is that it's not self-consistent
14:17 < muurkha> has there ever been a time when people didn't interpret morality through self-serving double standards?
14:18 < fenn> ok i'm disengaging because there is no point to this
14:18 < muurkha> I wouldn't be surprised if you could find defenses of the right of self-defense going back several millennia, though it gets harder to untangle in very martial societies which don't consider certain kinds of violence to require any justification
14:19 < muurkha> but nearly all societies have heavily sanctioned some kinds of homicide, so it should be possible to find something
14:22 < docl> we live in a heavily engineered (or evolved) multipolar system. people in a weaker position getting special considerations is part of that. used to be less of that in the world, with kings peasants and such
14:24 < kanzure> boring, bring me the heads of 95 year old emeritus professors to study
14:43 < kanzure> "The second of Chargaff's rules is that the composition of DNA varies from one species to another, in particular in the relative amounts of A, G, T, and C bases. Such evidence of molecular diversity, which had been presumed absent from DNA, made DNA a more credible candidate for the genetic material than protein."
15:02 < alethkit> fenn: I mean, part of the USSRs exports was technical education to non-aligned states
15:03 < alethkit> And considering the Soyuz-Apollo program, I would say the lack of progress is entirely self inflicted by the Americans
15:05 < alethkit> kanzure: https://yewtu.be/watch?v=av_0PhJdw9M
15:06 < docl> https://en.wikipedia.org/wiki/Chargaff%27s_rules
15:07 < docl> is that second rule distinct from the second parity rule? "that both Α% ≈ Τ% and G% ≈ C% are valid for each of the two DNA strands"
15:09 < docl> first parity rule (that A:T and C:G = 1:1) turned out to be caused by base pairs
15:13 < docl> second parity rule is weirder, as it applies to single strands (and is approximate). doesn't apply to mitochondria
15:16 < hprmbridge> docl> (from a textbook): "The eminent nucleic acid biochemist Chargaff’s famous ratios—that the amount of adenine in a DNA sample matched that of thymine, and the amount of cytosine matched that of guanine—were later understood in the context of Watson and Crick’s DNA double helix structure. Perhaps frustrated that he had never come up with base pairs himself, he became a bitter critic of molecular
15:16 < hprmbridge> docl> biology, an occupation he described as “essentially the practice of biochemistry without a license.”"
15:16 < yashgaroth> there's the bit about base-pairing causing A=T and G=C ratios, but also most organisms' genomes tend toward having a 50/50 ratio of AT/GC (with many exceptions because biology)
15:17 < yashgaroth> oh wait that appears to be talking about a separate thing
15:17 < docl> is kanzure talking about the same second rule I'm looking at?
15:18 -!- flooded [flooded@gateway/vpn/protonvpn/flood/x-43489060] has joined #hplusroadmap
15:23 < yashgaroth> I don't know where the quote is from, so I couldn't say. Ignore the stuff about https://en.wikipedia.org/wiki/GC-content since it just adds separate complexity
15:27 < docl> what I'm thinking is, if you pick one side of each pair, say A and G, you get a ratio that is basically randomly distributed, making each species tend to be distinct. with lots of exceptions to "random" because biology, perhaps. but that overall variation in ratio was one of the data points that helped discover that DNA encodes the information, back in the day.
15:27 -!- TMM_ [hp@amanda.tmm.cx] has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
15:28 -!- TMM_ [hp@amanda.tmm.cx] has joined #hplusroadmap
15:28 < hprmbridge> docl> oh, and the RNA molecule doesn't use thymine but uracil, so that was probably a confusing bit
15:34 < yashgaroth> yeah different ratios of bases in different organsism helped indicate that they encoded information
15:37 < fenn> imagine taking chomsky seriously on artificial intelligence, when he utters phrases like, "Unlike humans, for example, who are endowed with a universal grammar that limits the languages we can learn to those with a certain kind of almost mathematical elegance, these programs learn humanly possible and humanly impossible languages with equal facility."
15:37 < fenn> and considers it a fault
15:40 < fenn> chomsky seemed to be under the impression that the goal of the field of AI was to help linguists
15:42 < fenn> annoyed that i can't search twitter anymore
15:43 < hprmbridge> kanzure> is there any redeeming value of chomsky at all?
15:46 < fenn> teaching people how to be skeptical of mass media
15:46 < hprmbridge> kanzure> did it work?
15:47 < fenn> the overton window is moving away from mass media driven narratives, but not because of chomsky
15:49 < muurkha> how do we know that these programs can learn humanly impossible languages?
15:50 < fenn> but now we have manufactured consent through government and corporate meddling in social media algorithms, which is basically the same thing. i don't know if chomsky had any role in fostering skepticism of social media content
15:53 < muurkha> chomsky's greatest achievement was demolishing behaviorism
15:53 < fenn> i had a really cringey AI quote by chomsky but now i can't find it, so i had to dig up something else
15:53 < muurkha> not that what followed behaviorism was scientific yet
15:57 < fenn> "Chomsky rejects the radical behaviorist psychology of B. F. Skinner, who viewed speech, thought, and all behavior as a completely learned product of the interactions between organisms and their environments. ... Chomsky has maintained that syntactic knowledge is at least partially inborn."
15:57 < fenn> "Chomsky's nativist theory was ultimately rejected by the mainstream child language acquisition research community owing to its inconsistency with research evidence."
15:58 < fenn> obviously he'd have a beef with a "blank" AI that has no formal grammar encoded into it, demonstrating superhuman levels of language skill
15:58 < muurkha> yup
16:11 -!- codaraxis [~codaraxis@user/codaraxis] has joined #hplusroadmap
16:39 -!- test_ [~flooded@146.70.202.67] has joined #hplusroadmap
16:43 -!- flooded [flooded@gateway/vpn/protonvpn/flood/x-43489060] has quit [Ping timeout: 265 seconds]
16:51 -!- codaraxis [~codaraxis@user/codaraxis] has quit [Ping timeout: 252 seconds]
18:17 -!- codaraxis [~codaraxis@user/codaraxis] has joined #hplusroadmap
19:33 -!- yashgaroth [~ffffffff@c-73-147-55-120.hsd1.va.comcast.net] has quit [Quit: Leaving]
19:49 -!- flooded [~flooded@146.70.174.243] has joined #hplusroadmap
19:52 -!- test_ [~flooded@146.70.202.67] has quit [Ping timeout: 240 seconds]
19:53 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has quit [Ping timeout: 248 seconds]
21:19 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has quit [Ping timeout: 248 seconds]
22:44 < alethkit> fenn: I mean, is it not?
22:44 < alethkit> part of the interesting thing about AI for me is that it would help us understand how we process language
22:45 < alethkit> (and yes, me no likey LLM goin brrr)
22:57 < fenn> no, AI was not made for linguists exclusively
22:58 -!- faceface_ [~faceface@user/faceface] has quit [Remote host closed the connection]
22:59 < fenn> it's so bizarre he would even think that in the first place
23:02 < alethkit> Really?
23:02 < alethkit> I mean, Hofstadter also has a point
23:03 < alethkit> (replace linguists with linguists/cognitive scientists, because we're trying to figure out how *we* work by having a clean implementation)
23:04 < alethkit> and besides, LLMs have bad reliability guarantees
23:31 < muurkha> yeah, I think the way that LLMs are most interesting is linguistics
23:33 < muurkha> it's still unclear how much they will teach us about, say, physics or post-traumatic stress disorder, because so far they kind of suck at logic and emotional intelligence
23:33 < muurkha> but they're flabbergastingly powerful at language
23:34 < alethkit> good luck with that
23:34 < alethkit> because i'm not touching the cesspit that is llm training at all
23:36 < muurkha> moreover, they are in some sense fundamentally linguistic entities; to the extent they generalize to areas like logic, math, physics, etc., I think it strongly reinforces the hypothesis that those higher aspects of human cognition are in some sense implemented linguistically, or can be
23:42 -!- codaraxis__ [~codaraxis@user/codaraxis] has joined #hplusroadmap
23:43 < alethkit> muurkha: funnily enough, there's a few papers that show it's not the case
23:46 < fenn> the impressive thing is not that it does logic, math, etc. but that it can reason about common sense principles
23:46 -!- codaraxis [~codaraxis@user/codaraxis] has quit [Ping timeout: 260 seconds]
23:47 < fenn> e.g. "how is a beach ball like a howitzer"
23:47 < fenn> "they're both inflatable"
23:47 < fenn> most humans wouldn't be able to come up with that
23:48 < fenn> computers have always been good at formal reasoning, but until now did very poorly with fuzzy concepts
23:51 < fenn> i think LLMs will get better at "emotional intelligence" once people start caring about that
23:51 < fenn> the raw capability is there, to the extent that emotions can be transmitted via text at all
23:57 < fenn> so this is a thing: https://costplusdrugs.com/medications/
23:59 < fenn> manufacturing cost + 15% + $8 S&H
--- Log closed Wed Apr 26 00:00:20 2023