Construction Capo (@TechCapo) header
Construction Capo (@TechCapo) profile picture

construction brother (former technology brother)

Joined December 2021
Photo by Construction Capo (@TechCapo)Photo by Construction Capo (@TechCapo)Photo by Construction Capo (@TechCapo)Photo by Construction Capo (@TechCapo)Photo by Construction Capo (@TechCapo)Photo by Construction Capo (@TechCapo)Photo by Construction Capo (@TechCapo)Photo by Construction Capo (@TechCapo)Photo by Construction Capo (@TechCapo)Photo by Construction Capo (@TechCapo)
Replying to @reasonisfun
yeah this is one of my favourites. they're all hyperengaging tho, i lost a few nights of sleep watching Claudia but weirdly anti-memetic in that she brainwashed me into spending more time outside and with friends and family. wholesome stuff and fuel for optimism
1
48
u have good opinions, good opinion haver. i still love the masses tho, u gotta love the masses
33
great points tbh. they still feel like exceptions, even tho stories about individual geniuses with strong taste and vision dominate our attention. the collective deserves more credit, i blame individualism. something feels wrong with the lone genius vs unwashed masses frame
1
1
20
i'd push back on the claim that it has nothing to do with the audence's response, artists without audiences starve. for every Van Gogh, there are thousands of deluded artists who think they are misubderstood geniuses, and some artists have legacy while being popular. idk feels like we glorify the lone genius and don't give enough credit to audiences as an integral part of the co-creation process
1
20
make sure u finish on the Bach, not Debussy
6
true. i think inventing a new musical genre is not hard, getting humans to like it is the tricky part lol. my point is genres are co-created with audiences and that human rapport is what's tricky to do mechanically. this is a different problem than LLMs not taking leaps
1
14
yes this is where the analogy breaks down, the reward is jazzy sound vs mindblowing party. whats common is the iterative process driven by reward signals
26
y'all framing it as a pro-human vs pro-AI debate and since OP is pro-human and i offer a counter-hypothetical i must be anti-human. my point indeed claims u can't invent jazz without human audiences. the OP claimes AI can't invent jazz in a vacuum. i agree. note, however, that individual humans can't invent jazz in a vacuum either. so i posit the active ingredient in creation is not the creativity of the creator but the taste of the audience. taste makes us unique in the universe, not creativity
1
29
true, partially. as an artist tho u still must earn applause from someone otherwise ur a nuisence thinking ur a misunderstood genius. i agree that applause from a handful of sophisticated taste-makers is often a better signal than applause from the unwashed masses
25
maybe he is not aware of such study, i certainly ain't. but we should do this experiment. @suno?
30
how would anyone really?
1
93
you can ask grok to translate sentences to basic english. here you go "Yes. No one person did it. A group of people worked on it together. They changed it again and again."
36
i'm not sure really, we've seen examples of both
421
can you get human artists to invent genres that don't exist yet? yes, obviously. how do they do it? one answer is that new stuff comes from the human soul and machines can't do it because they don't have human souls. another answer is you can mix existing stuff and see if people like it. we don't really know which one is true so we come up with thought experiments, like Goertzel in OP. my counter-thought-experiment is imagine an immortal J. S. Bach locked in a room for 2000 years. can he come up with jazz? no, obviously, you needed the 1900s zeitgeist to make jazz. so Goertzel's point is not really fair and i propose that new genres come from audiences/zeitgeit/environment/culture/tastes, not individual human souls
1
71
it's a hypothetical, the point is if 1900s AI had the opportunity to play with live audiences and observe their feedback (like the jazz pioneers had), it could come up with jazz, probably. the reward function trained on modern jazz is a proxy for this in lieu of a time machine
2
1
566
yeah this is true but the artist's ability to decode these signals and respond on-the-fly feels easier to replicate mechanistically (emotion detection has been going quite well even before transformers) compared to producing these signals. like, how do you build a robot that dances like a human would if they heard wham-pebedoobapp-de-whampemm-boh for the first time? so seeing that we're losing our edge as creators, i start to think our real edge is judgment. taste is the final moat indeed
21
i made the argument illegible on purpose to separate the wheat from the chaff :D just kidding but to be fair, the analogy between a reward function trained on modern jazz and the zeitgeist of the early 1900s does break down when you consider that the former is a static thing built in hindsight, while the latter is a living thing co-evolving with the creative process, the taste of the audience-as-reward-function is shaped by the inventions of new jazzy concepts i still like the analogy because it shares the credit between the creators and the audiences, and this helps ease my existential terror about AI taking over the creative process. creation is not what's uniquely valuable about humans, taste is
1
26
not really financial but probably a game theory thing. the jazz pioneers were poor in money rich in taste and taste drives status teapawt.org/mhjb/status/2123172699… "taste is the final moat" as slopcore pioneers say and i think it might be true. attributing all the credit to jazz artists and zero to the taste of their audiences warps our sense of human worth, focuses it on creativity, and that's what triggered everybody here. i crossed the sacred line drawn around human creativity. AI might take over creation but we are the taste-makers and so we drive discovery. slopcore is getting good because there are tasteful directors behind the wheels
Replying to @RichDecibels
New things emerging… endless possibilities at every moment; but you need high openness social network A noticing what might be good, trying it out, elevating what seems best, and if A has sufficient status, wider social network B might risk trying the new thing out
1
40
ikea deprecated this. literal rug pull
2
40