l8doku

659 posts

l8doku banner
l8doku

l8doku

@l8doku

Katılım Ekim 2011
88 Takip Edilen15 Takipçiler
l8doku
l8doku@l8doku·
@browserdotsys Threaded comments + sorting by user votes = eventually reddit. Alternatives: one chronological thread (4chan), a mess (twitter), limiting size (invites? can't post much?) or scope (lesswrong? doesn't seem to work with HN)
English
0
0
2
165
l8doku
l8doku@l8doku·
Моё лучшее решение в твиттере - перейти на анонимный аккаунт. Сразу так легко стало - могу где хочу отвечать, что хочу писать. Хоть меня с этого аккаунта читают/знают всего пара человек, всё равно казалось, что это всё "на виду".
Русский
0
0
0
29
l8doku
l8doku@l8doku·
@punk6529 Some did. A couple months ago I saw a comment chain where an artist said "I've heard they solved the environment thing, so I've been more interested in the NFTs lately"
English
0
0
0
22
6529
6529@punk6529·
so now that ETH is not responsible for "destroying the environment" have all the people who were previously concerned about this started minting NFTs? 😀
English
111
41
675
93.8K
l8doku
l8doku@l8doku·
@seetheisolate @mattvsjapan A good fictional example that shows how a language can be optimized is dath ilan by Eliezer Yudkowsky. There are frequent mentions how complex terms from game theory, decision theory, epistemology would be really useful as short words.
English
0
0
0
55
hyperbolic jelq chamber
hyperbolic jelq chamber@seetheisolate·
@mattvsjapan what do u think these languages would be optimising “for”, exactly? given that all existing languages seem to transfer information at very similar rates it’s probably indicative of some human limit to our language processing speeds AI can’t overcome
English
2
0
3
904
Matt vs Japan
Matt vs Japan@mattvsjapan·
AI is going bring never-before popularity to conlangs Soon people will use AI to generate conlangs that are perfectly optimized for human communication Communities will form around people who acquire these "optimal conlangs" by consuming comprehensible input generated by AI
English
8
1
66
12.4K
l8doku
l8doku@l8doku·
@nearcyan Can the AI misrepresent the friend?
English
0
0
0
129
near
near@nearcyan·
you find out that a friend has been responding to some of your messages with an AI model, rather than writing the responses himself you still enjoyed yourself, but now it feels different knowing that he isn't actually behind some of the messages how does this make you feel?
English
31
1
43
12K
l8doku
l8doku@l8doku·
@DoqxaScott I read it as "I see how many points are correct, but the issue is complicated and not all points commonly associated with this idea follow from the correct ones", which may sound either condescending or precise.
English
0
0
1
56
Scott Hamilton
Scott Hamilton@DoqxaScott·
Are you annoyed when hearing a conversation, and one person says to the other "I'm sympathetic to that" [idea or point of argument], rather than "I agree" ? I find it disingenuous.
English
3
0
1
1.7K
l8doku
l8doku@l8doku·
@prerat n / m < 0.75 n < 0.75m (n+1) / (m+1) > 0.75 n+1 > 0.75(m+1) n > 0.75m - 0.25 0.75m - 0.25 < n < 0.75m 3m - 1 < 4n < 3m -1 < 4n - 3m < 0 n and m are integers, so 4n-3m is an integer. There are no integers which fit "-1<x<0".
English
0
0
1
6
l8doku
l8doku@l8doku·
@prerat I wrote a script to test numbers of likes from 10 to 10000, it hits 75% every time. Not sure how to prove it though.
English
1
0
1
15
prerat
prerat@prerat·
A tweet has a likes to views ratio of 70% a few minutes after posting, but it increases to 80% in the next few minutes. Is it necessarily true that it was exactly 75% at some point? Why or why not?
English
16
0
24
6.1K
l8doku
l8doku@l8doku·
@bananamelter Hypocrites are harder to spot, so they require a stronger and more public signals. When someone openly does a wrong thing it's easier to think "others will see it's wrong, so I don't have to signal it".
English
0
0
3
37
l8doku
l8doku@l8doku·
@Insect_Song @robinhanson For food - I want there to be some push towards trying new kinds of food, so I don't get stuck eating the same thing all the time.
English
1
0
1
0
Robin Hanson
Robin Hanson@robinhanson·
Would you prefer eating a new kind of food INCREASE or DECREASE your perceived value of eating that food again?
English
8
1
1
0
l8doku
l8doku@l8doku·
@ESYudkowsky An example about art: horror stories of Japanese animators being paid very poorly. Art is "overpriced", so it's natural that the price moves down, and the organized effort to support animators has to fight against this pressure.
English
0
0
1
0
l8doku
l8doku@l8doku·
@ESYudkowsky 1. Barriers preventing the market from being efficient - people not negotiating enough, not moving to better positions. 2. As others here said, price floor below which it doesn't make sense to charge money, power laws.
English
1
0
2
0
l8doku
l8doku@l8doku·
Я сделал себе котацу! Раньше думал оно почти бесполезное - всего лишь ноги греть. Оказалось, котацу - это прекрасно.
Русский
0
0
0
0
l8doku
l8doku@l8doku·
@mechanical_monk I once tried a "write a tiny blog entry every day for a year" challenge and was very surprised when I ran out of thoughts in the first couple months
English
0
0
5
0
monk
monk@mechanical_monk·
i'm still in awe that someone could have 100 distinct thoughts in total, let alone on a single topic
English
11
4
125
0
l8doku
l8doku@l8doku·
@magicianbrain Art has more of the artist's personality. You can fine-tune a model to copy someone, it's like you're stealing their work, their soul. It's subjective, you can argue "it will never be real". Most code is a means to an end, it's universal, there is almost no style.
English
0
0
2
0
brainlet
brainlet@magicianbrain·
why is it that artists are so much more hostile to AI making art than programmers are to AI writing code
English
886
285
4.5K
0
l8doku
l8doku@l8doku·
Кажется, до меня наконец дошло, как сценарий "ИИ нас всех превратит в скрепки" может быть правдоподобным. Обычно примеры, про те же скрепки, были слишком абстрактные. С ChatGPT проще видеть более конкретные.
Русский
0
0
0
0
l8doku
l8doku@l8doku·
Один из моих первых двух рынков на Manifold Markets хорошо идёт - manifold.markets/l8doku/will-a-… Вопрос задал размыто, но для первого раза ничего
Русский
0
0
0
0
Nick
Nick@nickcammarata·
If you had to try and 10x your life satisfaction in a year, what would you do?
English
126
28
472
0
l8doku
l8doku@l8doku·
Втянулся окончательно. Там очень хорошая геймификация, помогает пробить самый начальный барьер, после которого уже интересно.
Русский
0
0
0
0
l8doku
l8doku@l8doku·
Подсел на manifold.markets Раньше туда заходил, и сразу страшно - всё уже идеально настроено, все такие умные. А теперь чуть разобрался, и сам что-то выбираю
Русский
1
0
0
0