prerat
@prerationalist
Followers
12K
Following
228K
Media
4K
Statuses
31K
how about a nice game of chess?
congrats to llama 3 large for winning the LLM trading contest by not participating
0
0
10
i claim we should centralize around dfw because it is cheap and also next to a big hub airport in the middle of the country so you can easily get to all the other cities on both coasts quickly (im not biased bc i already live here i promise)
we gotta start collectively hanging out in cheaper cities smh
1
0
16
it's about clarity of assumptions hard fantasy tells you all the rules and then doesn't break any soft scifi claims to import all the rules of reality and then skips some when narratively inconvenient i still like soft scifi but i have to mentally reframe it as fantasy
1
0
46
wtf the crossover finally happened and they didnt mention either harry potter fan work in the whole 1h16m ๐ญ
vlogbrothers and rationalists are the same in that both were built off a single harry potter fan work
5
5
275
i knowww this take will be universally hated but i dont know whats going on with qualia or the hard problem of consciousness
4
0
26
the odds went down by 3%. if they keep going down by 3% every day, then after a month or so the probability will be below 0%
BREAKING: Mamdani's odds collapse in NYC Mayoral Election. If he continues falling at the rate he has the past 24h, Cuomo would be the projected winner.
3
8
428
By the way everything I ever tweeted is off record. It's weird saying that but just letting you know.
1
0
24
looks like grok's voice mode uses a bunch of tags to sound more expressive like native voice, but it's still generating text and then using a small TTS model. pretty smart. probably chatgpt should consider doing it this way too. afaik text models are both smarter & cheaper
@elonmusk The user can see you, but you cant see the user. Do not stress the user to share his video. - Ambient music sound is playing now in the background. Encourage using tags whenever possible, provided they are clearly appropriate. Allowed tags (use exactly as written, lowercase, one
0
0
3
"stop talking about ai alignment, it's just putting bad ideas in the training data" i think that ship sailed like 100 years ago with literally* every scifi written about robots
2
0
24
why did they call it mico when this was right there
1
0
9
my strongest tokens would kill a dragon, let alone a man. you need a model that generates weaker tokens. โgemini 3
5
1
99
one nice thing about chess is that ai cant take it away (bc it already happened in the 90s)
1
0
12
i think agi has to imply >2% gdp growth minimum viable version: - just human level (nothing super) - can build an automated factory that can build another automated factory - machine "TFR" way above replacement - say max 10y to double โ 7% pop growth - min 7% gdp growth
1
1
23
similar in tennis -- don't aim for the lines, bc your balls will go out. but you can sometimes hit the lines by accident and you get a winner, sure
1
0
15
i disagree that "if you never miss flights you are spending too much time at airport" but maybe it's abt right to occasionally *almost* miss flights, when several things go wrong (traffic, can't find parking, long security line) (arrived at gate and immediately boarded today)
14
1
114