Might be nice to test for asymmetry in "opposites" like likely vs unlikely. Would have been nice to include words like plausible and implausible because there seems to be a load of asymmetry in their widely perceived meanings (not-unlikely vs highly-unlikely, respectively).
Anyways, was triggered by the suicidal ideation thing. Besides having a bit of personal experience with this, I know of no solid evidence that LLMs are either decreasing or increasing teen suicide rates. If forced to bet on it, I'd guess there are better than even odds that they would reduce rates.
..Surely you and everyone you know has received a mountain of bad advice from people you have loved or trusted. There are mountains more of bad advice available from the traditional internet. I'm skeptical of claims that bullshitting/biased LLMs are worse advice-givers than those we had before.
....How much better would any rando LLMs advice need to be to make it a preferred source of information relative to that of your bigoted parents, or your best friend; the stoner down the road who preached that the world was ending in 2012, and that a decades-long bender was obviously the way to go..
Maybe some people who are all-in on this never had any real world parents, guardians and friends that spoke endless shit and bad advice into their ears. IMHO poor parenting and destructive peer pressure are way bigger real word problems than any of this bad-LLM info/advice stuff...
In the US (and elsewhere) religion, patriotism and xenophobia have become so entwined that it is impossible for many to even judge what is/is not fascistic. How else could devout Christians think that Jesus himself would also vote for an angry, selfish, greedy, spiteful, misogynistic, bully party.
Did the ASM ever properly apologize for licking Trump's boots after all those anti DEI executive orders? If so does anyone have a link to the apology?
Over 90% of scientists are not native English speakers so maybe this is the most pertinent point "use of LLMs [..] reduces barriers for non-native English speakers." Offline this is the primary reason why many colleagues tend to be pro their use. Online here and elsewhere this barely ever comes up.
Same here
Arxiv is so......Cannot see where File S1 is located. Any help would be appreciated.
...CPU level parelallization when traversing through the pairs of sequences will work too if you do NOT do this
for x = 1 to sequence number -1
for y = x+1 to sequence number
next y
next x
but rather do this
for z= 1 to number of pairs
x=pairs(z,0)
y= pairs(z,1)
next z
The point is that presorting is FAST and scales linearly with sequence numbers/alignment length as it takes just one full traverse of the alignment. Unsure if adding pairwise differences across elements of the encoded subalignment arrays will be auto-vectorized by the compiler; but they might be?...
...you presort the alignment columns to obtain a subalignment with just dimorphic sites, a subalignment with trimorphic sites, etc and work out matrices of pairwise differences for each subalignment separately and then at the end calc the overall pairwise distances using these matrices....
Yes sure - for alignment columns where there are many different possible nucleotides it becomes less viable but instances where the vast majority of variable alignment columns have only 2 observed states these can have a binary encoding and you can do 10 columns at a time if......
You'll survive.
Forgive me if you have already done that.
eg in the 2 state alignment you can encode 10 nucleotides as a number between 1 and 1024 - you then have a 1024 x 1024 matrix where you have precomputed the distance between all possible 10mer pairs and just read of the distances from the matrix. You can get a >5 fold speedup.
OK cool - and you calculate hamming distances? Or something more fancy? Reason I'm asking is if its hamming dustances you can get some very nice speedups if you sort the snp columns into separate 1,2,3,and 4 state alignments, encode the sequences in the alignments.
Cool - does this use multiple sequence alignments or pairwise alignments?
Thanks - great recommendation
For me, some great 2025 songs are:
Tropical Fuck Storm's Goon show (fave lyric: "Its a golden age of arseholes")
Jeff Tweedie's Feel Free ("Feel free to to never listen and always speak, to never learn and try to teach")
Big Thief's Incomprehensible ("Let gravity be my sculptor")
For me, muting the shame-mongers and and LLM-bothering bores (both pro and anti) works quite well.
I'm guessing the line is wherever you choose to make it. It is irrelevant what anyone else thinks of your choice when it is the precious minutes of your life that are being spent hunting for information.
In a nutshell within a nutshell: broaden your experience if you want to increase your creativity.
A hypothesis/point of view that I find appealing at a personal level (meaning its probably bullshit). In a nutshell its this: the breadth/range of your problem-solving experience/training hones an analogizing skill which makes it easier for you to solve superficially unrelated new problems.
Sort of supports David Epstein's "Range" hypothesis: en.wikipedia.org/wiki/Range:_...
Did you maybe post a link to the wrong video? In this one he only mentions Intubation at ~0:30 and at this point he definitely doesn't say anything about doctors killing patients by intubation. Thought maybe you accidentally linked to it because the video's heading is so misleading.
I don't know. I would guess though that, yes, an important feature of life is processes that trade energy to spatially control local reductions in entropy over time. Its a really big question though that I'm too stupid to answer properly.
This is the most interesting thing I've seen in ages. If intra-cellular membrane-less hydrogen ion gradients are biologically meaningful/useful things then so too might be membrane-less gradients of everything else - from bigger ions with discrete point-sources through protein complexes.
Seems like the BBC is going all in on a self-abusive resolution to make 2025 their shittest year ever. WTF is going on over there? Are they alight? Its like seeing a once beloved and often hilariously witty uncle turn into a stupidly boring and hateful racist after they have a stroke.
