|
idonotlikepeas posted:Clay Bennett There ya go Bennett! A good one!
|
# ? Feb 5, 2024 15:04 |
|
|
# ? Jun 3, 2024 10:32 |
|
Oh, that Branco is just begging to be edited
|
# ? Feb 5, 2024 15:18 |
|
Alhazred posted:I'm just gonna wait until you've googled what Japan did during WWII. Yes, I am aware, but I would posit that it is still stupid for an Asian American to be a neo-nazi. Not that he necessarily is, but he seems to be inching ever closer.
|
# ? Feb 5, 2024 15:34 |
|
World Famous W posted:what the hell utah, who wants to be represented by gulls?! You will not be surprised to discover it's a Mormon thing.
|
# ? Feb 5, 2024 17:14 |
|
So, this is actually a very fascinating topic! To put things very, very simply a lot of "AI" scripts use multiple nodes that are activated to varying degrees by different inputs. There are much more inputs it needs to handle than nodes, so different end concepts are essentially encoded in complex combinations of nodes that may seem unrelated at first. This makes it very hard to diagnose why an AI model output the precise response it did to a given input, which has some significant implications for training AI to do what we actually want it to do (such as not saying "kill all black people" when asked "how do we reduce crime" just because its training data set included racist websites). One way around this is to use much more powerful but also much simpler AI system to analyze the "main" system, and identify what causes it to output false or misleading information. This is made much more complicated because our current AI systems aren't people! They can't actually lie, because there's no internal world view or personality. BUT! Analyses have consistently found that AI systems like Chat GPT have specific nodes that activate when they are making up false information (even THAT isn't quite accurate, because there's no mind to "make up" information. Our language is VERY BAD at talking about what AI actually does!) so there's some sense in which the system "knows" it is lyi- oh wait, you're just a hack trying to make a dumb cartoon about AI. Carry on
|
# ? Feb 5, 2024 19:37 |
|
HootTheOwl posted:There ya go Bennett! A good one! Every so often, the old Bennett manages to emerge and remind us what he's made of.
|
# ? Feb 5, 2024 19:51 |
Kaza42 posted:So, this is actually a very fascinating topic! To put things very, very simply a lot of "AI" scripts use multiple nodes that are activated to varying degrees by different inputs. There are much more inputs it needs to handle than nodes, so different end concepts are essentially encoded in complex combinations of nodes that may seem unrelated at first. This makes it very hard to diagnose why an AI model output the precise response it did to a given input, which has some significant implications for training AI to do what we actually want it to do (such as not saying "kill all black people" when asked "how do we reduce crime" just because its training data set included racist websites). One way around this is to use much more powerful but also much simpler AI system to analyze the "main" system, and identify what causes it to output false or misleading information. Do you have a recommended source on these functions? I'd like to read more.
|
|
# ? Feb 5, 2024 19:56 |
|
Skios posted:Michael Ramirez There we have it folks, one of the few instances where a label would not be unnecessary on a Ramirez. WTF is this horseshit all about? I think his Biden is doing the old trope of lady scared of a mouse jumping up on a chair, but we have no idea why.
|
# ? Feb 5, 2024 20:19 |
|
Guardian: "Ella Baron on US and British airstrikes on Iranian-backed militias – The Pentagon is continuing its attacks in reprisal for the killing of three US soldiers at a military base on the Jordan-Syria border" Telegraph: Army fails to meet recruitment targets at all training centres for five years After Lady Butler. Times: Evening Standard: O’Neill expects poll on Irish unity to take place in next decade; Love Island: When does the All Stars series finish and is there a Casa Amor twist this year?
|
# ? Feb 5, 2024 20:58 |
|
logger posted:There we have it folks, one of the few instances where a label would not be unnecessary on a Ramirez. Something about ladies on chairs being inherently weak and fearful, which is shameful, and therefore Biden is shamed by the comparison? But it doesn't work well because the jumping on a chair is a reaction and there's nothing here he's being afraid of. He's just afraid for no reason and in a dress to emphasise how weak and fearful is.
|
# ? Feb 5, 2024 21:29 |
|
Pope: Labor promises to save you $1000 and more choice of EVs. It could also fuel a Dutton attack (SMH). Knight: Rowe: Australian academic Yang Hengjun given suspended death sentence by Chinese court (Guardian). Wilcox: Leak, Son of Leak: Albanese was called a "handsome boy" by Chinese Premier Li Qiang.
|
# ? Feb 5, 2024 21:40 |
|
Discendo Vox posted:Do you have a recommended source on these functions? I'd like to read more. Here is a primer on the concept of AI systems using complex interactions of nodes to "store" concepts. This is more specifically about AI's "lying" and trying to build systems that detect it. Usual caveats apply: a lot of people who write about and discuss AI topics are weird, and I am not specifically endorsing other posts on this blog.
|
# ? Feb 5, 2024 23:17 |
Kaza42 posted:Here is a primer on the concept of AI systems using complex interactions of nodes to "store" concepts. Thanks, those are as disappointing (and the comments as pathetic) as I expected. I'll use some of this to burst some executives' bubbles in the near future.
|
|
# ? Feb 5, 2024 23:39 |
|
Kaza42 posted:not saying "kill all black people" when asked "how do we reduce crime" Oh, so now AIs can't get jobs as political cartoonists?! I think we all see who the REAL bigot is here.
|
# ? Feb 6, 2024 00:29 |
|
idonotlikepeas posted:Amjad Rasmi The middle east is actually stable and can't blow up no matter how many people try. An interesting take.
|
# ? Feb 6, 2024 01:28 |
|
Adam Zyglis Amjad Rasmi Bill Bramhall Dave Whamond Jeff Danziger Jeff Koterba Kevin "KAL" Kallaugher Matt Wuerker Mike Peters Paul Berge Randall Enos Rivers Walt Handelsman Ward Sutton This is from 2020, but he thought it was relevant again.
|
# ? Feb 6, 2024 01:59 |
|
https://twitter.com/Lubchansky/status/1754524085789344074
|
# ? Feb 6, 2024 02:46 |
|
What?
|
# ? Feb 6, 2024 05:16 |
rodbeard posted:What? It’s pretty straightforward, Mattie opposes the legalization of cannabis.
|
|
# ? Feb 6, 2024 05:19 |
|
I can't I disagree with anything in this because I have no clue what on Earth Lub is trying to say. Kellies Nomination: Most Incomprehensible Vincent Van Goatse fucked around with this message at 05:33 on Feb 6, 2024 |
# ? Feb 6, 2024 05:27 |
|
Very clearly a comic about "Genocide" Joe Biden and Voting Blue No Matter Who IMO
|
# ? Feb 6, 2024 05:29 |
|
Youremother posted:Very clearly a comic about "Genocide" Joe Biden and Voting Blue No Matter Who IMO The sad thing is you're probably right.
|
# ? Feb 6, 2024 05:34 |
|
Neito posted:Kellies Nomination: Most easily reinterpreted. I had to double check the comic to make sure I understood what was going on, that's not sending the message you think it is, Branco. If you’ve ever used the term “reverse racism,” it means you don’t have a problem with racism so long as certain conditions are met.
|
# ? Feb 6, 2024 05:35 |
|
Youremother posted:Very clearly a comic about "Genocide" Joe Biden and Voting Blue No Matter Who IMO Seems likely. Biden is frequently cited as the "harm reduction candidate."
|
# ? Feb 6, 2024 05:44 |
|
rodbeard posted:What?
|
# ? Feb 6, 2024 05:46 |
|
Youremother posted:Very clearly a comic about "Genocide" Joe Biden and Voting Blue No Matter Who IMO It's definitely this. Lubchansky has done many cartoons on this theme before, usually with the message "voting is pointless, might as well not bother anyway". "Harm reduction" is the term used to describe a specific voting strategy where you vote for the least bad candidate in the general election to minimize the damage that will be done to the world by that person. Obviously, there's a lot of debate (here and elsewhere) on that topic and its morality and effectiveness. On the one hand, I know my own opinion is that the harm reduction strategy is the correct one, and so I want to excoriate Lubchansky here. On the other hand, I can empathize with their feelings on the topic; it's exhausting and demoralizing to be asked to vote for the less bad candidate over and over again. It's hard to get upset with people who are just depressed about the world they have to live in, although I do hope they come around and go for harm reduction in the longer term, since some of that harm will actually be done to people I love if Trump is elected in 2024.
|
# ? Feb 6, 2024 06:07 |
|
It is also demoralizing because there is an upper limit to the number of times we can do this dance where they say "vote for us to prevent horrible things," and then the horrible thing happens anyway, and they go "well just vote harder." I distinctly recall, right after Roe was struck down, they said "vote to protect abortion rights" and the first response was "WE DID." (USER WAS PUT ON PROBATION FOR THIS POST)
|
# ? Feb 6, 2024 06:41 |
|
Is it bad that one of my first thoughts about this comic was, "drat, that dude has a thicc rear end"?
|
# ? Feb 6, 2024 08:04 |
|
Harm reduction/lesser-evilism is an admission that we have already lost and there is no hope of things ever getting better. Things will forever get worse, more evil and more harmful, and the only choice we get is rate of increase. And then the people espousing it seem baffled that that makes people depressed and demotivated. (USER WAS PUT ON PROBATION FOR THIS POST) Byzantine fucked around with this message at 16:45 on Feb 6, 2024 |
# ? Feb 6, 2024 08:28 |
|
I can't get everything that I want, so why should I ever vote or otherwise participate in society? - me, a genius (USER WAS PUT ON PROBATION FOR THIS POST)
|
# ? Feb 6, 2024 10:08 |
|
Nenonen posted:I can't get everything that I want, so why should I ever vote or otherwise participate in society? Nobody has said this.
|
# ? Feb 6, 2024 11:22 |
|
A.F. Branco Andy Marlette Bob Gorrell Mark the date, Gorrell actually drew something new. Chip Bok Gary Markstein Gary Varvel John Deering Michael Ramirez Mike Luckovich Steve Breen I really wish people would get Noah's ark right... quote:Take with you seven pairs of every kind of clean animal, a male and its mate, and one pair of every kind of unclean animal, a male and its mate, and also seven pairs of every kind of bird, male and female, to keep their various kinds alive throughout the earth. Tom Stiglich Thank you for telling us you have the palate of a child, Tom.
|
# ? Feb 6, 2024 12:15 |
|
Skios posted:Tom Stiglich
|
# ? Feb 6, 2024 12:25 |
|
Skios posted:
Rhinos (?), bears and zebras are all unclean animals. The only clean mammals are the ruminants, they have both cloven hooves and chew cud.
|
# ? Feb 6, 2024 13:09 |
|
Skios posted:Gary Varvel A really stupid message presented in a decently compelling way, I assume there's no category for that
|
# ? Feb 6, 2024 14:16 |
|
Vincent Van Goatse posted:I can't I disagree with anything in this because I have no clue what on Earth Lub is trying to say.
|
# ? Feb 6, 2024 14:37 |
|
Phlegmish posted:A really stupid message presented in a decently compelling way, I assume there's no category for that
|
# ? Feb 6, 2024 14:40 |
|
Byzantine posted:Harm reduction/lesser-evilism is an admission that we have already lost and there is no hope of things ever getting better. Things will forever get worse, more evil and more harmful, and the only choice we get is rate of increase. And then the people exposing it seem baffled that that makes people depressed and demotivated. No, it's not? It's an acknowledgement that there are near-term measures that can and should be implemented quicker while advocating for longer-term solutions. Nobody advocating harm reduction is saying "We get to this part and then stop"; it's not about giving up halfway and assuring yourself you did a good job. It's basic first aid; you do what you can to minimize damage now, then get to the next level of care where something can be actually solved. If someone's shot in the leg, I can staunch the bleeding with a tourniquet, but nobody is under the illusion that the tourniquet is the solution, it's simply a stop-gap until I can get you to a hospital or at least a first responder.
|
# ? Feb 6, 2024 15:40 |
|
I've been waiting for the Revolution™ for 40 years at this point and it's no closer today than it ever was so yeah, harm reduction voting seems like a good idea to me.
|
# ? Feb 6, 2024 16:16 |
|
|
# ? Jun 3, 2024 10:32 |
|
Neito posted:No, it's not? It's an acknowledgement that there are near-term measures that can and should be implemented quicker while advocating for longer-term solutions. Nobody advocating harm reduction is saying "We get to this part and then stop"; it's not about giving up halfway and assuring yourself you did a good job. Vincent Van Goatse posted:I've been waiting for the Revolution™ for 40 years at this point and it's no closer today than it ever was so yeah, harm reduction voting seems like a good idea to me.
|
# ? Feb 6, 2024 16:21 |