But I'd say the Great Filter of the modern moment is more aimed at the technology than the people. What we're stress-testing is traditional modes of communication. How much noise can be generated before the signal is lost?
We have enough people without the general ability to critically think that the fallout will be that we will rush into a probable Nuclear war coupled with ignoring climate change till it’s too late. Meanwhile Billionaires are undermining democracy across the globe chasing short term profits and power. A filter event doesn’t have to be a single thing, more likely it’s a bunch of events all building to the actual “filter”.
We have enough people without the general ability to critically think
Everyone has the ability to think critically. What you often stumble on is their priors. People who have accumulated a body of understanding or a logic flow that is totally alien to you don't seem to follow any kind of rational behavior patterns.
the fallout will be that we will rush into a probable Nuclear war coupled with ignoring climate change till it’s too late
We've been predicting imminent interstate thermonuclear war since the last bomb dropped on Nagasaki. Nearly a century later, we've never used one in an overt act of aggression. There's a whole litany of reasons why. But the hard truth of it is that nuclear weapons aren't any more efficient than nuclear power plants. You can get a lot more bang for your buck doing violence the conventional way. And we've done just that for decades.
Meanwhile Billionaires are undermining democracy across the globe
What we're discovering is how little democracy there was to undermine. But, again, this isn't news to anyone who has been paying attention. What's shifted is demographics, not politics. The formerly hegemonic upper-middle class white male voter is no longer the dominant political face or voice, outside of the increasingly cloistered echo chambers of reactionary state-sanctioned media.
Even the fucking billionaires don't have the decency to be interchangeable WASP stereotypes. So now we're seeing a real crisis in continuity of governance, as Middle Easterners and East Asians and South Africans and Latin Americans all pop their heads up into the top rungs of the bureaucracy.
What we are seeing in the modern era isn't a degraded democracy. It's a civil war between the various plutocrats who they they should inherit the American Throne.
We are well and truly fucked.
The blinders are coming off your eyes as you learn more about the real function of the world at-large. And this can be immediately shocking, to the point that you assume "We're at the end of the world!"
But to quote from the sage words of Mark Fisher, author of Capitalist Realism:
Capitalist realism as I understand it cannot be confined to art or to the quasi-propagandistic way in which advertising functions. It is more like a pervasive atmosphere, conditioning not only the production of culture but also the regulation of work and education, and acting as a kind of invisible barrier constraining thought and action.
and, therefore
it is easier to imagine an end to the world than an end to capitalism
What we're witnessing is the end of an era. It's a thing every generation gets to bare witness to over enough time. It isn't the end of The World, just the end of the world you were born into.
I really enjoy when people dont know the difference between democracy and republic. Democracy isnt failing, it simply hasnt been tried since athens. Whats failing is the system of corruption people like to call a "republic". Why tf you would elect people to represent your interests instead of representing your interests is beyond me. Republics are the original 'outsourcing your thinking' creation.
Democracy isnt failing, it simply hasnt been tried
Jesus Christ dude, crack a civics book. We have a litany of mechanisms for direct popular governance already in place. My state constitution gets updated with a raft of popularly chosen amendments every election cycle.
Why tf you would elect people to represent your interests instead of representing your interests
Elected bureaucracy is not the only mode of civil engagement
Like the past few decades of stripping education and anti-intellectual culture and anti-solidarity propaganda isnt a factor. Weve created a perfect storm to keep people obedient and subservient.
Weve all decided that giving all planning permissions and resource allocation to people whose way of life requires we be like this is righteous.
We need to exterminate the billionaires or life on earth does not survive.
Funny thing, every now and then someone makes these questionnaire websites where you answer questions about policy and it tells you which party is most in alignment. Pretty much universally most people turn out to be more to the left than their voting history and intentions would suggest.
Honestly, I wish(not like this, of course). Most people vote based on propaganda and not actual policy so it would be kinda sweet to see people’s vote actually alogn with their goals for fucking once.
You know what else costs a fraction of traditional polling and takes a fraction of the time?
Lying, making shit up. Which conveniently is basically what AI slop does, and having a person lie is even cheaper than licensing some random AI to do it.
That is exactly it, the AI gives them an excuse to blame someone else even as they had every reason to know, and they did know, we all know they know but the courts pretend like they didn't know because the Federalist Society.
That bullshit line of thinking only remotely works if you let AI make the decisions from the beginning. Somebody still made the decision to ask AI for bullshit stats. That's the problem here, the human decisions, not necessarily the AI output.
Thanks! This was a spur of the moment decision, but I have thought about moderating a comm for a while. This seems like a good candidate to start with as I have a decades-long back catalogue of tech nonsense to draw from, I'll probably sit down an write out a list sometime this week and plan some posts out. At one post a day I bet I could keep it going myself for a few months, before I take into account the eternal firehose of techbrobabble I drink from every day lol.
“opinions” formed from a mix of stolen books and movie scripts, terminally online shutins and fanfic writers, and politics comment sections cannot be considered a holistic look at humanity.
We’re absolutely going extinct. I’m out of hope at this point.
When you are right, you are right. This dumpster fire needs to burn itself out. It is so much worse than we think though, and we could stop it, by organizing.
Still pretty bad though. "Hey, this study agrees with what we want to publish, let's do this thing without fact checking!" Nobody gives a fuck anymore. We're doomed.
Any scientific publications accidentally posting an article not based on the actual scientific method should be immediately punished by law or we are lost. It's time they used the hordes of money they accumulated during the easy part to now prove themselves and perform their actual function in society.
Cool, so everything is just fucking made up now. Why even bother with the AI at that point? Just make up stats that say what you want right there on the spot. Its the same fucking different. Bullshit from humans or bullshit from AI, its all still bullshit.
Holy fuck. It can simulate large samplings or it can just hallucinate some nonsensical BS that completely misinterprets the data it gathers in order to agree with the phrasing of the person who created the prompt.
Do the majority of people trust their doctors and nurses? Maybe. Or, maybe it depends on the context of the question.
Do I trust my doctors and nurses are a better source of information than random internet advice and AI generated slop? I would hope so.
Do I trust that the American healthcare system is set up to prioritize the health and well-being of the patient over maximizing profits and forcing healthcare workers to adhere to standardized time allotments of 10 to 15 minutes for every patient interaction regardless of the individual case? Absolutely not.
Yep. AI, such as it is, can be useful for some things. I had had a song kinda stuck in my head for years, couldn't remember the lyrics, artist, or song name - only that it was in french and done stuff from the video. ChatGPT worked it out quickly. On the other hand, I was using it to look up definitions for a legal case I'm kinda involved in that hasn't really happened before - I know I shouldn't, but I'm not a lawyer! - and I found it was using my own case which has not been decided as evidence.
So yeah, they can be impressive but you can't trust these programmes with anything important.
Ever notice that they're just doing what Clavicular or whatever his name is doing? They're inventing lingo to sound like its not bullshit.
Its not hitting yourself with a hammer its looksmaxxing. Its not standing around being a dork, its mogging. Its not a context window, its the chat scrollback. Its not asking chatgpt its "silicon sampling".
They're making it seem legit by making its own terminology and in-group lingo.
We had the photospread, but nothing directly uttered by the 29-year-old. With a driving need for a feature, we had to be inventive. Harnessing our creative license, we pulled his verbatim from previous interviews and fed them through an AI programme to formulate new responses.
Are these the words we expect from Mackenyu? Or are they just replies from an echo chamber of celebrity-hood that we want to believe is from him?
With the absence of information, can new insights be gained?
Nature abhors a vacuum, and in its place, a story fills the hollow.
Somehow it is currently accepted by a certain portion of people that LLM-based systems can be used to replace actually existing human beings.
Lately, I've had some coworkers empowered by AI in really cool ways, building mockups using code they can't personally write to present to me, an actual engineer. Not with the expectations of a final product, but to express their thoughts and ideas and how they would envision a project moving forward. I think that's a really cool and exciting use of AI, allowing non-technical people to better communicate with technical people.
Then I see crap like this and think "we need to burn this shit down immediately."
Polls have always been leveraged as a form of propaganda.
We had Push Polling from back during the early Bush Era, where the ostensible polling cold call was just a marketing tool. We had "Unskewed Polls" during the Obama/Romney election, wherein Republicans tried to insist they were far more popular in order to influence everyone else through bandwagon appeal. Polling about Transgender Athletes was used as an excuse to dismantle civil protections for the LGBTQ community. Polls online are used to gather information on the public through responses and attendant metadata. Call-in shows are a form of engagement bait.
You can talk about the useful information gleaned from a public survey. But by and large, we only take polls when we want to change people's opinions. Its the first step in market research that ends with a blizzard of advertisements.
The thing is, logically the distribution of opinions or individual situations/beliefs which lead to those opinions, has been baked into the model when the data used to train it was captured, which means that at best and if the entire principle of the thing works (which itself isn't mathematically proven in any way form or shape) they're still getting only poll results for the past and which will not actually change beyond some random noise until the next time data is captured and the model is retrained.
It's like repeatedly using an old picture of a street to make realtime claims about the traffic there.
It’s like repeatedly using an old picture of a street to make realtime claims about the traffic there.
It is worse then that really, as the data used has shown to be heavily siloed (think a subreddit with heavy modderation or a facebook group). So its like using a old biased propaganda photo of a street to make realtime claims about the traffic there.
So, use AI to just make shit up, then report that as information? I wouldn't expect anything less in our post-truth era. However, come on Axios, I thought you were better than that.
Tempted to start a AI survey company, only take big corp customers. Then always give them the opposite of intended survey results I assume they want.
Really if its fuckin "AI" doing the survey why even go through to steps. Just make some graphs, demographic breakdowns and what not, then let a random number generator do the rest. Don't even need gen ai.
This is dumb. But, there's a hint of an interesting idea in there. If LLMs sample all human text and produce statistical averages from it, there's a sense in which they contain a statistically average opinion.
It's basically like how if you use Google to search for "how many calories are there in a" it will suggest the next word. The word it suggests is the statistically average way to finish that sentence. That also means it's the food item for which people most want to know the calories. At least, it's the item they most type into a Google search box. It's just matching text patterns, but it reveals something about people that say a fast food company might find useful.
If you scale up the population of humanity to 8 trillion people and have thousands of years of data in these LLMs, maybe you actually do get useful insights about what people care about. And, maybe that's how you get psychohistory from Asimov's foundation series.
Even if it did spit out the average value, it would be the average of the training data. I don't think people/opinions are evenly distributed in LLM training data.
Just look into how racist computer vision models can be.
What's misleading? They are claiming it's an opinion poll, when in reality they didn't ask humans but asked LLM agents, because they think that LLM agents will output similar things as humans.
That's literally faking an opinion poll using LLM, since most reasonable people would think that an opinion poll polls the opinion of people. (LLMs don't have opinions, that's a human concept.)
probably that it wasn't axios that actually performed the survey, they just cited it. although leaving it up is dumb, and their "clarification" just makes it dumber.
Hmm, I see you've linked to an image and included the citation as text. While a lot of people may like that format, and in a way it seems like the best of both worlds, I automatically downvote it because on my home page it looks like an image submission, which is a poor source.
I could start clicking the expando next to the title to check if a source was provided, but images are going to start getting auto-blocked in the next lemmy update anyway, so this format is not good.
Naw, this has been wafted around as a potential new feature allowing users to autoblock images for a while now. The funny part is that even if it does come to be, who cares? If people don't want pictures, whatever, it is not going to stop pictures being used and there is not really much concern if someone breaks their own use of lemmy (hell it is looking like people are going to dunk on those users).
Yes no big deal having the option to autoblock images, a literal nothing burger. Having to change the format of posts and comments to pander to a small minority of users that think they should have a say about other users is an issue. The format police are not popular and most users would use more images just to spite them.
If or when that feature rolls out Lemmy wide, it better be an option and not a locked setting.
Is this not just "have the image be blacked out with a 'hidden' overlay, then you can click it to unhide it (with a button to re-hide it again if needed)"? That's how it works over on Masto, and it doesn't change the format of anything at all.
I don't know much about Lemmy development so have no idea what this new feature is like. If it's more like "filters out posts with images so you don't even see they exist in the first place", I could see that being annoying.
Lotta pearl clutching in this thread, as if every poll has always been rigorous and checked against bias. At least when they run a sloppy simulation you automatically have a massive grain of salt instead of requiring every reader to sniff their methods.
Would you rather read the headline "Simulated poll implies people could love Hitler" or "Poll finds people love Hitler*poll of KKK headquarters, sample size: Kyle"?
Seems like Axios fucked up by not properly vetting their source but it's no better or worse than them failing to catch a dishonest pollster.
I'm not both sidesing, I'm pointing out that misinformation and biased studies have existed forever. It's the responsibility of the media to suss that out and verify it before reporting. That's just a plain fact of the job, nobody can offload that onto the source.
regular polls are sometimes wrong too
"Sometimes wrong" is an incredibly generous way to put it. If reporters had got out ahead in vetting the bullshit vaccine-autism study before amplifying it, we wouldn't have people dying of measles today. Asking them to do the bare minimum of checking that a human poll even happened is not asking much.
lied about everything
The source didn't lie about anything, they used Ai mysticism to project plausible sounding answers. Anyone can do that and it's not that subversive. The people who believe them either fundamentally don't understand the tech or are too lazy to look closely. Don't condone that passivity by pretending baseline media and technical literacy are impossible to achieve (especially for professional reporters).
Great, so now they're outsourcing our own fucking opinions to them
This just in: 95% of poll respondents think tech bro oligarchy is a really swell idea.
We are keen on ai allright. Until we can burn it to the ground.
Wait till these Silicon Sampling reports start popping up in election exit-polls.
We're going to get levels of "Proven Election Fraud" and "Illegal Immigrant Voter Invasion" hysteria the likes of which you've never seen.
We are living through a great filter event and we aren’t going to come through the other side
Pretty much by definition. Nobody lives forever.
But I'd say the Great Filter of the modern moment is more aimed at the technology than the people. What we're stress-testing is traditional modes of communication. How much noise can be generated before the signal is lost?
We have enough people without the general ability to critically think that the fallout will be that we will rush into a probable Nuclear war coupled with ignoring climate change till it’s too late. Meanwhile Billionaires are undermining democracy across the globe chasing short term profits and power. A filter event doesn’t have to be a single thing, more likely it’s a bunch of events all building to the actual “filter”.
We are well and truly fucked.
Everyone has the ability to think critically. What you often stumble on is their priors. People who have accumulated a body of understanding or a logic flow that is totally alien to you don't seem to follow any kind of rational behavior patterns.
We've been predicting imminent interstate thermonuclear war since the last bomb dropped on Nagasaki. Nearly a century later, we've never used one in an overt act of aggression. There's a whole litany of reasons why. But the hard truth of it is that nuclear weapons aren't any more efficient than nuclear power plants. You can get a lot more bang for your buck doing violence the conventional way. And we've done just that for decades.
What we're discovering is how little democracy there was to undermine. But, again, this isn't news to anyone who has been paying attention. What's shifted is demographics, not politics. The formerly hegemonic upper-middle class white male voter is no longer the dominant political face or voice, outside of the increasingly cloistered echo chambers of reactionary state-sanctioned media.
Even the fucking billionaires don't have the decency to be interchangeable WASP stereotypes. So now we're seeing a real crisis in continuity of governance, as Middle Easterners and East Asians and South Africans and Latin Americans all pop their heads up into the top rungs of the bureaucracy.
What we are seeing in the modern era isn't a degraded democracy. It's a civil war between the various plutocrats who they they should inherit the American Throne.
The blinders are coming off your eyes as you learn more about the real function of the world at-large. And this can be immediately shocking, to the point that you assume "We're at the end of the world!"
But to quote from the sage words of Mark Fisher, author of Capitalist Realism:
and, therefore
What we're witnessing is the end of an era. It's a thing every generation gets to bare witness to over enough time. It isn't the end of The World, just the end of the world you were born into.
The climate scientists say otherwise.
I really enjoy when people dont know the difference between democracy and republic. Democracy isnt failing, it simply hasnt been tried since athens. Whats failing is the system of corruption people like to call a "republic". Why tf you would elect people to represent your interests instead of representing your interests is beyond me. Republics are the original 'outsourcing your thinking' creation.
Jesus Christ dude, crack a civics book. We have a litany of mechanisms for direct popular governance already in place. My state constitution gets updated with a raft of popularly chosen amendments every election cycle.
Elected bureaucracy is not the only mode of civil engagement
You act like people=people.
Like the past few decades of stripping education and anti-intellectual culture and anti-solidarity propaganda isnt a factor. Weve created a perfect storm to keep people obedient and subservient.
Weve all decided that giving all planning permissions and resource allocation to people whose way of life requires we be like this is righteous.
We need to exterminate the billionaires or life on earth does not survive.
We weren't flush with quality education and rich intellectual culture and a staunch labor movement several decades ago.
None of these problems are new or unique to the current moment.
They are worse on purpose. There has been conscious effort. They started teaching children not to read instead of how.
How do you interact with anyone or anything in this country as an illiterate?
Everything Is Computer
Should just have it handle voting as well. They could call it Automatic Democracy.
None of that pesky informed voting, you can just instruct an AI company on what your stance is, and it'll vote in your stead.
Funny thing, every now and then someone makes these questionnaire websites where you answer questions about policy and it tells you which party is most in alignment. Pretty much universally most people turn out to be more to the left than their voting history and intentions would suggest.
Honestly, I wish(not like this, of course). Most people vote based on propaganda and not actual policy so it would be kinda sweet to see people’s vote actually alogn with their goals for fucking once.
See a dude in a grey hoodie. Ask if he works in tech.
If he says yes, beat the shit out of him. Break his fingers and tell him he should learn to vibe code.
This was always the point. Did anyone not see this coming?
You know what else costs a fraction of traditional polling and takes a fraction of the time?
Lying, making shit up. Which conveniently is basically what AI slop does, and having a person lie is even cheaper than licensing some random AI to do it.
Ah, but licensing the AI lets you blame the AI when it inevitably blows up
That is exactly it, the AI gives them an excuse to blame someone else even as they had every reason to know, and they did know, we all know they know but the courts pretend like they didn't know because the Federalist Society.
That bullshit line of thinking only remotely works if you let AI make the decisions from the beginning. Somebody still made the decision to ask AI for bullshit stats. That's the problem here, the human decisions, not necessarily the AI output.
Ah, yes, lies.
Technically, these are damned lies because they've been summarized.
Is there a community for cataloging tech bullshit like "silicon sampling"? If not I'll make one. First thought
c/techbrobabbleEDIT: 'tis done ![email protected]
I hope you're ready to be the only poster until the community catches on, and then that you're ready to play moderator. best of luck
Thanks! This was a spur of the moment decision, but I have thought about moderating a comm for a while. This seems like a good candidate to start with as I have a decades-long back catalogue of tech nonsense to draw from, I'll probably sit down an write out a list sometime this week and plan some posts out. At one post a day I bet I could keep it going myself for a few months, before I take into account the eternal firehose of techbrobabble I drink from every day lol.
That's fucking stupid.
You beat me to it. I was going to comment that this is literally the stupidest shit I have ever heard in my goddamn life.
I could go even cheaper by just thinking about it really hard and guessing
honestly it'd probably be better unless you're actively hallucinating bullshit
Hey at least this is one poll result rather than zero
“opinions” formed from a mix of stolen books and movie scripts, terminally online shutins and fanfic writers, and politics comment sections cannot be considered a holistic look at humanity.
We’re absolutely going extinct. I’m out of hope at this point.
I figure ill be around to see a shitshow, and then I'm out. I'm not leaving any kids behind. Making a clean break this time.
Hmmmmm
When you are right, you are right. This dumpster fire needs to burn itself out. It is so much worse than we think though, and we could stop it, by organizing.
I have an environment friendly alternative to this method. It involves tea leaves…
It costs less to make shit up that "mimics" the real information. Who would have thought
Misleading title, Axios did not do this, but rather the referenced a study that they later discovered did this.
It’s on them for not learning this sooner, but let’s not act like they’re the ones who sent it up to try and manipulate political reporting.
Still pretty bad though. "Hey, this study agrees with what we want to publish, let's do this thing without fact checking!" Nobody gives a fuck anymore. We're doomed.
Any scientific publications accidentally posting an article not based on the actual scientific method should be immediately punished by law or we are lost. It's time they used the hordes of money they accumulated during the easy part to now prove themselves and perform their actual function in society.
Cool, so everything is just fucking made up now. Why even bother with the AI at that point? Just make up stats that say what you want right there on the spot. Its the same fucking different. Bullshit from humans or bullshit from AI, its all still bullshit.
What the actual fuck??
Holy fuck. It can simulate large samplings or it can just hallucinate some nonsensical BS that completely misinterprets the data it gathers in order to agree with the phrasing of the person who created the prompt.
Do the majority of people trust their doctors and nurses? Maybe. Or, maybe it depends on the context of the question.
Do I trust my doctors and nurses are a better source of information than random internet advice and AI generated slop? I would hope so.
Do I trust that the American healthcare system is set up to prioritize the health and well-being of the patient over maximizing profits and forcing healthcare workers to adhere to standardized time allotments of 10 to 15 minutes for every patient interaction regardless of the individual case? Absolutely not.
Yep. AI, such as it is, can be useful for some things. I had had a song kinda stuck in my head for years, couldn't remember the lyrics, artist, or song name - only that it was in french and done stuff from the video. ChatGPT worked it out quickly. On the other hand, I was using it to look up definitions for a legal case I'm kinda involved in that hasn't really happened before - I know I shouldn't, but I'm not a lawyer! - and I found it was using my own case which has not been decided as evidence.
So yeah, they can be impressive but you can't trust these programmes with anything important.
Unfucking believable that a legitimate media outlet would do such a thing. That’s some Breitbart shit.
Ever notice that they're just doing what Clavicular or whatever his name is doing? They're inventing lingo to sound like its not bullshit.
Its not hitting yourself with a hammer its looksmaxxing. Its not standing around being a dork, its mogging. Its not a context window, its the chat scrollback. Its not asking chatgpt its "silicon sampling".
They're making it seem legit by making its own terminology and in-group lingo.
Bullshit artists.
Sadly, I'm not too surprised. Check this shit out, published back in November 2025: https://arxiv.org/pdf/2510.25137.
"We simulated 151 million American workers [using LLMs] to see what proportion of tasks they do that can also be done by AI".
Much more recently, Esquire couldn't get ahold of an actor for an interview and so decided to generate the actors responses using Claude: https://esquiresg.com/mackenyu-one-piece-roronoa-zoro-interview/.
Somehow it is currently accepted by a certain portion of people that LLM-based systems can be used to replace actually existing human beings.
Doing an interview with an LLM trained on a real person feels like libel.
What the absolute fuck. Even if I was pro-AI, I'd find this to be incredibly unethical.
Lately, I've had some coworkers empowered by AI in really cool ways, building mockups using code they can't personally write to present to me, an actual engineer. Not with the expectations of a final product, but to express their thoughts and ideas and how they would envision a project moving forward. I think that's a really cool and exciting use of AI, allowing non-technical people to better communicate with technical people.
Then I see crap like this and think "we need to burn this shit down immediately."
Making shit up, but with extra steps.
Sure, our polls mean nothing, but think about the money you can save!!!
This might be the dumbest fucking thing I've ever heard.
The ideal would be that clients who actually want useful information will stop paying the pollsters for their useless crap.
The reality will be that slack will be more than picked up by people who want sham poll results to back up their agenda.
Polls have always been leveraged as a form of propaganda.
We had Push Polling from back during the early Bush Era, where the ostensible polling cold call was just a marketing tool. We had "Unskewed Polls" during the Obama/Romney election, wherein Republicans tried to insist they were far more popular in order to influence everyone else through bandwagon appeal. Polling about Transgender Athletes was used as an excuse to dismantle civil protections for the LGBTQ community. Polls online are used to gather information on the public through responses and attendant metadata. Call-in shows are a form of engagement bait.
You can talk about the useful information gleaned from a public survey. But by and large, we only take polls when we want to change people's opinions. Its the first step in market research that ends with a blizzard of advertisements.
63% of all statistics are made up on the spot
Now you just made it 63.0001%
Now I just made it 63.0002%
...
65% of all statistics are made up right there on the spot!
You wouldn’t know my survey results, she goes to another school
Crossposting to fuck_ai if you don't mind.
::: spoiler nsfw :::
We all wish this was as fulfilling as that.
Not that I would know, but I've heard it feels a lot more like sucking dick than it does having your dick sucked.
Both scenarios seem to be a lot more desirable then an AI fake poll.
Why bother asking people when you can just ask AI to make answers.
The thing is, logically the distribution of opinions or individual situations/beliefs which lead to those opinions, has been baked into the model when the data used to train it was captured, which means that at best and if the entire principle of the thing works (which itself isn't mathematically proven in any way form or shape) they're still getting only poll results for the past and which will not actually change beyond some random noise until the next time data is captured and the model is retrained.
It's like repeatedly using an old picture of a street to make realtime claims about the traffic there.
It is worse then that really, as the data used has shown to be heavily siloed (think a subreddit with heavy modderation or a facebook group). So its like using a old biased propaganda photo of a street to make realtime claims about the traffic there.
Kinda sounds like numbers pulled out of your ass…
I've read a lot of fucked up shit in the last few years - that's the first time I've thrown my phone in response!
(My phone's fine - I just threw it onto the sofa beside me, but still..)
Here, have one of these:
Sorry, wrong pic:
What could possibly go wrong?
So, use AI to just make shit up, then report that as information? I wouldn't expect anything less in our post-truth era. However, come on Axios, I thought you were better than that.
Well this might go a ways in explaining why polling has been so off in the last few years.
Tempted to start a AI survey company, only take big corp customers. Then always give them the opposite of intended survey results I assume they want.
Really if its fuckin "AI" doing the survey why even go through to steps. Just make some graphs, demographic breakdowns and what not, then let a random number generator do the rest. Don't even need gen ai.
Their enshittification was inevitable once they were acquired by Cox media
Ah that explains it.
I was wondering why it seemed everyone soured on Axios.
GFC.
Surprisingly unrelated to the recently compromised npm package.
I relate to these the same as almost any poll.
This is dumb. But, there's a hint of an interesting idea in there. If LLMs sample all human text and produce statistical averages from it, there's a sense in which they contain a statistically average opinion.
It's basically like how if you use Google to search for "how many calories are there in a" it will suggest the next word. The word it suggests is the statistically average way to finish that sentence. That also means it's the food item for which people most want to know the calories. At least, it's the item they most type into a Google search box. It's just matching text patterns, but it reveals something about people that say a fast food company might find useful.
If you scale up the population of humanity to 8 trillion people and have thousands of years of data in these LLMs, maybe you actually do get useful insights about what people care about. And, maybe that's how you get psychohistory from Asimov's foundation series.
Even if it did spit out the average value, it would be the average of the training data. I don't think people/opinions are evenly distributed in LLM training data.
Just look into how racist computer vision models can be.
Its one thing if you present your findings like that, but pretending like its a poll is complete bullshit.
Yeah, definitely.
Misleading title
What's misleading? They are claiming it's an opinion poll, when in reality they didn't ask humans but asked LLM agents, because they think that LLM agents will output similar things as humans.
That's literally faking an opinion poll using LLM, since most reasonable people would think that an opinion poll polls the opinion of people. (LLMs don't have opinions, that's a human concept.)
probably that it wasn't axios that actually performed the survey, they just cited it. although leaving it up is dumb, and their "clarification" just makes it dumber.
Axios didn't say what their source was until they were called out.
Hmm, I see you've linked to an image and included the citation as text. While a lot of people may like that format, and in a way it seems like the best of both worlds, I automatically downvote it because on my home page it looks like an image submission, which is a poor source.
I could start clicking the expando next to the title to check if a source was provided, but images are going to start getting auto-blocked in the next lemmy update anyway, so this format is not good.
Images are going to get auto blocked?
Naw, this has been wafted around as a potential new feature allowing users to autoblock images for a while now. The funny part is that even if it does come to be, who cares? If people don't want pictures, whatever, it is not going to stop pictures being used and there is not really much concern if someone breaks their own use of lemmy (hell it is looking like people are going to dunk on those users).
Over on Mastodon we block all images by default and it's SUPER nice. We click the picture to see it, it's no big deal.
-- Frost
Yes no big deal having the option to autoblock images, a literal nothing burger. Having to change the format of posts and comments to pander to a small minority of users that think they should have a say about other users is an issue. The format police are not popular and most users would use more images just to spite them.
If or when that feature rolls out Lemmy wide, it better be an option and not a locked setting.
Is this not just "have the image be blacked out with a 'hidden' overlay, then you can click it to unhide it (with a button to re-hide it again if needed)"? That's how it works over on Masto, and it doesn't change the format of anything at all.
I don't know much about Lemmy development so have no idea what this new feature is like. If it's more like "filters out posts with images so you don't even see they exist in the first place", I could see that being annoying.
-- Frost
There will be a new user option to block image posts.
Lotta pearl clutching in this thread, as if every poll has always been rigorous and checked against bias. At least when they run a sloppy simulation you automatically have a massive grain of salt instead of requiring every reader to sniff their methods.
Would you rather read the headline "Simulated poll implies people could love Hitler" or "Poll finds people love Hitler *poll of KKK headquarters, sample size: Kyle"?
Seems like Axios fucked up by not properly vetting their source but it's no better or worse than them failing to catch a dishonest pollster.
You're really both sides-ing facts???
"Sure, they completely made it up and lied about everything, but regular polls are sometimes wrong too. So it's basically the same thing."
I'm not both sidesing, I'm pointing out that misinformation and biased studies have existed forever. It's the responsibility of the media to suss that out and verify it before reporting. That's just a plain fact of the job, nobody can offload that onto the source.
"Sometimes wrong" is an incredibly generous way to put it. If reporters had got out ahead in vetting the bullshit vaccine-autism study before amplifying it, we wouldn't have people dying of measles today. Asking them to do the bare minimum of checking that a human poll even happened is not asking much.
The source didn't lie about anything, they used Ai mysticism to project plausible sounding answers. Anyone can do that and it's not that subversive. The people who believe them either fundamentally don't understand the tech or are too lazy to look closely. Don't condone that passivity by pretending baseline media and technical literacy are impossible to achieve (especially for professional reporters).