AI Answers in Ask MetaFilter? April 23, 2025 8:33 PM Subscribe
I am wondering what the memberships' sentiments are on the use of AI services to answer questions in Ask MetaFilter. Is this something that people think is great or grating? Is this something that we should just accept going forward or ban for the foreseeable future? Maybe you have a nuanced take on the matter?
I am not trying to stir up trouble though I recognize that this may rouse the passions of many here, so please, let us all keep civility in mind.
I am not trying to stir up trouble though I recognize that this may rouse the passions of many here, so please, let us all keep civility in mind.
I am like 200% on board with our policy. I think it's safe to assume that anybody who would be delighted to receive an LLM generated answer will just ask their question to an LLM, and so everyone asking questions to metafilter is doing so because they want an answers from humans.
posted by aubilenon at 9:54 PM on April 23 [49 favorites]
posted by aubilenon at 9:54 PM on April 23 [49 favorites]
I am personally glad to avoid LLM usage completely. I find them, in their current forms, to be without demonstrable benefits for their users. However I have noticed more people using them and I don't want to give into my knee jerk reaction of, "Kill it! Kill it with fire." I would like to imagine that I could be open to well thought out arguments as to why these systems are of actual value in answering queries.
posted by Ignorantsavage at 1:30 AM on April 24 [1 favorite]
posted by Ignorantsavage at 1:30 AM on April 24 [1 favorite]
There are certain things where Mefi’s wisened fuddy duddiness really bothers me and makes clear why we have hemorrhaged members and activity. But the decisions and discussion about current “AI” resulting in the faq linked above is definitely something that makes me proud to be a wisened fuddy duddy.
I think that it is entirely possible that within the next decade or so we will have reason to revisit this policy, as LLMs and related programs become more refined, more ethical, and less trendy. Hopefully mefi will be around then to fight about it some more.
posted by Mizu at 1:57 AM on April 24 [7 favorites]
I think that it is entirely possible that within the next decade or so we will have reason to revisit this policy, as LLMs and related programs become more refined, more ethical, and less trendy. Hopefully mefi will be around then to fight about it some more.
posted by Mizu at 1:57 AM on April 24 [7 favorites]
Even setting aside the many ethical issues around AI use, I think the rule we have is a good one because AI responses seem a lot like dropping a link to the first page of google results or a lmgtfy.com link. People use AskMe to get the benefit of the knowledge and experience of actual humans. Presuming that the asker will find a generic AI response to their question useful is presuming that they are incapable of doing even the most basic research on their question, and it's insulting in the same way that "hey, I don't know anything about this topic but here's the first result on the search page when I drop your title in to Google" is insulting. Even in cases where an LLM's output might be useful and the OP hasn't considered using one, saying that is a better answer than just dumping the output into the comment box, in a "giving a MeFite a fish vs teaching them to fish" sort of way.
posted by firechicago at 3:17 AM on April 24 [8 favorites]
posted by firechicago at 3:17 AM on April 24 [8 favorites]
Personally I get immense power out of LLMs, and lean on them heavily as my assistant, as my language teacher, as my researcher, etc etc. But I also put a lot of effort into crafting prompts to lean into the strengths of the LLM and avoid its weaknesses.
I don't think "paste the question in an LLM and extract the result" answers make sense on Metafilter any more than a shelf full of electrical tools makes sense in the shoe store. As the policy says, if people want such a low effort response they can get it from an LLM themselves.
People can always post "help me write a great LLM prompt for this problem" questions if they want to, although I suspect that crafting a really good LLM prompt for your average AskMe question would require more back and forth than is currently allowed on AskMe.
posted by quacks like a duck at 3:59 AM on April 24 [5 favorites]
I don't think "paste the question in an LLM and extract the result" answers make sense on Metafilter any more than a shelf full of electrical tools makes sense in the shoe store. As the policy says, if people want such a low effort response they can get it from an LLM themselves.
People can always post "help me write a great LLM prompt for this problem" questions if they want to, although I suspect that crafting a really good LLM prompt for your average AskMe question would require more back and forth than is currently allowed on AskMe.
posted by quacks like a duck at 3:59 AM on April 24 [5 favorites]
I don't think "paste the question in an LLM and extract the result" answers make sense on Metafilter any more than a shelf full of electrical tools makes sense in the shoe store. As the policy says, if people want such a low effort response they can get it from an LLM themselves.
I'm generally fine with this take, and how we're enforcing things today. But that said, it feels like there's been a rise in questions that could easily be answered by a minute or two on Google, and I'm having a hard time getting worked up about someone taking a question, dumping it into ChatGPT or whatever, and then giving the poster what they're looking for, especially when they're absolutely transparent about what they've done.
posted by NotMyselfRightNow at 4:15 AM on April 24 [3 favorites]
I'm generally fine with this take, and how we're enforcing things today. But that said, it feels like there's been a rise in questions that could easily be answered by a minute or two on Google, and I'm having a hard time getting worked up about someone taking a question, dumping it into ChatGPT or whatever, and then giving the poster what they're looking for, especially when they're absolutely transparent about what they've done.
posted by NotMyselfRightNow at 4:15 AM on April 24 [3 favorites]
NotMyselfRightNow - I've noticed that google is becoming absolutely rubbish. What I used to be confident I could just google for a decade ago, I'm finding frustrating and difficult to craft a search that yields what I actually want, instead of what the algorithm has "helpfully" decided I actually need/want.
So I definitely can see there being an uptick in previously lmgtfy questions that are there genuinely, not from laziness.
I don't trust AI to not hallucinate, especially when I don't know enough to verify the response. So, I don't think it's fair to accept "hey that's easy just ask the machine" LLM answers.
I think, too, that Metafilter pushing back on the starry eyed "hey check this cool AI out" impulse is something important. People might not know it's problematic until someone says "hey, don't do that."
posted by freethefeet at 4:34 AM on April 24 [7 favorites]
So I definitely can see there being an uptick in previously lmgtfy questions that are there genuinely, not from laziness.
I don't trust AI to not hallucinate, especially when I don't know enough to verify the response. So, I don't think it's fair to accept "hey that's easy just ask the machine" LLM answers.
I think, too, that Metafilter pushing back on the starry eyed "hey check this cool AI out" impulse is something important. People might not know it's problematic until someone says "hey, don't do that."
posted by freethefeet at 4:34 AM on April 24 [7 favorites]
NotMyselfRightNow: "But that said, it feels like there's been a rise in questions that could easily be answered by a minute or two on Google, and I'm having a hard time getting worked up about someone taking a question, dumping it into ChatGPT or whatever, and then giving the poster what they're looking for, especially when they're absolutely transparent about what they've done."
There have always been questions that could be answered by a minute or two on Google. That's not the reason people come to AskMe - they're looking for answers on AskMe because they trust humans over computers, because humans understand things like nuance. Giving them a computer-generated answer when they've expressly sought human answers is a disservice.
(Tangentially, speaking of AI and AskMe - dude, how has that current question about "how can I get AI to write an article for me" not been flagged to within an inch of its life?)
posted by EmpressCallipygos at 4:37 AM on April 24 [11 favorites]
There have always been questions that could be answered by a minute or two on Google. That's not the reason people come to AskMe - they're looking for answers on AskMe because they trust humans over computers, because humans understand things like nuance. Giving them a computer-generated answer when they've expressly sought human answers is a disservice.
(Tangentially, speaking of AI and AskMe - dude, how has that current question about "how can I get AI to write an article for me" not been flagged to within an inch of its life?)
posted by EmpressCallipygos at 4:37 AM on April 24 [11 favorites]
I said this in the previous thread but I think it's incredibly rude and insulting to respond to a person asking for help with an automatically generated "answer" especially if they could simply generate it themselves if that's what they wanted. I think bad answers are worse than no answers and if you aren't willing to put in actual effort or share actual knowledge with another human walk on by, you don't have to respond.
posted by an octopus IRL at 4:41 AM on April 24 [15 favorites]
posted by an octopus IRL at 4:41 AM on April 24 [15 favorites]
how has that current question about "how can I get AI to write an article for me" not been flagged to within an inch of its life?)
It's an answerable question, not the sort of use of AI that is directly banned. If we delete that we should delete the "get Google AI to explain a made-up idiom" thread on the front page that everyone was enjoying.
posted by EndsOfInvention at 5:12 AM on April 24 [3 favorites]
It's an answerable question, not the sort of use of AI that is directly banned. If we delete that we should delete the "get Google AI to explain a made-up idiom" thread on the front page that everyone was enjoying.
posted by EndsOfInvention at 5:12 AM on April 24 [3 favorites]
The AI article-writing question is a perfectly cromulent use of AskMe.
posted by Klipspringer at 5:26 AM on April 24 [1 favorite]
posted by Klipspringer at 5:26 AM on April 24 [1 favorite]
The prior Metatalk thread had some really good discussion that gives the background of the policy Brandon references.
And for once, I look back on Past Me in that thread without being like, what on earth was I thinking back then? (Of course, it was only two months ago. Summer Me might be firmly on board with "Only AI Answers On AskMe, Please.")
posted by mittens at 5:33 AM on April 24
And for once, I look back on Past Me in that thread without being like, what on earth was I thinking back then? (Of course, it was only two months ago. Summer Me might be firmly on board with "Only AI Answers On AskMe, Please.")
posted by mittens at 5:33 AM on April 24
Klipspringer: "The AI article-writing question is a perfectly cromulent use of AskMe."
(pouts) Okay, I guess.
posted by EmpressCallipygos at 5:33 AM on April 24 [2 favorites]
(pouts) Okay, I guess.
posted by EmpressCallipygos at 5:33 AM on April 24 [2 favorites]
When it comes to wrong answers, I would much rather have confidently wrong answers from a human Mefite than confidently wrong answers from a stochastic parrot.
posted by rmd1023 at 5:57 AM on April 24 [7 favorites]
posted by rmd1023 at 5:57 AM on April 24 [7 favorites]
how has that current question about "how can I get AI to write an article for me" not been flagged to within an inch of its life?
This seems like a “you’re not supposed to use Ask to learn how to break the law” objection. And while LLM use is unpopular here, it is not in fact illegal.
If the person would be guilty of plagiarism in the eyes of their teacher or editor, that would be the teacher’s or editor’s job to deal with.
posted by Lemkin at 6:04 AM on April 24 [1 favorite]
This seems like a “you’re not supposed to use Ask to learn how to break the law” objection. And while LLM use is unpopular here, it is not in fact illegal.
If the person would be guilty of plagiarism in the eyes of their teacher or editor, that would be the teacher’s or editor’s job to deal with.
posted by Lemkin at 6:04 AM on April 24 [1 favorite]
That Ask does seem a little like a bit of chatfilter should be allowed, along the lines of: If you let it write the article for you, you'll quickly find out why AIs are not good at writing articles--or, worse, you'll think you have a good article and it will be very sad when someone informs you otherwise.
The fact that LLMs can be okay summarizers but are terrible, terrible synthesizers seems like it should be just one of those things people know. But the sheer volume of Let AI Do All The Writing For You articles is like some kind of brainwashing.
posted by mittens at 6:24 AM on April 24 [5 favorites]
The fact that LLMs can be okay summarizers but are terrible, terrible synthesizers seems like it should be just one of those things people know. But the sheer volume of Let AI Do All The Writing For You articles is like some kind of brainwashing.
posted by mittens at 6:24 AM on April 24 [5 favorites]
I’ve been one to post a Google result in response to an Ask. Sometimes it’s been deleted. Other times it’s been fine. I agree with freethefeet and quacks like a duck — sometimes to get the answer you want, you need to ask the question in a specific way, crafting a particular string out of specific vocabulary. And in that case, maybe it would be helpful to allow a little bit more thread sitting for refinement and clarification. The times I’ve posted a Google result have been kind of like that — I felt the asker needed help creating a string, bc the question said something like “how do I find what I’m looking for.”
posted by toodleydoodley at 6:31 AM on April 24
posted by toodleydoodley at 6:31 AM on April 24
As the instigator of that ChatGPT MeTa I was cautiously optimistic at the speed and vehemence of the response.
It's a shame, then, that there are still a few users who continue to ignore the rules of the site and post ChatGPT answers, despite that MeTa's conclusion.
Add to that the fact that, as already noted, there are FPPs openly encouraging users to play with AI technology, and I'm not particularly convinced that anyone actually gives a fuck.
posted by fight or flight at 7:22 AM on April 24 [2 favorites]
It's a shame, then, that there are still a few users who continue to ignore the rules of the site and post ChatGPT answers, despite that MeTa's conclusion.
Add to that the fact that, as already noted, there are FPPs openly encouraging users to play with AI technology, and I'm not particularly convinced that anyone actually gives a fuck.
posted by fight or flight at 7:22 AM on April 24 [2 favorites]
there are still a few users who continue to ignore the rules of the site
[high horse]
From the policy: Please do not query ChatGPT or other LLMs and then post the output unless the discussion itself is about those types of tools, and in those cases, please clearly label what you are posting as coming from an AI tool.
[/high horse]
posted by mittens at 7:27 AM on April 24 [13 favorites]
[high horse]
From the policy: Please do not query ChatGPT or other LLMs and then post the output unless the discussion itself is about those types of tools, and in those cases, please clearly label what you are posting as coming from an AI tool.
[/high horse]
posted by mittens at 7:27 AM on April 24 [13 favorites]
In re: the questions that seem easy to answer. Aside from the enshittification of search these days, there are places you just know to go and giving your answer along with the way you found is very helpful. I don't use search for movie and tv info - I go right to IMDB or epguides. I usually start with Wikipedia for historical or geographical facts. So, if someone asks the population of Indonesia, you can answer and also explain the Wikipedia is the best place to find this kind of fact quickly. This is one of my "kids these days" pet peeves. Start with the right source and you will spend much less time finding your info.
posted by soelo at 7:29 AM on April 24 [1 favorite]
posted by soelo at 7:29 AM on April 24 [1 favorite]
freethefeet: "I've noticed that google is becoming absolutely rubbish"
Did you read this post? The gist is that if you add profanity to your question, the summaries often go away. Personally, I use DuckDuckGo and other options as much as I use Google. I set different engines as the default in different browsers and devices on purpose.
posted by soelo at 7:34 AM on April 24
Did you read this post? The gist is that if you add profanity to your question, the summaries often go away. Personally, I use DuckDuckGo and other options as much as I use Google. I set different engines as the default in different browsers and devices on purpose.
posted by soelo at 7:34 AM on April 24
MetaFilter has only ever been special to me as a function of the erudite, empathetic* humanity--humane-ness!-- of its members. LLM glurge is everywhere else, and getting everywhere-r by the minute.
* yes there may be gaps here and there hahaaa
posted by working_objects at 8:38 AM on April 24 [5 favorites]
* yes there may be gaps here and there hahaaa
posted by working_objects at 8:38 AM on April 24 [5 favorites]
there are FPPs openly encouraging users to play with AI technology
When the FAQ says “MetaFilter is a place for people to share their own knowledge and ideas with each other, not to interact with algorithms, so using ChatGPT or other generative AI tools is discouraged”, I don’t believe it is talking about using AI on your own time.
posted by Lemkin at 10:14 AM on April 24 [3 favorites]
When the FAQ says “MetaFilter is a place for people to share their own knowledge and ideas with each other, not to interact with algorithms, so using ChatGPT or other generative AI tools is discouraged”, I don’t believe it is talking about using AI on your own time.
posted by Lemkin at 10:14 AM on April 24 [3 favorites]
If I want an answer from AI, I'll ask AI. That will be when I'm ice-skating in hell and need tips.
If you have a question to ask or a comment to make, make it yourself, don't outsource your thinking to a corrupt, fascist-funding machine.
If there's some hidden "I will test this by making AI posts and see if people spot them or engage in good faith" thing going on somewhere on site, I hope that person is permabanned the instant it becomes apparent.
If I got a sense that there was regular copying of chatGPT or AI garbage onto this site, I would leave and cancel my membership instanter.
There is literally nothing in the world that keeps me from getting AI content if I want it - indeed, I have to go to considerable effort NOT to get AI content. I pay for metafilter because it is guaranteed human. If we are not a human site, what separates us from the rest of the internet? Why should people pay for membership when they can just type some keywords into an interface and get their own custom-generated slop?
posted by Frowner at 10:40 AM on April 24 [14 favorites]
If you have a question to ask or a comment to make, make it yourself, don't outsource your thinking to a corrupt, fascist-funding machine.
If there's some hidden "I will test this by making AI posts and see if people spot them or engage in good faith" thing going on somewhere on site, I hope that person is permabanned the instant it becomes apparent.
If I got a sense that there was regular copying of chatGPT or AI garbage onto this site, I would leave and cancel my membership instanter.
There is literally nothing in the world that keeps me from getting AI content if I want it - indeed, I have to go to considerable effort NOT to get AI content. I pay for metafilter because it is guaranteed human. If we are not a human site, what separates us from the rest of the internet? Why should people pay for membership when they can just type some keywords into an interface and get their own custom-generated slop?
posted by Frowner at 10:40 AM on April 24 [14 favorites]
I add: one thing I like about metafilter is the people. There are a few that I've met in real life, there are a few I've corresponded with on metafilter or elsewhere, there are many whose posting style and concerns I recognize. If Genjiandproust or brainwane posts something about SFF, for instance, I know that I can trust their judgement. If advicepig posts something about Minneapolis or bikes, I know they're a reliable source.
I have been on metafilter for years and years now and I like that people (or at least the people who talk about the things I like to talk about) - for good or for ill - have a general sense of who I am. Maybe that means they dismiss me or maybe that means they think I'm okay on SFF but awful on politics, but they base that on what I post and the reputation I've created here.
Metafilter isn't just a place to go to get the very finest in prompt-engineering; it's a place where I can get advice from people with a history on site which helps me to evaluate what they say.
If it's going to be me typing into a hellpit of content "crafted" by AI, this place loses about 4/5 of its interest for me.
posted by Frowner at 10:46 AM on April 24 [9 favorites]
I have been on metafilter for years and years now and I like that people (or at least the people who talk about the things I like to talk about) - for good or for ill - have a general sense of who I am. Maybe that means they dismiss me or maybe that means they think I'm okay on SFF but awful on politics, but they base that on what I post and the reputation I've created here.
Metafilter isn't just a place to go to get the very finest in prompt-engineering; it's a place where I can get advice from people with a history on site which helps me to evaluate what they say.
If it's going to be me typing into a hellpit of content "crafted" by AI, this place loses about 4/5 of its interest for me.
posted by Frowner at 10:46 AM on April 24 [9 favorites]
If there's some hidden "I will test this by making AI posts and see if people spot them or engage in good faith" thing going on somewhere on site, I hope that person is permabanned the instant it becomes apparent.
There were some (now deleted) comments in the metafilter subreddit that indicated that people were trying this.
posted by bunton at 11:19 AM on April 24 [1 favorite]
There were some (now deleted) comments in the metafilter subreddit that indicated that people were trying this.
posted by bunton at 11:19 AM on April 24 [1 favorite]
There were some (now deleted) comments in the metafilter subreddit that indicated that people were trying this.
That's what I figured. I've been wondering about this because I feel like there's been an uptick in poorly written vaguely corporate-writing posts (not wanting to call anything out in specific but it is not invisible day to day by any means) by users without the kind of posting history that make them make sense. I'd tried to put it down to low quality writing, but various kinds of gotcha smuggery have crossed my mind.
I'm telling you, the more this place turns into boring, bland, mid-brain AI slop, the closer it will get to death. It might have an undead existence where AI boosters post just plausible enough to avoid deletion bot questions and answers and people get directed to the answers from google, making ad dollars or whatever, but it will kill the site itself.
I often wonder what these AI types plan to do all day once everything is AI. Are they just going to sit in a tub and stare into space, satisfied that a fully automated internet is generating and "reading" content? It just feels like real death drive stuff.
posted by Frowner at 11:39 AM on April 24 [5 favorites]
That's what I figured. I've been wondering about this because I feel like there's been an uptick in poorly written vaguely corporate-writing posts (not wanting to call anything out in specific but it is not invisible day to day by any means) by users without the kind of posting history that make them make sense. I'd tried to put it down to low quality writing, but various kinds of gotcha smuggery have crossed my mind.
I'm telling you, the more this place turns into boring, bland, mid-brain AI slop, the closer it will get to death. It might have an undead existence where AI boosters post just plausible enough to avoid deletion bot questions and answers and people get directed to the answers from google, making ad dollars or whatever, but it will kill the site itself.
I often wonder what these AI types plan to do all day once everything is AI. Are they just going to sit in a tub and stare into space, satisfied that a fully automated internet is generating and "reading" content? It just feels like real death drive stuff.
posted by Frowner at 11:39 AM on April 24 [5 favorites]
don't outsource your thinking to a corrupt, fascist-funding machine.
But my company already signed the contract! /s I am doing some training this week, and it is all about using the embedded AI. I want to learn how to use this, not be told how to use it. It is very annoying.
posted by soelo at 11:55 AM on April 24 [1 favorite]
But my company already signed the contract! /s I am doing some training this week, and it is all about using the embedded AI. I want to learn how to use this, not be told how to use it. It is very annoying.
posted by soelo at 11:55 AM on April 24 [1 favorite]
If we are not a human site, what separates us from the rest of the internet?
"My name is MetaFilter, Site of Sites" (etc)
posted by ginger.beef at 12:28 PM on April 24
"My name is MetaFilter, Site of Sites" (etc)
posted by ginger.beef at 12:28 PM on April 24
The only AI in AskMefi should be Accurate Information.
posted by star gentle uterus at 12:35 PM on April 24 [1 favorite]
posted by star gentle uterus at 12:35 PM on April 24 [1 favorite]
Add to that the fact that, as already noted, there are FPPs openly encouraging users to play with AI technology, and I'm not particularly convinced that anyone actually gives a fuck.
Going from your house to the playground is very different from bringing the playground into your house. If people want to go play (and a lot do) then that's their business.
posted by Tell Me No Lies at 12:48 PM on April 24 [5 favorites]
Going from your house to the playground is very different from bringing the playground into your house. If people want to go play (and a lot do) then that's their business.
posted by Tell Me No Lies at 12:48 PM on April 24 [5 favorites]
"My name is MetaFilter, Site of Sites" (etc)
Look upon my beans, ye mighty, and despair!
posted by Lemkin at 1:34 PM on April 24 [6 favorites]
Look upon my beans, ye mighty, and despair!
posted by Lemkin at 1:34 PM on April 24 [6 favorites]
If I want an answer from AI, I'll ask AI. That will be when I'm ice-skating in hell and need tips.
When it comes to wrong answers, I would much rather have confidently wrong answers from a human Mefite than confidently wrong answers from a stochastic parrot.
rmd1023
Ah, yep. I'm with Frowner and rmd1023. I come here to interact with real people not ChatGPT.
When it comes to being human, sometimes a thoughtful wrong answer can be more informative than a soulless right answer generated from AI.
posted by BlueHorse at 2:12 PM on April 24 [4 favorites]
When it comes to wrong answers, I would much rather have confidently wrong answers from a human Mefite than confidently wrong answers from a stochastic parrot.
rmd1023
Ah, yep. I'm with Frowner and rmd1023. I come here to interact with real people not ChatGPT.
When it comes to being human, sometimes a thoughtful wrong answer can be more informative than a soulless right answer generated from AI.
posted by BlueHorse at 2:12 PM on April 24 [4 favorites]
Mod note: One comment removed. Seriously, please avoid using ChatGPT or other LLMs on the site.
posted by Brandon Blatcher (staff) at 2:39 PM on April 24 [3 favorites]
posted by Brandon Blatcher (staff) at 2:39 PM on April 24 [3 favorites]
When I see "I asked (genAI tool name here) and here's what it said" I feel a level of intense rage and disdain at such a high level that I'm not able to explain how much I hate it. Like 'flames on my face' from Clue, or the 'hate on every nano-angstrom' from I Have No Mouth And I Must Scream level of hatred.
posted by bolda at 2:41 PM on April 24 [13 favorites]
posted by bolda at 2:41 PM on April 24 [13 favorites]
Ironically, asking ChatGPT this question gives a pretty good summary of the current site policy.
posted by ssg at 3:13 PM on April 24 [1 favorite]
posted by ssg at 3:13 PM on April 24 [1 favorite]
When I see "I asked (genAI tool name here) and here's what it said" I feel a level of intense rage and disdain at such a high level that I'm not able to explain how much I hate it.
Wow, you are going to hate pretty much every second from here to the end of you life.
posted by Tell Me No Lies at 3:28 PM on April 24 [3 favorites]
Wow, you are going to hate pretty much every second from here to the end of you life.
posted by Tell Me No Lies at 3:28 PM on April 24 [3 favorites]
I feel a level of intense rage and disdain
same, same
you are going to hate pretty much every second from here to the end of you life
that is how it's shaping up, yes
posted by working_objects at 3:41 PM on April 24 [15 favorites]
same, same
you are going to hate pretty much every second from here to the end of you life
that is how it's shaping up, yes
posted by working_objects at 3:41 PM on April 24 [15 favorites]
... oof.
You know what? You got it.
I hereby forswear this rhetorical device forever in the interest of peace between peoples.
posted by Lemkin at 5:20 PM on April 24
You know what? You got it.
I hereby forswear this rhetorical device forever in the interest of peace between peoples.
posted by Lemkin at 5:20 PM on April 24
you have a nuanced take on the matter?
mittens, these are questions from my previously there
mittens, these are questions from my previously there
[philpapers (pdf:)] Vision is always a question of the power to see – & perhaps of the violence implicit in our visualizing practices. With whose blood were my eyes crafted?posted by HearHere at 6:08 PM on April 24 [2 favorites]
👀
These technologies are ways of life, social orders, practices of visualization. Technologies are skilled practices. How to see? Where to see from? What limits to vision? What to see for? Whom to see with? Who gets to have more than one point of view? Who gets blinded? Who wears blinders? Who interprets the visual field? What other sensory powers do we wish to cultivate besides vision?
For those trying to avoid AI in their google results, there is a trick where you can just set it up once and it works consistently, you don't have to swear, and you don't even get 'sponsored' results at the top.
Instructions here - https://tenbluelinks.org/. I'm moderately computer savvy and it took me about one minute to do this for each of the three browsers I have on each of my various devices (okay two of them are different flavours of Firefox but still). And it works really well.
posted by ngaiotonga at 2:04 AM on April 25 [12 favorites]
Instructions here - https://tenbluelinks.org/. I'm moderately computer savvy and it took me about one minute to do this for each of the three browsers I have on each of my various devices (okay two of them are different flavours of Firefox but still). And it works really well.
posted by ngaiotonga at 2:04 AM on April 25 [12 favorites]
How to see? Where to see from? What limits to vision?
I wish I'd had time to participate in your original Haraway thread, HearHere. I feel that I would have made a long, long comment--one that is totally inappropriate for this metatalk thread about AI where we all seem to agree with one another (mirabile dictu!).
But what I would have said is something like, isn't it funny how a feminist essay from 1988 turns out to be simply...right?
Think about how many people have rolled their eyes at the silliness of the idea of embodied, situational knowledge--knowledge 'written on the body,' god, what a relic of a stupider age--only to come to understand that that's literally how knowledge works. We don't know anything until it's written in our synapses, until our neurons connect in a certain way, and the sureness of that knowledge is reflected in the strength and repetition of those connections. And sometimes you will have wanted to know something that you cannot know, because your neurons took another path, so you are always living in a kind of self-imposed ignorance--not even self-imposed, as it's happening at a level far lower than something we could call a self.
And vision itself! The history of the science of vision is finding out just how little we see. How full of blind, blurry, colorless spots! The jitteriness embedded in a sense that is so weirdly set up that if it sits still it ceases to function. Vision is movement about light. How much of the brain is devoted to trying to hallucinate a world out of these brief fleeting grainy glimpses. What a joke to posit a god's-eye view.
To use vision as a metaphor, you either make it complicated, partial and true; or objective, simple, and dumb and wrong.
One would then say that most of our AI discourse is marred by a similar error, assuming a stability and concreteness to intelligence that does not exist. We are hallucinating all the time, and we make a mistake criticizing AI for doing the same.
(actually i probably would have quoted this bit above what you just quoted, about the impossibility of entertaining innocent 'identity' politics and epistemologies as strategies for seeing from the standpoints of the subjugated in order to see well and then everybody would've gotten mad at me. it's kind of funny that nobody in that thread got down to the 'objects as actors' part, otherwise we could've had another Metafilter Is Bad At Gender discussion!)
posted by mittens at 7:22 AM on April 25 [2 favorites]
I wish I'd had time to participate in your original Haraway thread, HearHere. I feel that I would have made a long, long comment--one that is totally inappropriate for this metatalk thread about AI where we all seem to agree with one another (mirabile dictu!).
But what I would have said is something like, isn't it funny how a feminist essay from 1988 turns out to be simply...right?
Think about how many people have rolled their eyes at the silliness of the idea of embodied, situational knowledge--knowledge 'written on the body,' god, what a relic of a stupider age--only to come to understand that that's literally how knowledge works. We don't know anything until it's written in our synapses, until our neurons connect in a certain way, and the sureness of that knowledge is reflected in the strength and repetition of those connections. And sometimes you will have wanted to know something that you cannot know, because your neurons took another path, so you are always living in a kind of self-imposed ignorance--not even self-imposed, as it's happening at a level far lower than something we could call a self.
And vision itself! The history of the science of vision is finding out just how little we see. How full of blind, blurry, colorless spots! The jitteriness embedded in a sense that is so weirdly set up that if it sits still it ceases to function. Vision is movement about light. How much of the brain is devoted to trying to hallucinate a world out of these brief fleeting grainy glimpses. What a joke to posit a god's-eye view.
To use vision as a metaphor, you either make it complicated, partial and true; or objective, simple, and dumb and wrong.
One would then say that most of our AI discourse is marred by a similar error, assuming a stability and concreteness to intelligence that does not exist. We are hallucinating all the time, and we make a mistake criticizing AI for doing the same.
(actually i probably would have quoted this bit above what you just quoted, about the impossibility of entertaining innocent 'identity' politics and epistemologies as strategies for seeing from the standpoints of the subjugated in order to see well and then everybody would've gotten mad at me. it's kind of funny that nobody in that thread got down to the 'objects as actors' part, otherwise we could've had another Metafilter Is Bad At Gender discussion!)
posted by mittens at 7:22 AM on April 25 [2 favorites]
If the idea behind Ask is that people are looking for answers from people with relevant experience, people chiming in with “I asked an LLM for an answer, and it said this” all that different from (the already discouraged) people chiming in to say they have no relevant experience or knowledge, yet still offering their unhelpful advice all the same?
Ask is for people seeking help from other people with relevant experience and expertise. If they wanted to use an LLM, they would have. Ask serves a purpose that LLMs don’t, and people trying to shoehorn them in aren’t really doing anyone a service.
posted by Ghidorah at 8:04 AM on April 25 [4 favorites]
Ask is for people seeking help from other people with relevant experience and expertise. If they wanted to use an LLM, they would have. Ask serves a purpose that LLMs don’t, and people trying to shoehorn them in aren’t really doing anyone a service.
posted by Ghidorah at 8:04 AM on April 25 [4 favorites]
For those trying to avoid AI in their google results, there is a trick where you can just set it up once and it works consistently, you don't have to swear, and you don't even get 'sponsored' results at the top.
I finally got around to doing this and it's like having Ye Olde Google back. Hooray!
But it only works when you start a search from the browser's search field; if you try to refine the search from within the Google page that results, you'll start seeing AI crap again. Still, better than nothing.
posted by rory at 8:16 AM on April 25 [1 favorite]
I finally got around to doing this and it's like having Ye Olde Google back. Hooray!
But it only works when you start a search from the browser's search field; if you try to refine the search from within the Google page that results, you'll start seeing AI crap again. Still, better than nothing.
posted by rory at 8:16 AM on April 25 [1 favorite]
Absolutely fucking not. I come here to get away from the rest of the internet.
posted by reedbird_hill at 10:52 AM on April 25 [1 favorite]
posted by reedbird_hill at 10:52 AM on April 25 [1 favorite]
What about making ai.metafilter.com, where posters must first fail a CAPTCHA in order to post, and shunting all the useless drivel there? /s
posted by axiom at 10:59 AM on April 25 [1 favorite]
posted by axiom at 10:59 AM on April 25 [1 favorite]
You are not logged in, either login or create an account to post comments
posted by Brandon Blatcher (staff) at 8:36 PM on April 23 [24 favorites]