If this is a subject you’re interested in, there are some decent links in the article, but the article itself has a lot of signs of being AI-generated.
If this is a subject you’re interested in, there are some decent links in the article, but the article itself has a lot of signs of being AI-generated.
It's not as easy as you might think. A few years ago, I tried using Midjourney to generate pictures for all my blog posts, because I thought they looked pretty bare without images. It took me lots...
It's not as easy as you might think. A few years ago, I tried using Midjourney to generate pictures for all my blog posts, because I thought they looked pretty bare without images. It took me lots of tries to generate images that I somewhat liked and were appropriate to the theme of the article.
At this point some of them look pretty cringe, but that's mostly because I've seen too much in a similar style since then.
If an article is written and adorned using images based off the works of others, it's without care. As a visual artist, AI images are not used by people or business I want to associate with. They...
If an article is written and adorned using images based off the works of others, it's without care. As a visual artist, AI images are not used by people or business I want to associate with. They do not care about me or swaths of other people that were used to train their models. They do not care about how they are perceived for using stolen work.
In the case of clip art, that's its intended purpose. In the case of memes, I'd say the majority of them are transformative in a meaningful way. Using someone's art to train an AI to replace them...
In the case of clip art, that's its intended purpose. In the case of memes, I'd say the majority of them are transformative in a meaningful way. Using someone's art to train an AI to replace them is not meaningfully transformative.
It's not transformative if you just copy a meme. I mean, nobody worries about whether it's copyright infringement (although it might be), but from an artistic standpoint, it's still pretty...
It's not transformative if you just copy a meme. I mean, nobody worries about whether it's copyright infringement (although it might be), but from an artistic standpoint, it's still pretty thoughtless.
Ah, so you're talking about finding a meme somewhere that conveys an idea relevant to the subject and putting it in as a visual element, not making a meme yourself. In that case, I'd argue that...
Ah, so you're talking about finding a meme somewhere that conveys an idea relevant to the subject and putting it in as a visual element, not making a meme yourself. In that case, I'd argue that the creation of the meme is creative and transformative, and the meme should be attributed in any quasi-journalistic work if possible. "Thoughtless" isn't categorical here, since it can take a fair bit of sifting to find le meme juste, to coin a phrase. Additionally, most memes are created and published anonymously, because the manner of distribution is itself a major part of the artistic medium. "I collaged this thing together specifically for you to throw into online conversations" is kinda the raison d'etre of the whole medium.
I don't see how one's judgment of the artistic merit of a work pertains, but for what it's worth, I wouldn't think too highly of a blog that exclusively used memes scrounged from various Internet sources either. I would still rate a blog that leans on AI-generated images lower though, because there are so many serious concerns about environmental and societal damage being caused by the industry that to blithely ignore them for the convenience of having easy access to bland and weirdly uncanny visual elements strikes me as scummy. It feels not so much like rearranging deck chairs or striking up the band on the Titanic, as taking this whole ship sinking thing as an excuse to sneak off to steerage for a quick wank.
Anti-AI populists are pretty quick to single out AI as being particularly bad for the environment when it doesn't seem worse than a lot of other things people do every day, like, say, driving or...
Anti-AI populists are pretty quick to single out AI as being particularly bad for the environment when it doesn't seem worse than a lot of other things people do every day, like, say, driving or eating meat. I think there's a lot that could be done to reduce the environment impact of AI, but when it's used as a reason to stigmatize AI-generated images, I'm fine with ignoring that.
That's a pretty blatant whataboutism, but let's humor it anyway. Driving gets a person where they need or want to be, and eating meat can sustain them. Those are both pretty important things, and...
... it doesn't seem worse than a lot of other things people do every day, like, say, driving or eating meat.
That's a pretty blatant whataboutism, but let's humor it anyway.
Driving gets a person where they need or want to be, and eating meat can sustain them. Those are both pretty important things, and while they could be accomplished more sustainably, they need to happen somehow. Are you prepared to claim the same for generated images? Hell, take it further: could you say the same about coding agents, the most persuasive use-case of LLMs and related technologies?
More importantly, society has been structured to fulfill transportation and food needs in a particular way, and the specifics of how those needs are provided for were determined by the exigencies of history and the limited understanding of the people who furthered development at various times. It's a fine and necessary thing to note that cattle ranching has a deleterious environmental effect, but you also have to acknowledge thousands of years of human history that's wrapped up with beef as a food source, and decades or centuries of infrastructure development devoted to providing it as cheaply at the point-of-purchase as possible. Those don't excuse the problems, but they should be taken into account.
What's happening right now is different: a powerful subset of society is propelling development of these tools in ways that are pretty categorically unsustainable (in multiple senses), despite well-founded, logical criticisms. It's disingenuous to equate the current development of a technology in full understanding of its drawbacks to systems that were built up for decades, centuries or millennia without that same understanding.
If you're simply looking for reasons to disregard criticisms, you're welcome to have that one, though it's not what I would call a respectable position.
Environmental costs are about costs and they need to be measured or at least estimated to do it right. We need to be doing comparisons between different approaches. One way to do this properly...
Environmental costs are about costs and they need to be measured or at least estimated to do it right. We need to be doing comparisons between different approaches. One way to do this properly might be to compare making images in different ways. I’m not going to do the study myself, but the environment costs of things like film or art supplies might be a relevant comparison.
So is this better or worse than just downloading a random piece of art off Google Images and slapping it on there with no credit, the previous stage of the art in half-arsed blog post illustration?
So is this better or worse than just downloading a random piece of art off Google Images and slapping it on there with no credit, the previous stage of the art in half-arsed blog post illustration?
I mean, what if were? Honestly, there are so AI-generated text nowadays, even from people who do know how to write, that… whatever. There are a ton of awful text that would be better with some AI...
I mean, what if were? Honestly, there are so AI-generated text nowadays, even from people who do know how to write, that… whatever. There are a ton of awful text that would be better with some AI guidance.
For me it’s less about whether it’s actually AI generated and more about whether the sentence patterns are distracting because I’ve learned to recognize them from using AI. Certain phrases become...
For me it’s less about whether it’s actually AI generated and more about whether the sentence patterns are distracting because I’ve learned to recognize them from using AI. Certain phrases become cliches through overuse, and AI accelerates that process.
It’s like how good special effects in movies are the ones you don’t notice.
That's interesting! I usually ignore the quality of the text when reading something for the information it contains (except when really low). When reading for pleasure, even non-fiction, I'm more...
That's interesting! I usually ignore the quality of the text when reading something for the information it contains (except when really low). When reading for pleasure, even non-fiction, I'm more concerned about what you mentioned.
Signs of AI writing are to me a huge warning that the information being presented can't be trusted. I absolutely skip out of most such articles, I've got better things to do.
Signs of AI writing are to me a huge warning that the information being presented can't be trusted. I absolutely skip out of most such articles, I've got better things to do.
In this case, even if the author used AI assistance to write, they provide external links that validate their argument. I thought that to be good enough.
In this case, even if the author used AI assistance to write, they provide external links that validate their argument. I thought that to be good enough.
I understand it's good enough for you, but simply saying "that should be good enough" doesn't make that so, and certainly doesn't define that line for other folks. It certainly lessens my interest...
I understand it's good enough for you, but simply saying "that should be good enough" doesn't make that so, and certainly doesn't define that line for other folks. It certainly lessens my interest in what is functionally "old news".
Why bring this old topic up again now if you're not going to put in human effort for my attention? Or if they write in a way I consider to be lower quality, why do I as a person want to read their article on the topic.
I don't think it's so binary. I often spend hours cowriting things with an LLM. I double-check the tone and accuracy and stand behind it. But I also don't try to hide that I used AI for it and...
if you're not going to put in human effort for my attention
I don't think it's so binary. I often spend hours cowriting things with an LLM. I double-check the tone and accuracy and stand behind it. But I also don't try to hide that I used AI for it and there are certainly AI tells left behind.
I share your annoyance though when someone doesn't put in the effort and just presents something that they didn't take the time to double check or polish at all. I just didn't get that sense from this particular article.
It depends some for me but I really don't like the use of LLMs and hate that I'm being asked to use them at work. I am sure there are things I can't clock. But when I do, I immediately will feel...
It depends some for me but I really don't like the use of LLMs and hate that I'm being asked to use them at work.
I am sure there are things I can't clock. But when I do, I immediately will feel like my time was wasted. Particularly if it's something intended to be persuasive or to educate me, it just feels like the writer just doesn't actually care if I read it. I could just have an LLM tell me what it said in that case, ya know?
I am not speaking to your writing or about you specifically fwiw, just generally. I really like learning and reading new perspectives, and I think what gets me the most is that if you just regurgitate the LLM, I could have looked that up myself. Posting an article or blog or whatever with that content is just pointless.
Well, because it was new for me and I thought could be for more people? I guess you're overcomplicating a detail that shouldn't be the discussion focus…
Well, because it was new for me and I thought could be for more people? I guess you're overcomplicating a detail that shouldn't be the discussion focus…
I meant why did the author bring it up now In those circumstances I understand you found it useful, and that you think potential AI use - and correlated, writing quality and ethics - is not a...
I meant why did the author bring it up now In those circumstances
I understand you found it useful, and that you think potential AI use - and correlated, writing quality and ethics - is not a relevant conversation point.
For me those things matter, but also I wasn't who you were talking to before so I don't think I'm complicating anything. Any discussion will necessarily add aspects and layers and tangents.
Further evidence, is that I constantly gripe at friends who share links that have tracking URL bits on them. The average user is savvy enough to just copy-paste a URL, but the technobabble that...
Further evidence, is that I constantly gripe at friends who share links that have tracking URL bits on them. The average user is savvy enough to just copy-paste a URL, but the technobabble that are tracking URL params is indecipherable to them, and they just leave them.
The article focuses on share buttons specifically, but I almost never see these anymore. Instead, I see icon links to social media pages. I believe this still has some value today: If users...
The article focuses on share buttons specifically, but I almost never see these anymore. Instead, I see icon links to social media pages. I believe this still has some value today:
If users subscribe to a social page, it reminds them of the company on a regular basis, potentially driving sales.
It acts as a social proof, showing the company is not fly-by-night and can be trusted.
It clarifies that these are your social pages, which discourages any imitators that might be trying to scam with a knockoff account.
This is the approach I've taken even since share buttons were popular. I've never used a plugin or any company's offered JavaScript code. Those are slow and ugly. A few tiny SVGs in a common style in the header or footer is all you've ever needed.
I always assumed the whole point was to track users, not for them to actually click “share”, and have website owners believe that it would make their content go viral.
I always assumed the whole point was to track users, not for them to actually click “share”, and have website owners believe that it would make their content go viral.
Yeah, that's the goal for platforms. But what drawn site owners to add them is to facilitate sharing of their content, which explain simpler buttons without platforms' tracking.
Yeah, that's the goal for platforms. But what drawn site owners to add them is to facilitate sharing of their content, which explain simpler buttons without platforms' tracking.
If this is a subject you’re interested in, there are some decent links in the article, but the article itself has a lot of signs of being AI-generated.
The image at the top is certainly AI generated. That's enough for me to skip reading the article.
I don’t read blogs for the art.
They're choosing to represent their article with a piece of art. If that art is chosen without care, I expect the article to be written without care.
It's not as easy as you might think. A few years ago, I tried using Midjourney to generate pictures for all my blog posts, because I thought they looked pretty bare without images. It took me lots of tries to generate images that I somewhat liked and were appropriate to the theme of the article.
At this point some of them look pretty cringe, but that's mostly because I've seen too much in a similar style since then.
If an article is written and adorned using images based off the works of others, it's without care. As a visual artist, AI images are not used by people or business I want to associate with. They do not care about me or swaths of other people that were used to train their models. They do not care about how they are perceived for using stolen work.
Do you have the same reaction to memes or clip art, though? That’s also someone else’s work.
In the case of clip art, that's its intended purpose. In the case of memes, I'd say the majority of them are transformative in a meaningful way. Using someone's art to train an AI to replace them is not meaningfully transformative.
It's not transformative if you just copy a meme. I mean, nobody worries about whether it's copyright infringement (although it might be), but from an artistic standpoint, it's still pretty thoughtless.
Ah, so you're talking about finding a meme somewhere that conveys an idea relevant to the subject and putting it in as a visual element, not making a meme yourself. In that case, I'd argue that the creation of the meme is creative and transformative, and the meme should be attributed in any quasi-journalistic work if possible. "Thoughtless" isn't categorical here, since it can take a fair bit of sifting to find le meme juste, to coin a phrase. Additionally, most memes are created and published anonymously, because the manner of distribution is itself a major part of the artistic medium. "I collaged this thing together specifically for you to throw into online conversations" is kinda the raison d'etre of the whole medium.
I don't see how one's judgment of the artistic merit of a work pertains, but for what it's worth, I wouldn't think too highly of a blog that exclusively used memes scrounged from various Internet sources either. I would still rate a blog that leans on AI-generated images lower though, because there are so many serious concerns about environmental and societal damage being caused by the industry that to blithely ignore them for the convenience of having easy access to bland and weirdly uncanny visual elements strikes me as scummy. It feels not so much like rearranging deck chairs or striking up the band on the Titanic, as taking this whole ship sinking thing as an excuse to sneak off to steerage for a quick wank.
Anti-AI populists are pretty quick to single out AI as being particularly bad for the environment when it doesn't seem worse than a lot of other things people do every day, like, say, driving or eating meat. I think there's a lot that could be done to reduce the environment impact of AI, but when it's used as a reason to stigmatize AI-generated images, I'm fine with ignoring that.
That's a pretty blatant whataboutism, but let's humor it anyway.
Driving gets a person where they need or want to be, and eating meat can sustain them. Those are both pretty important things, and while they could be accomplished more sustainably, they need to happen somehow. Are you prepared to claim the same for generated images? Hell, take it further: could you say the same about coding agents, the most persuasive use-case of LLMs and related technologies?
More importantly, society has been structured to fulfill transportation and food needs in a particular way, and the specifics of how those needs are provided for were determined by the exigencies of history and the limited understanding of the people who furthered development at various times. It's a fine and necessary thing to note that cattle ranching has a deleterious environmental effect, but you also have to acknowledge thousands of years of human history that's wrapped up with beef as a food source, and decades or centuries of infrastructure development devoted to providing it as cheaply at the point-of-purchase as possible. Those don't excuse the problems, but they should be taken into account.
What's happening right now is different: a powerful subset of society is propelling development of these tools in ways that are pretty categorically unsustainable (in multiple senses), despite well-founded, logical criticisms. It's disingenuous to equate the current development of a technology in full understanding of its drawbacks to systems that were built up for decades, centuries or millennia without that same understanding.
If you're simply looking for reasons to disregard criticisms, you're welcome to have that one, though it's not what I would call a respectable position.
Environmental costs are about costs and they need to be measured or at least estimated to do it right. We need to be doing comparisons between different approaches. One way to do this properly might be to compare making images in different ways. I’m not going to do the study myself, but the environment costs of things like film or art supplies might be a relevant comparison.
So is this better or worse than just downloading a random piece of art off Google Images and slapping it on there with no credit, the previous stage of the art in half-arsed blog post illustration?
Why does that have to be a choice? Both suck.
I mean, what if were? Honestly, there are so AI-generated text nowadays, even from people who do know how to write, that… whatever. There are a ton of awful text that would be better with some AI guidance.
I articulated my point of view in this article, if anyone is interested.
For me it’s less about whether it’s actually AI generated and more about whether the sentence patterns are distracting because I’ve learned to recognize them from using AI. Certain phrases become cliches through overuse, and AI accelerates that process.
It’s like how good special effects in movies are the ones you don’t notice.
That's interesting! I usually ignore the quality of the text when reading something for the information it contains (except when really low). When reading for pleasure, even non-fiction, I'm more concerned about what you mentioned.
Signs of AI writing are to me a huge warning that the information being presented can't be trusted. I absolutely skip out of most such articles, I've got better things to do.
In this case, even if the author used AI assistance to write, they provide external links that validate their argument. I thought that to be good enough.
I understand it's good enough for you, but simply saying "that should be good enough" doesn't make that so, and certainly doesn't define that line for other folks. It certainly lessens my interest in what is functionally "old news".
Why bring this old topic up again now if you're not going to put in human effort for my attention? Or if they write in a way I consider to be lower quality, why do I as a person want to read their article on the topic.
I don't think it's so binary. I often spend hours cowriting things with an LLM. I double-check the tone and accuracy and stand behind it. But I also don't try to hide that I used AI for it and there are certainly AI tells left behind.
I share your annoyance though when someone doesn't put in the effort and just presents something that they didn't take the time to double check or polish at all. I just didn't get that sense from this particular article.
It depends some for me but I really don't like the use of LLMs and hate that I'm being asked to use them at work.
I am sure there are things I can't clock. But when I do, I immediately will feel like my time was wasted. Particularly if it's something intended to be persuasive or to educate me, it just feels like the writer just doesn't actually care if I read it. I could just have an LLM tell me what it said in that case, ya know?
I am not speaking to your writing or about you specifically fwiw, just generally. I really like learning and reading new perspectives, and I think what gets me the most is that if you just regurgitate the LLM, I could have looked that up myself. Posting an article or blog or whatever with that content is just pointless.
Well, because it was new for me and I thought could be for more people? I guess you're overcomplicating a detail that shouldn't be the discussion focus…
I meant why did the author bring it up now In those circumstances
I understand you found it useful, and that you think potential AI use - and correlated, writing quality and ethics - is not a relevant conversation point.
For me those things matter, but also I wasn't who you were talking to before so I don't think I'm complicating anything. Any discussion will necessarily add aspects and layers and tangents.
Further evidence, is that I constantly gripe at friends who share links that have tracking URL bits on them. The average user is savvy enough to just copy-paste a URL, but the technobabble that are tracking URL params is indecipherable to them, and they just leave them.
I just don't understand how they can share a 300-character URL that uses plaintext to show what it's tracking without even being curious.
I imagine they think it's just computer stuff. Doesn't matter as long as the click takes one to the intended destination.
The article focuses on share buttons specifically, but I almost never see these anymore. Instead, I see icon links to social media pages. I believe this still has some value today:
This is the approach I've taken even since share buttons were popular. I've never used a plugin or any company's offered JavaScript code. Those are slow and ugly. A few tiny SVGs in a common style in the header or footer is all you've ever needed.
Yeah, links to social profiles kinda worth it. The article focuses on sharing links, however — and I agree with the author.
I always assumed the whole point was to track users, not for them to actually click “share”, and have website owners believe that it would make their content go viral.
Yeah, that's the goal for platforms. But what drawn site owners to add them is to facilitate sharing of their content, which explain simpler buttons without platforms' tracking.
Related: https://neal.fun/share-this-page/
Instructions unclear, shared with whales, am being eaten by orcas