I asked ChatGPT to write a comment, but it was too annoying for me to inflict on you even as a joke.
Seriously: great post. AI is going to be a huge deal for online communities that aren't paying attention, and anyone that wants to maintain real interactions is going to have to be that much more vigilant.
Adding to what you said, because I agree with the need for discretion and vigilance.
I think anonymity in large communities full of strangers, such as substack, denies the kinds of personal accountability that would make someone think twice about acting in bad faith — assuming the rules are clear.
At the same time, there’s good reason for anonymity given that you can be held accountable for controversial speech if it’s inconvenient to your employer under an at will arrangement.
Please consider using a different motto for your Substack, the one that I'm seeing (" ideas so good they kill themselves") seems oddly gratuitously repellant.
Ahahaha I recognise that LLM spammer account! I received a comment from it a couple of weeks ago and blocked it after doing a similar search to you (I didn't think of reporting it for spam). I definitely see people interacting with its comments; I wonder how many blocks/deletions it accumulates.
After looking at their account, I think a lot of people have gotten comments from it. I’d also love to know the stats on their blocks/subscribes/reports
Same here. I thought it was legit when the comment appeared in a recent post of mine because I had seen it join in a conversation with writers I know to be real! The weird thing is that if you search for it, you get the newsletter's homepage, but if you click on the comment you get sent to the Profile Not Found page.
Turns out the comment I was really excited about on my most recent post was this same user. And I thought I was decently good at recognizing AI slop. I’m curious if this user is using an LLM to help compose comments in a large number from articles they have genuinely read or if they are feeding our articles into an LLM. The first is still bad but the second is a huge invasion—especially since (I assume) most of us have spent a lot of energy trying not to have our IP stolen or used to train ChatGPT. 😒
There is no reason to think that a human is actually reading the articles unfortunately. Look at this one, and unhinged ramble of nonsense because the article was about word puzzles and the script had no idea how to handle it
I really enjoyed what you wrote. Personally my take on AI as a whole is muddled and mixed and unformed… I find that many people on this platform are either very pro or vary anti ai and I thinks that there is more grey than we think.
That being said, when I do come hear to read I look for that human interaction and reflection (as do many others I think- this is the only “social” media I still consume and I left the others in the hope of escaping the algorithm and posts filled by people who aren’t real) and it disappoints me on the rare occasion that I “fall” and read an article that ended up being ai.
Either way the last paragraph- leaving your own words behind- is a beautiful sentiment and I don’t think it’s too late. Esp if you don’t look at it as something that will paint a picture of you for the future but rather something that shows people (in this case people like me, your readers) who you are right now
Thank you for your kind words. You made so many great point.
I also have a middle-road position on AI, but I think if we aren’t careful with its adoption we will lose things we took for granted. I totally feel the betrayal when you are expecting a human voice, but you end up reading a machine. It just feels so hollow.
I feel you on the loneliness of trying to be a writer, or even just exchanging ideas on social media. Social media has always had problems, but in the 2010s version it was much easier to get some attention and start building some sort of community and engagement with your ideas.
My old tumblr account had dozens of followers, many of whom actively engaged with what I wrote, within a few months. And I had a similar experience during my one stint of being active on Twitter. It was also a lot easier to get responses from bigger accounts and writers even as a reply guy. I’ve been trying to be more active on Substack lately, and to lesser extent Twitter, and it’s not the same.
Slop has made big accounts wary of engaging with new and small accounts, and that’s when it simply doesn’t drown us out.
The other detrimental part is that the more AI slop fills Substack, the less that real authors and readers can appreciate each other. A couple weeks back I was actually in a rush to get out a post about how I'm not using LLMs, my voice is my voice ( https://ramblingafter.substack.com/p/voice ) because I wanted so badly for people to understand and have confidence that my blog isn't bullshit.
>> We aren’t surprised when a Yale professor like Dr. Daniel Greco, who writes brilliantly but infrequently, is lapped in influence by someone like Bentham's Bulldog
Bentham's Bulldog is the reason I have any subscribers at all outside of real life acquaintances, so I should feel grateful to him, and the dude's a nice guy, but he's so dang WRONG about the whole SIA thing. My next post is actually going to be one that explains what's incorrect about a recent post of his ( https://benthams.substack.com/p/you-need-self-locating-evidence ), but I really hope he gives me the time of day again.
It's so annoying to see someone else rake in all the subscribers when you're the one with truth on your side and they're not...
Totally agree & understand. BB’s value comes from his publishing cadence (I could never), and from occasionally making good points. Respect it in a lot of ways, but usually just feel like social media optimizes for the wrong thing.
Fascinating. Thanks for calling it out. Getting informed on the AI traps that exist around us is the first step in gaining agency and learning how to deal with them
This is unsettling. New writers are vulnerable (I’ve been on Substack less than a month), and this kind of AI “validation” feels especially predatory. Really appreciate you calling this out and pushing for boundaries before it becomes normalized. I’ll keep an eye out.
I also have had a different account do this to my account but only my account, all their activity is liking and commenting on my stories. I am a bit worried and unsure if they are a real person or not because an account that seems to be them followed me on another platform and messaged to tell me it’s them. I thought it was just someone lonely but not eloquent enough to write comments so they used ChatGTP, and I ignored them. Some people do it to get more traffic for sure but what I think is sadder is that some people may use it in place of social skill/confidence.
I had a similar experience with this same bot account. Instead I had noticed his comment on a really weird theological essay that was purportedly Lutheran but contained no real Lutheran theology. This commenter though lauded his argument and use of scripture (while I knew this was an essay incorrectly interpreting the Bible from the Lutheran lens as the author purported). This commenter erroneously added credibility to this unsound essay. )For context it was on how the Bible apparently supports alt right ideals).
I got a comment on my post by NF, too. It's been making rounds, commenting on essays of other writers I read. Could it just be a person behind the account, one who can't really string a sentence without filtering it through AI? Sadly, those exist, sigh.
I think it is a person behind the account, but they are using spam/AI comments on hundreds of accounts per day and tricking some percent of the users into believing that they’re a real person
It's a crisis. Optimistically, we combine face-to-face contacts, old-fashioned social networks, and technological systems for identity verification (plus maybe trust in legacy institutions) to separate the human wheat from the AI chaff. But people need to care about the difference.
One problem being that the norms of civil conversation go rather strongly against saying that you think a post is coming from a bot. Plus the trolls are already making bot accusations.
I think Substack has been trying to reign these accounts in, though they’ve not told me anything or responded to any of my reports. Have you had any problems with this in the last couple weeks?
Yeah that looks like the same script that was run on mine. Wish Substack would say something about it but I think they might be banning at least the smaller accounts doing it if they get reported
It’s not something I’ve noticed, but I hope you’re right! I think it would be nice if they were more transparent about their algorithm, but of course they may not really know what’s inside the black box either.
I asked ChatGPT to write a comment, but it was too annoying for me to inflict on you even as a joke.
Seriously: great post. AI is going to be a huge deal for online communities that aren't paying attention, and anyone that wants to maintain real interactions is going to have to be that much more vigilant.
Thank you so much! I thought about running the same script on my post and commenting under it, but figured I’d let someone else do it.
Adding to what you said, because I agree with the need for discretion and vigilance.
I think anonymity in large communities full of strangers, such as substack, denies the kinds of personal accountability that would make someone think twice about acting in bad faith — assuming the rules are clear.
At the same time, there’s good reason for anonymity given that you can be held accountable for controversial speech if it’s inconvenient to your employer under an at will arrangement.
Please consider using a different motto for your Substack, the one that I'm seeing (" ideas so good they kill themselves") seems oddly gratuitously repellant.
Ahahaha I recognise that LLM spammer account! I received a comment from it a couple of weeks ago and blocked it after doing a similar search to you (I didn't think of reporting it for spam). I definitely see people interacting with its comments; I wonder how many blocks/deletions it accumulates.
After looking at their account, I think a lot of people have gotten comments from it. I’d also love to know the stats on their blocks/subscribes/reports
Same here. I thought it was legit when the comment appeared in a recent post of mine because I had seen it join in a conversation with writers I know to be real! The weird thing is that if you search for it, you get the newsletter's homepage, but if you click on the comment you get sent to the Profile Not Found page.
Turns out the comment I was really excited about on my most recent post was this same user. And I thought I was decently good at recognizing AI slop. I’m curious if this user is using an LLM to help compose comments in a large number from articles they have genuinely read or if they are feeding our articles into an LLM. The first is still bad but the second is a huge invasion—especially since (I assume) most of us have spent a lot of energy trying not to have our IP stolen or used to train ChatGPT. 😒
Sorry to hear that :/
There is no reason to think that a human is actually reading the articles unfortunately. Look at this one, and unhinged ramble of nonsense because the article was about word puzzles and the script had no idea how to handle it
I really enjoyed what you wrote. Personally my take on AI as a whole is muddled and mixed and unformed… I find that many people on this platform are either very pro or vary anti ai and I thinks that there is more grey than we think.
That being said, when I do come hear to read I look for that human interaction and reflection (as do many others I think- this is the only “social” media I still consume and I left the others in the hope of escaping the algorithm and posts filled by people who aren’t real) and it disappoints me on the rare occasion that I “fall” and read an article that ended up being ai.
Either way the last paragraph- leaving your own words behind- is a beautiful sentiment and I don’t think it’s too late. Esp if you don’t look at it as something that will paint a picture of you for the future but rather something that shows people (in this case people like me, your readers) who you are right now
Thank you for your kind words. You made so many great point.
I also have a middle-road position on AI, but I think if we aren’t careful with its adoption we will lose things we took for granted. I totally feel the betrayal when you are expecting a human voice, but you end up reading a machine. It just feels so hollow.
It’s really sad, and we do need to take action.
I feel you on the loneliness of trying to be a writer, or even just exchanging ideas on social media. Social media has always had problems, but in the 2010s version it was much easier to get some attention and start building some sort of community and engagement with your ideas.
My old tumblr account had dozens of followers, many of whom actively engaged with what I wrote, within a few months. And I had a similar experience during my one stint of being active on Twitter. It was also a lot easier to get responses from bigger accounts and writers even as a reply guy. I’ve been trying to be more active on Substack lately, and to lesser extent Twitter, and it’s not the same.
Slop has made big accounts wary of engaging with new and small accounts, and that’s when it simply doesn’t drown us out.
The other detrimental part is that the more AI slop fills Substack, the less that real authors and readers can appreciate each other. A couple weeks back I was actually in a rush to get out a post about how I'm not using LLMs, my voice is my voice ( https://ramblingafter.substack.com/p/voice ) because I wanted so badly for people to understand and have confidence that my blog isn't bullshit.
>> We aren’t surprised when a Yale professor like Dr. Daniel Greco, who writes brilliantly but infrequently, is lapped in influence by someone like Bentham's Bulldog
Bentham's Bulldog is the reason I have any subscribers at all outside of real life acquaintances, so I should feel grateful to him, and the dude's a nice guy, but he's so dang WRONG about the whole SIA thing. My next post is actually going to be one that explains what's incorrect about a recent post of his ( https://benthams.substack.com/p/you-need-self-locating-evidence ), but I really hope he gives me the time of day again.
It's so annoying to see someone else rake in all the subscribers when you're the one with truth on your side and they're not...
...thanks for letting me vent, hah.
Totally agree & understand. BB’s value comes from his publishing cadence (I could never), and from occasionally making good points. Respect it in a lot of ways, but usually just feel like social media optimizes for the wrong thing.
“When I was so young, two weeks ago…” aieeeee! This is so depressing. But I’m grateful you brought it out.
Thanks Susie! If it helps, I figured out how to sometimes engage actual humans on this app, so robots are no longer making me depressed.
Fascinating. Thanks for calling it out. Getting informed on the AI traps that exist around us is the first step in gaining agency and learning how to deal with them
This is unsettling. New writers are vulnerable (I’ve been on Substack less than a month), and this kind of AI “validation” feels especially predatory. Really appreciate you calling this out and pushing for boundaries before it becomes normalized. I’ll keep an eye out.
I also have had a different account do this to my account but only my account, all their activity is liking and commenting on my stories. I am a bit worried and unsure if they are a real person or not because an account that seems to be them followed me on another platform and messaged to tell me it’s them. I thought it was just someone lonely but not eloquent enough to write comments so they used ChatGTP, and I ignored them. Some people do it to get more traffic for sure but what I think is sadder is that some people may use it in place of social skill/confidence.
I had a similar experience with this same bot account. Instead I had noticed his comment on a really weird theological essay that was purportedly Lutheran but contained no real Lutheran theology. This commenter though lauded his argument and use of scripture (while I knew this was an essay incorrectly interpreting the Bible from the Lutheran lens as the author purported). This commenter erroneously added credibility to this unsound essay. )For context it was on how the Bible apparently supports alt right ideals).
I got a comment on my post by NF, too. It's been making rounds, commenting on essays of other writers I read. Could it just be a person behind the account, one who can't really string a sentence without filtering it through AI? Sadly, those exist, sigh.
I think it is a person behind the account, but they are using spam/AI comments on hundreds of accounts per day and tricking some percent of the users into believing that they’re a real person
It's a crisis. Optimistically, we combine face-to-face contacts, old-fashioned social networks, and technological systems for identity verification (plus maybe trust in legacy institutions) to separate the human wheat from the AI chaff. But people need to care about the difference.
Completely agree. The tension between surveillance and proving that you are a real person is going to be a difficult needle to thread.
One problem being that the norms of civil conversation go rather strongly against saying that you think a post is coming from a bot. Plus the trolls are already making bot accusations.
You can always tell because the first sentence evaluates some point you made. I've been getting these too. It's weird and infuriating
I think Substack has been trying to reign these accounts in, though they’ve not told me anything or responded to any of my reports. Have you had any problems with this in the last couple weeks?
yeah i posted a note about how one account in particular spams comments under my essays, but haven't seen anything about regulation
Yeah that looks like the same script that was run on mine. Wish Substack would say something about it but I think they might be banning at least the smaller accounts doing it if they get reported
Substack already promotes locals to our feeds. I noticed because I spend too much time here. Blush. I meant to quit for Lent.
I block frequently too just to keep my feed on topic. (What I’m currently into reading- it’s not personal.)
It’s not something I’ve noticed, but I hope you’re right! I think it would be nice if they were more transparent about their algorithm, but of course they may not really know what’s inside the black box either.
I really noticed when I canceled 200 subscriptions with one click and had to start fresh. Suddenly my feed was filled with the wonder of New England.