Yes, perhaps I am overreacting. This article came out of frustration with a completely different situation about someone questioning my actual face as AI in a video, so I jumped online to write this article and it somehow turned into a rant about Substack’s AI tool.
This is what happens when I don’t have a plan going into a writing session.
I still think there’s a valid point to be made about the real danger to society when everything is synthetic, where human creators fit into that new world, and the hard work it takes to create the stuff I do. Saying something is AI is an instant way to disparage the insane amount of work that goes into writing and video work.
By the way, if you’re not on Medium, I’ve been writing there since 2018 and some of my best writing is over there. I don’t normally post the same thing on Medium and Substack, which means if you only know me from Substack, you’re missing out on 50% of the stuff I write. I know some writers copy and paste whatever they post here to Medium as well. I’ve tried to avoid that to give people a reason to read at both places. Unlike Substack, Medium is only $5/month for ALL their writers. So its value is unmatched.
First off, if you migrate to Medium I'll follow you without hesitation. I greatly appreciate your accurate, clear writing.
I have a BS in Tech Writing and did it for over 13 years for DoD and VA after a few years writing manuals in the airlines. I'm glad I'm retired now, because having to go through all this bullshit AI grading would drive me to drink! You aren't the only person dealing with the AI estimates. Our youngest daughter is a Ph.D. teaching at the university level. Her institution has multiple AI detectors and she still struggles with her student's papers. If misery loves company, you have an auditoriumfull of fellow sufferers.
You already cross post. And Medium has a serious AI problem, almost certainly worse than Substack. Some day, they may try and address it the way Twitter and FB pretend to address bots.
I admire your work, re-stack pretty regularly, and think you’re over-reacting. You use AI to do research and help organize and structure your article. So what ? Here the 85/15 ratio is critical. You’re using AI. You’re doing good work. Keep going.
That's because you work hard at it and leave no stone unturned. And that can only mean you're a perfectionist, a professional, and an all-round great guy. ;)
I’ll go further and say that if a writer doing the kind of technical reporting you do, Wes, DOESN’T use AI, THAT makes me like them less! AI is a great tool and it absolutely elevates the final result for those who know how to use it properly. A true professional should always be challenging themselves to use the best tools available to make the best version of whatever it is they are making. Don’t be ashamed of being the best you can be!
Wes, a good article and I fully understand your frustration but I would gently suggest this is possibly an over reaction. I use AI to help research and structure papers but the words are my own and I do reference AI at the end of the paper. Your Monday posts also clearly state how you use OSINT systems so I have no issues. As you have described the way you produce your "unique" content in this article the 15/85 split was NOT something that raised any concerns with me but I do accept that it is rather telling about the potential error rates of the system that Substack is using.
Perhaps the best approach for Substack would be to use percentile gradings. You may want to suggest this in your feedback and as you say it provides a non-specific figure. Either way as someone who reads many formal papers, which are human authored, I am very happy with what you produce and would ask that you continue your very detailed analysis of current military technical subjects.
Wes, don’t leave! Please! Your posts are an education and a highlight of my week. Take the third option and ignore this BS. By the way, here’s a wrap on your writing: it’s not only very clear, it is occasionally elegant. Phillips O’Brian, whose content I like, is extremely irritating because he can’t spell (at all) and he declines to use Grammerly and an editor, and having to wade through the sludge is off-putting. And I don’t know Medium so can’t go there if you abandon Substack. Did you pass the Bar? Best wishes, Will
I think you could add one or two sentences on how you use AI in your research but personally read all sources and do your own writting to your personal note at the end of your articles and leave it at that.
“Organization isn’t a “stylistic flourish” in this line of work. It’s the difference between a reader understanding how a jammer defeats satellite navigation and a reader quietly closing the tab and opening up Pornhub.” This reads EXACTLY like AI except for the last word. AI always says “that isn’t A, it’s all kinds of modifiers B with a cherry on top.” It’s my main tell.
Here’s a vote for not stopping these written pieces—whether on Substack or somewhere else. I’m not a video guy (or a podcast guy)—I’m a reader. Also, FWIW, the fragments I’ve seen of your videos leave me uninterested (that man should cut back on the caffeine—it’s a style, and I assume it works for a lot of your viewers, but I dislike it). But your writing—your writing is awesome. It’s interesting to read (in this essay) how you structure your pieces, and they are generally very clear, and I’m a big fan of your wit. Also, excellent content. And it’s obvious that you know a lot and put a lot of work into your Substack writing, Pangram be damned. I hope you continue doing this.
It seems to me that a better approach Substack and others should take is to use their measure to predict whether a piece of writing is creative content or machine made. 15% would suggest a low probability that the piece is not creative content.
I can't imagine how they can parse a piece and conclude that a specific percentage of the writing was AI. Weird. BTW, anyone who thinks your video are AI is an idiot.
Wes, I feel your pain! In what sane world do we ask AI to check work to see if AI is involved in writing it? I had a funny and disturbing discussion with google AI about my Dad. The question was who were the pioneers of the mushroom industry in Australia. It came up with a response that omitted my dad, so I prompted it and it came back with an apology and an excuse (it kept saying “I” as well). It then invented a brother that I never knew I had, but that was okay because my siblings and I became grandchildren of my father??? so he would have been my uncle. Then when this was pointed out it apologised again and came up with another excuse (still saying “I”). So how can anyone trust this technology?
When I read your work I can hear your voice, if that stops then that will be my AI detector. That’s assuming it is you on your videos and not AI generated 🤯🤯🤯
Wes, I’m a professional economist and data scientist with 45 years of experience. I have written hundreds of technical and scientific papers. I’m pretty sure an AI engine would look at my body of work and draw a similar conclusion. I have never used AI in the production of my work.
A recent article in either the Times, the Post noted that AI was making papers by students all sound similar and less distinctive. A march, if you will, toward stylistic conformity and intellectual void.
If you leave Substack, you won’t be the first. I follow a baseball writer, Craig Calcaterra, who left for a different platform. The platform he chose works smoothly from a subscriber perspective. I know I’ll follow you as I value your work. I also read you on Medium.
The AI intellectual bubble, like the associated economic one, will collapse, and a different path for use will emerge. Until then, we must stand up for humanity and human principles in the face of thinking machine evangelism.
And Wes, if you do read any of my replies of your substacked words, I really have written some sci-fi. Nothing professional, but if you want a laugh, then I'd be happy to pass it on to you. That way, maybe I can get on with my King Halfer's Hamalot stories and one other silly Polish story. ;)
Maybe I shouldn't say this, but I had to use Google AI and Gemini for some of my mad scientist ideas just to make some of my writing a bit more plausible. LOL
15% AI is as good as human from where I sit… no worries. I never even thought for a second you could be AI; you write to make a point, not to sound like some smarta** without a clue
For someone who already has a large corpus of their own pre-AI writing voice, I would suggest feeding that into a model (either ask it to produce an editing prompt or get someone to train/fine tune a model for you) and then do the first draft yourself, and have the LLM edit it but with explicit instructions not to modify the voice.
On holiday so late to this one. I also hope you don't move solely to Medium. I think you have to hold on for just a bit longer until the AI circular-financing ponzi scheme eventually implodes and the stupidly-costly-to-run, LLM-based AI 'helpers' as we currently know them die off.
Another thing, I haven't even noticed this Substack AI-percentage score so I wouldn't worry about it. (he says)
Yes, perhaps I am overreacting. This article came out of frustration with a completely different situation about someone questioning my actual face as AI in a video, so I jumped online to write this article and it somehow turned into a rant about Substack’s AI tool.
This is what happens when I don’t have a plan going into a writing session.
I still think there’s a valid point to be made about the real danger to society when everything is synthetic, where human creators fit into that new world, and the hard work it takes to create the stuff I do. Saying something is AI is an instant way to disparage the insane amount of work that goes into writing and video work.
By the way, if you’re not on Medium, I’ve been writing there since 2018 and some of my best writing is over there. I don’t normally post the same thing on Medium and Substack, which means if you only know me from Substack, you’re missing out on 50% of the stuff I write. I know some writers copy and paste whatever they post here to Medium as well. I’ve tried to avoid that to give people a reason to read at both places. Unlike Substack, Medium is only $5/month for ALL their writers. So its value is unmatched.
First off, if you migrate to Medium I'll follow you without hesitation. I greatly appreciate your accurate, clear writing.
I have a BS in Tech Writing and did it for over 13 years for DoD and VA after a few years writing manuals in the airlines. I'm glad I'm retired now, because having to go through all this bullshit AI grading would drive me to drink! You aren't the only person dealing with the AI estimates. Our youngest daughter is a Ph.D. teaching at the university level. Her institution has multiple AI detectors and she still struggles with her student's papers. If misery loves company, you have an auditoriumfull of fellow sufferers.
Thanks Bender
You already cross post. And Medium has a serious AI problem, almost certainly worse than Substack. Some day, they may try and address it the way Twitter and FB pretend to address bots.
I admire your work, re-stack pretty regularly, and think you’re over-reacting. You use AI to do research and help organize and structure your article. So what ? Here the 85/15 ratio is critical. You’re using AI. You’re doing good work. Keep going.
I do get a little hot headed sometimes and type out some rash words. It's just frustrating. It's the Irish temper
That's because you work hard at it and leave no stone unturned. And that can only mean you're a perfectionist, a professional, and an all-round great guy. ;)
I’ll go further and say that if a writer doing the kind of technical reporting you do, Wes, DOESN’T use AI, THAT makes me like them less! AI is a great tool and it absolutely elevates the final result for those who know how to use it properly. A true professional should always be challenging themselves to use the best tools available to make the best version of whatever it is they are making. Don’t be ashamed of being the best you can be!
Wes, a good article and I fully understand your frustration but I would gently suggest this is possibly an over reaction. I use AI to help research and structure papers but the words are my own and I do reference AI at the end of the paper. Your Monday posts also clearly state how you use OSINT systems so I have no issues. As you have described the way you produce your "unique" content in this article the 15/85 split was NOT something that raised any concerns with me but I do accept that it is rather telling about the potential error rates of the system that Substack is using.
Perhaps the best approach for Substack would be to use percentile gradings. You may want to suggest this in your feedback and as you say it provides a non-specific figure. Either way as someone who reads many formal papers, which are human authored, I am very happy with what you produce and would ask that you continue your very detailed analysis of current military technical subjects.
Wes, don’t leave! Please! Your posts are an education and a highlight of my week. Take the third option and ignore this BS. By the way, here’s a wrap on your writing: it’s not only very clear, it is occasionally elegant. Phillips O’Brian, whose content I like, is extremely irritating because he can’t spell (at all) and he declines to use Grammerly and an editor, and having to wade through the sludge is off-putting. And I don’t know Medium so can’t go there if you abandon Substack. Did you pass the Bar? Best wishes, Will
I think you could add one or two sentences on how you use AI in your research but personally read all sources and do your own writting to your personal note at the end of your articles and leave it at that.
“Organization isn’t a “stylistic flourish” in this line of work. It’s the difference between a reader understanding how a jammer defeats satellite navigation and a reader quietly closing the tab and opening up Pornhub.” This reads EXACTLY like AI except for the last word. AI always says “that isn’t A, it’s all kinds of modifiers B with a cherry on top.” It’s my main tell.
Here’s a vote for not stopping these written pieces—whether on Substack or somewhere else. I’m not a video guy (or a podcast guy)—I’m a reader. Also, FWIW, the fragments I’ve seen of your videos leave me uninterested (that man should cut back on the caffeine—it’s a style, and I assume it works for a lot of your viewers, but I dislike it). But your writing—your writing is awesome. It’s interesting to read (in this essay) how you structure your pieces, and they are generally very clear, and I’m a big fan of your wit. Also, excellent content. And it’s obvious that you know a lot and put a lot of work into your Substack writing, Pangram be damned. I hope you continue doing this.
I want is accurate information about the things you write about and I don't care how you get it. If AI helps, I think you should use it.
It seems to me that a better approach Substack and others should take is to use their measure to predict whether a piece of writing is creative content or machine made. 15% would suggest a low probability that the piece is not creative content.
I can't imagine how they can parse a piece and conclude that a specific percentage of the writing was AI. Weird. BTW, anyone who thinks your video are AI is an idiot.
Wes, I feel your pain! In what sane world do we ask AI to check work to see if AI is involved in writing it? I had a funny and disturbing discussion with google AI about my Dad. The question was who were the pioneers of the mushroom industry in Australia. It came up with a response that omitted my dad, so I prompted it and it came back with an apology and an excuse (it kept saying “I” as well). It then invented a brother that I never knew I had, but that was okay because my siblings and I became grandchildren of my father??? so he would have been my uncle. Then when this was pointed out it apologised again and came up with another excuse (still saying “I”). So how can anyone trust this technology?
When I read your work I can hear your voice, if that stops then that will be my AI detector. That’s assuming it is you on your videos and not AI generated 🤯🤯🤯
Wes, I’m a professional economist and data scientist with 45 years of experience. I have written hundreds of technical and scientific papers. I’m pretty sure an AI engine would look at my body of work and draw a similar conclusion. I have never used AI in the production of my work.
A recent article in either the Times, the Post noted that AI was making papers by students all sound similar and less distinctive. A march, if you will, toward stylistic conformity and intellectual void.
If you leave Substack, you won’t be the first. I follow a baseball writer, Craig Calcaterra, who left for a different platform. The platform he chose works smoothly from a subscriber perspective. I know I’ll follow you as I value your work. I also read you on Medium.
The AI intellectual bubble, like the associated economic one, will collapse, and a different path for use will emerge. Until then, we must stand up for humanity and human principles in the face of thinking machine evangelism.
What everyone said. LOL
And Wes, if you do read any of my replies of your substacked words, I really have written some sci-fi. Nothing professional, but if you want a laugh, then I'd be happy to pass it on to you. That way, maybe I can get on with my King Halfer's Hamalot stories and one other silly Polish story. ;)
Maybe I shouldn't say this, but I had to use Google AI and Gemini for some of my mad scientist ideas just to make some of my writing a bit more plausible. LOL
15% AI is as good as human from where I sit… no worries. I never even thought for a second you could be AI; you write to make a point, not to sound like some smarta** without a clue
AI writes everything like it's a marketing blurb.
For someone who already has a large corpus of their own pre-AI writing voice, I would suggest feeding that into a model (either ask it to produce an editing prompt or get someone to train/fine tune a model for you) and then do the first draft yourself, and have the LLM edit it but with explicit instructions not to modify the voice.
On holiday so late to this one. I also hope you don't move solely to Medium. I think you have to hold on for just a bit longer until the AI circular-financing ponzi scheme eventually implodes and the stupidly-costly-to-run, LLM-based AI 'helpers' as we currently know them die off.
Another thing, I haven't even noticed this Substack AI-percentage score so I wouldn't worry about it. (he says)