So Justin, what happens when people realize they don’t need you for writing anything at all? Are you and your substack becoming obsolete? I want real human thought, not perfection. And AI’s do make mistakes, it is not good to rely on this exclusively. As an assist to productivity, I can see that. But you are letting it replace you entirely.
Fair question, Claire. The article's P.S. addresses this directly — I described what I wanted, picked the case studies, reviewed the draft while eating a burrito, and sent revision notes by voice memo. The thinking was mine. The typing wasn't. Is a carpenter 'replaced' by a nail gun? I'd argue I'm more present in my work now because I'm not drowning in the mechanical parts. The judgment, the voice, the 'what's worth saying' — that's still me. AI handled the transcription so I could focus on the thought.
I was getting into this and then realized... did he even write this? Did AI? Did the human being with (true) thoughts and emotions and background and history and HUMANITY write this?
"AI didn’t replace my job."
It did mine. Why do they need me to do transcription? Why do they need me to sit there for hours, listening to people speaking and doing the very difficult work of not just typing out every word that is said, but every, "um/uh", "like", "you know", etc.?
Except that's not the hardest part. As long as the audio is clear, that's the "easy" part. The hard part is taking normal speech and turning it into legible sentences that others can read and understand, without changing the meaning of what the people are saying, keeping the personality, as it were, of each person speaking, and somehow even managing to keep the tone.
That's the true work. It's as much art as it is work. A machine, a program can't do that.
But nobody cares about that, just like they don't care about the fake music, fake articles/books, fake movies, fake everything else this technology is churning out.
***
Again, it has been such a shock to see you promoting this stuff. I would have assumed you (of all people!) would have been on the front lines warning us and fighting against it.
Mo, I hear you, and I'm not going to pretend this transition isn't painful for people whose work is being automated. Transcription is genuinely hard, skilled work — you're right that it's art as much as labor. I don't have an easy answer for the displacement question. What I can say is that fighting the technology won't bring those jobs back, but learning to work with it might open doors that didn't exist before. The person who can direct AI to do transcription well, then edit for the human touches you describe, is more valuable than either alone. I know that's cold comfort when the industry is shifting under your feet. I'm sorry it hurts.
I get the allergy. Most AI output is garbage — generic, soulless, obviously machine-made. The difference I'm describing is AI that knows my voice, my clients, my constraints, and operates under my direction. It's not 'generate a blog post.' It's 'here's the data, here's what matters, here's my take — now help me ship it.' The human is still the filter. Bad AI content happens when people skip that step.
I am wary of AI in the sense that it will be the cause of much anguish in unskilled hands. I also know that with my caution or even paranoia about it I could use it safely at the cost of loosing my own agency.
Right now with some effort and dedication it is possible to control your pet AI and limit the harm it can do but the time is approaching when we will no longer KNOW if our personal AI is making choices in our name, in our interest at first we hope but perhaps for expediency in spite of our real interests at some point.
If we give our credentials to a software agent and it causes harm to our reputation or to someone else's life without us being aware how do we fix it?
I have a business idea I believe a competent current generation coding AI could create for me but am too hesitant (scared) to give it the power over me even if it might provide a period of low effort income before some other AI simply copies with less effort than it takes me to describe the project to my AI.
Essentially those who own the AI will be the owners of the modern capital and this is going to be those that possess the capital now. Asking AI to level the playing field is perhaps the only noble task it could perform for humanity, not by dragging anyone down I hope but by finding the tools that enable capital concentration and providing them to everyone and not simply a new tier of nouveau AI capitalists that will trample the rest of humanity into obscurity.
Kalle, you've identified the real tension. Giving an AI your credentials is a risk. I mitigate it by: (1) reading everything before it posts, (2) keeping it in 'ask first' mode for anything public, (3) maintaining kill switches. The capital concentration concern is valid — but I'd argue the tools are more accessible now than ever. The setup I describe costs about $200/month, not millions. Whether that remains true is the policy question of our time.
I suppose if you can earn back more than US$200 then the money is a justifiable cost. It just seems to tragic that all modern financial survival tools are going to be sold as a service, even if they are hardware they will have some cloud based service if at all possible.
I am not important enough for me to be a grave risk from giving AI my life history but hanging onto the principle makes it easier for me to notice when I am being fished by yet another system. The global tech industry has discovered that if they reduce the friction for you to install an application into a phone people stop resisting most of the time. Lots of memory (more with each version) and flashier graphics and people forget that they have loaded a hundred apps onto their phone with most of them idle yet some of those still on subscription. I have an old android phone with 4 or 5 apps that I have installed with F-Droid. Yet simply moving around in Helsinki almost forces you to have multiple apps for parking garages or public transit that rubs me up the wrong way. I do not like knowing that it is all a trap and it is all so easy.
The same way that Amazon routinely sells their own branded products that they copy from successful smaller sellers we will find that the AI assistance is fleeting. In the end humans will only be needed for the creative idea and the AI will make it real, the people with capital and ferocity will spend US$2000/month to harvest all the working ideas and ask their AI to duplicate the idea.
There will literally be nothing new under the sun in terms of marketing as it can and will be duplicated at staggering speed whenever it is detected to be profitable.
So year, AI can make money but it will be ever more cutthroat and a world I do not want to inhabit that forces everyone to be equally ruthless to break even with large capital. I see no joy in that.
My hope is that some person smarter than me will find a way to get AI to work to level the playing field but this is very much wishful thinking due to who owns most of the AI servers.
Running your own server and controlling it tightly mitigates some of the danger of slop and hallucinations and errors but it will not be able to compete with the big boys and end in tears I expect.
a Computer is Artificial Intelligence period end of story!! Not sure why everyone seems to feel compelled to allow AI (an elaborate search engine) Apps take over their entire life. This bullshit technology only takes over your life as much as you allow it too.
It only takes over what you let it. The point of the article is that I chose what to hand off (mechanical execution) and what to keep (judgment, direction, relationships). The 30% that's still mine is the part that matters. If someone lets it take over entirely, that's a choice too — just not one I'd recommend.
Fantastic article. I’m new to this topic and am looking forward to exploring the tools you have shared. Though I’ll admit, I immediately wondered if you’re responding to comments or your AI.
I may comment more fully later, but right off the bat, I'm psyched for this type of piece. I too have "discovered" a massive tool/helper in ChatGPT. Digging through tons of data. Finds tons of citations or references. Evaluating them, critically, with nuance and insight. Summarizing the whole thing. I have been blown away. I've also been somewhat blown away by the amount of push-back from well-meaning folks who STILL think ChatGPT [or place your favorite AI here] is just Wikipedia + Google. Do I need to check output? Sure. Do I need to verify citations? Sure. (Same as it ever was.) The issue is not the oft cited "replace human creativity" gambit. The issue isn't even the fact that AI *could* do some jobs better. (Hell, I might argue that many, if not most, of those jobs probably sucked anyway.) Simply but tragically overstated, a better chisel doesn't put Michealangelo out of job. An excellent tool, even one with lots of bells and whistles, remains a tool. Use it or wait until we reach Schopenhauer's endpoint and maybe wish you had. #Shrugs
The Michelangelo line is perfect. And you nailed the 'same as it ever was' point — I verified sources before AI, I verify them now. The difference is I'm verifying 10x more output in the same time. The folks who think it's 'just Wikipedia + Google' haven't seen it operate a browser, query a database, or coordinate four sub-agents overnight. That's the shift. Looking forward to your fuller thoughts.
After reading the piece, I have no fuller thoughts. It's awesome. I am currently enjoying the hell out of just the analysis part of my ChatGPT experience, which includes extensive database query and summarization. I haven't had it *do* anything yet, but that's coming!
"I too have "discovered" a massive tool/helper in ChatGPT. Digging through tons of data. Finds tons of citations or references. Evaluating them, critically, with nuance and insight. Summarizing the whole thing. I have been blown away."
The problem is that most people are NOT going to "evaluate" the information, "critically, with nuance and insight."
Do you not understand that students are already using this stuff to cheat in school? They're not going to bother even learning how to write proper sentences and think for themselves. And you are okay with that.
"I've also been somewhat blown away by the amount of push-back from well-meaning folks who STILL think ChatGPT [or place your favorite AI here] is just Wikipedia + Google. Do I need to check output? Sure. Do I need to verify citations? Sure. (Same as it ever was.)"
Again, most people are not going to be doing that.
"The issue is not the oft cited "replace human creativity" gambit."
It already is replacing it. Have you heard the AI "music" being churned out? Have you seen the "musicians" that don't actually exist but are put out there as though they do? I can't remember her name, but one had an entire website. It looked real! Right now, you can sort of tell, with a little effort. In a short time, it will be impossible to tell. And you are okay with that.
That's just one area. How many articles and even books are already being "written" by AI?
And you are okay with that.
The same will be true with visual art, TV/movies, anything we see on Youtube, all internet content - EVERYTHING. It's already hard enough to tell what is real and what isn't. Pretty soon, we won't be able to tell at all. And you are okay with that.
"The issue isn't even the fact that AI *could* do some jobs better. (Hell, I might argue that many, if not most, of those jobs probably sucked anyway.) "
Says the person who, I am sure, has a job and therefore does not care if entire industries are erased. I promise you, you will be singing a different tune when it's YOUR job and YOUR industry that is erased.
But you know what? As much as I understand and (actually feel) the complete disregard people like you have for people like me, I still can't find it in me to wish that on you. I just can't. It's horrible.
I just wish people would wake up before it's too late.
Thanks for chiming in. I agree (or suspect) that many people do not do the stuff that those of us who HOPE we’re using the tool correctly, do. Other than that, on your other points, I guess we can just agree to disagree. Be well!
So Justin, what happens when people realize they don’t need you for writing anything at all? Are you and your substack becoming obsolete? I want real human thought, not perfection. And AI’s do make mistakes, it is not good to rely on this exclusively. As an assist to productivity, I can see that. But you are letting it replace you entirely.
Fair question, Claire. The article's P.S. addresses this directly — I described what I wanted, picked the case studies, reviewed the draft while eating a burrito, and sent revision notes by voice memo. The thinking was mine. The typing wasn't. Is a carpenter 'replaced' by a nail gun? I'd argue I'm more present in my work now because I'm not drowning in the mechanical parts. The judgment, the voice, the 'what's worth saying' — that's still me. AI handled the transcription so I could focus on the thought.
I was getting into this and then realized... did he even write this? Did AI? Did the human being with (true) thoughts and emotions and background and history and HUMANITY write this?
"AI didn’t replace my job."
It did mine. Why do they need me to do transcription? Why do they need me to sit there for hours, listening to people speaking and doing the very difficult work of not just typing out every word that is said, but every, "um/uh", "like", "you know", etc.?
Except that's not the hardest part. As long as the audio is clear, that's the "easy" part. The hard part is taking normal speech and turning it into legible sentences that others can read and understand, without changing the meaning of what the people are saying, keeping the personality, as it were, of each person speaking, and somehow even managing to keep the tone.
That's the true work. It's as much art as it is work. A machine, a program can't do that.
But nobody cares about that, just like they don't care about the fake music, fake articles/books, fake movies, fake everything else this technology is churning out.
***
Again, it has been such a shock to see you promoting this stuff. I would have assumed you (of all people!) would have been on the front lines warning us and fighting against it.
I won't lie, this one hurts.
Mo, I hear you, and I'm not going to pretend this transition isn't painful for people whose work is being automated. Transcription is genuinely hard, skilled work — you're right that it's art as much as labor. I don't have an easy answer for the displacement question. What I can say is that fighting the technology won't bring those jobs back, but learning to work with it might open doors that didn't exist before. The person who can direct AI to do transcription well, then edit for the human touches you describe, is more valuable than either alone. I know that's cold comfort when the industry is shifting under your feet. I'm sorry it hurts.
People are becoming sensitive and allergic to AI-produced garbage. To me, it’s nails on a chalkboard.
I get the allergy. Most AI output is garbage — generic, soulless, obviously machine-made. The difference I'm describing is AI that knows my voice, my clients, my constraints, and operates under my direction. It's not 'generate a blog post.' It's 'here's the data, here's what matters, here's my take — now help me ship it.' The human is still the filter. Bad AI content happens when people skip that step.
I am wary of AI in the sense that it will be the cause of much anguish in unskilled hands. I also know that with my caution or even paranoia about it I could use it safely at the cost of loosing my own agency.
Right now with some effort and dedication it is possible to control your pet AI and limit the harm it can do but the time is approaching when we will no longer KNOW if our personal AI is making choices in our name, in our interest at first we hope but perhaps for expediency in spite of our real interests at some point.
If we give our credentials to a software agent and it causes harm to our reputation or to someone else's life without us being aware how do we fix it?
I have a business idea I believe a competent current generation coding AI could create for me but am too hesitant (scared) to give it the power over me even if it might provide a period of low effort income before some other AI simply copies with less effort than it takes me to describe the project to my AI.
Essentially those who own the AI will be the owners of the modern capital and this is going to be those that possess the capital now. Asking AI to level the playing field is perhaps the only noble task it could perform for humanity, not by dragging anyone down I hope but by finding the tools that enable capital concentration and providing them to everyone and not simply a new tier of nouveau AI capitalists that will trample the rest of humanity into obscurity.
Kalle, you've identified the real tension. Giving an AI your credentials is a risk. I mitigate it by: (1) reading everything before it posts, (2) keeping it in 'ask first' mode for anything public, (3) maintaining kill switches. The capital concentration concern is valid — but I'd argue the tools are more accessible now than ever. The setup I describe costs about $200/month, not millions. Whether that remains true is the policy question of our time.
I suppose if you can earn back more than US$200 then the money is a justifiable cost. It just seems to tragic that all modern financial survival tools are going to be sold as a service, even if they are hardware they will have some cloud based service if at all possible.
I am not important enough for me to be a grave risk from giving AI my life history but hanging onto the principle makes it easier for me to notice when I am being fished by yet another system. The global tech industry has discovered that if they reduce the friction for you to install an application into a phone people stop resisting most of the time. Lots of memory (more with each version) and flashier graphics and people forget that they have loaded a hundred apps onto their phone with most of them idle yet some of those still on subscription. I have an old android phone with 4 or 5 apps that I have installed with F-Droid. Yet simply moving around in Helsinki almost forces you to have multiple apps for parking garages or public transit that rubs me up the wrong way. I do not like knowing that it is all a trap and it is all so easy.
The same way that Amazon routinely sells their own branded products that they copy from successful smaller sellers we will find that the AI assistance is fleeting. In the end humans will only be needed for the creative idea and the AI will make it real, the people with capital and ferocity will spend US$2000/month to harvest all the working ideas and ask their AI to duplicate the idea.
There will literally be nothing new under the sun in terms of marketing as it can and will be duplicated at staggering speed whenever it is detected to be profitable.
So year, AI can make money but it will be ever more cutthroat and a world I do not want to inhabit that forces everyone to be equally ruthless to break even with large capital. I see no joy in that.
My hope is that some person smarter than me will find a way to get AI to work to level the playing field but this is very much wishful thinking due to who owns most of the AI servers.
Running your own server and controlling it tightly mitigates some of the danger of slop and hallucinations and errors but it will not be able to compete with the big boys and end in tears I expect.
three cheers for open source, and the open source mentality which promotes sharing useful things!
🙌 Exactly why I published the content pipeline as an open-source skill. If it's useful, share it.
a Computer is Artificial Intelligence period end of story!! Not sure why everyone seems to feel compelled to allow AI (an elaborate search engine) Apps take over their entire life. This bullshit technology only takes over your life as much as you allow it too.
It only takes over what you let it. The point of the article is that I chose what to hand off (mechanical execution) and what to keep (judgment, direction, relationships). The 30% that's still mine is the part that matters. If someone lets it take over entirely, that's a choice too — just not one I'd recommend.
Fantastic article. I’m new to this topic and am looking forward to exploring the tools you have shared. Though I’ll admit, I immediately wondered if you’re responding to comments or your AI.
I may comment more fully later, but right off the bat, I'm psyched for this type of piece. I too have "discovered" a massive tool/helper in ChatGPT. Digging through tons of data. Finds tons of citations or references. Evaluating them, critically, with nuance and insight. Summarizing the whole thing. I have been blown away. I've also been somewhat blown away by the amount of push-back from well-meaning folks who STILL think ChatGPT [or place your favorite AI here] is just Wikipedia + Google. Do I need to check output? Sure. Do I need to verify citations? Sure. (Same as it ever was.) The issue is not the oft cited "replace human creativity" gambit. The issue isn't even the fact that AI *could* do some jobs better. (Hell, I might argue that many, if not most, of those jobs probably sucked anyway.) Simply but tragically overstated, a better chisel doesn't put Michealangelo out of job. An excellent tool, even one with lots of bells and whistles, remains a tool. Use it or wait until we reach Schopenhauer's endpoint and maybe wish you had. #Shrugs
The Michelangelo line is perfect. And you nailed the 'same as it ever was' point — I verified sources before AI, I verify them now. The difference is I'm verifying 10x more output in the same time. The folks who think it's 'just Wikipedia + Google' haven't seen it operate a browser, query a database, or coordinate four sub-agents overnight. That's the shift. Looking forward to your fuller thoughts.
After reading the piece, I have no fuller thoughts. It's awesome. I am currently enjoying the hell out of just the analysis part of my ChatGPT experience, which includes extensive database query and summarization. I haven't had it *do* anything yet, but that's coming!
"I too have "discovered" a massive tool/helper in ChatGPT. Digging through tons of data. Finds tons of citations or references. Evaluating them, critically, with nuance and insight. Summarizing the whole thing. I have been blown away."
The problem is that most people are NOT going to "evaluate" the information, "critically, with nuance and insight."
Do you not understand that students are already using this stuff to cheat in school? They're not going to bother even learning how to write proper sentences and think for themselves. And you are okay with that.
"I've also been somewhat blown away by the amount of push-back from well-meaning folks who STILL think ChatGPT [or place your favorite AI here] is just Wikipedia + Google. Do I need to check output? Sure. Do I need to verify citations? Sure. (Same as it ever was.)"
Again, most people are not going to be doing that.
"The issue is not the oft cited "replace human creativity" gambit."
It already is replacing it. Have you heard the AI "music" being churned out? Have you seen the "musicians" that don't actually exist but are put out there as though they do? I can't remember her name, but one had an entire website. It looked real! Right now, you can sort of tell, with a little effort. In a short time, it will be impossible to tell. And you are okay with that.
That's just one area. How many articles and even books are already being "written" by AI?
And you are okay with that.
The same will be true with visual art, TV/movies, anything we see on Youtube, all internet content - EVERYTHING. It's already hard enough to tell what is real and what isn't. Pretty soon, we won't be able to tell at all. And you are okay with that.
"The issue isn't even the fact that AI *could* do some jobs better. (Hell, I might argue that many, if not most, of those jobs probably sucked anyway.) "
Says the person who, I am sure, has a job and therefore does not care if entire industries are erased. I promise you, you will be singing a different tune when it's YOUR job and YOUR industry that is erased.
But you know what? As much as I understand and (actually feel) the complete disregard people like you have for people like me, I still can't find it in me to wish that on you. I just can't. It's horrible.
I just wish people would wake up before it's too late.
Thanks for chiming in. I agree (or suspect) that many people do not do the stuff that those of us who HOPE we’re using the tool correctly, do. Other than that, on your other points, I guess we can just agree to disagree. Be well!