For me, a gentleman in his later work life, I can recall a time when my work product was written longhand, faxed to an assistant, then typed and returned to me for final edits, which I then returned for a final version for distribution. That seemed needlessly laborious even in the early 90s, and I suggested that we work on laptops with email. One was purchased for me, and I learned DOS because no one else knew what to do. I started to produce at a much greater pace, and I was encouraged to train co-workers so they could do likewise. Initially, there was some bump in pay, though not commensurate with our production. Eventually, our pay stagnated, though our production was not expected to diminish. This is a not uncommon experience.
Now come the LLMs, which do in fact make some elements of my job easier. I’m no longer in the avant-garde of these innovations, though my production is greater. My younger colleagues, OTOH, have used these innovations to create systems of production that are much faster than we have used in the past. Production is up and that initial market efficiency has resulted in a modest pay increase for me, and a larger one for my digital native colleagues who have developed these tools. The market efficiency is gradually diminishing, but the demand for production is unrelenting. I have cautioned my co-workers that we may be planning our own obsolescence once the market adjusts to the production expectations and the profit level can no longer exploit this temporary market efficiency. But the lure of immediate pay gains trumps any desire to wrest control of the means of production. This is America, and we will gladly pay Tuesday for a hamburger today.
I hate hate to reference Malcolm Gladwell, but I can usually tell when work product has relied on GPT based on a “Blink” test. There are many tells (rule of threes, “While x may be true, one must consider y”, etc.), and sometimes users will deliberately include typos for authenticity. But the Blink test can discern whether the voice is authentic, though that doesn’t help with calculations or coding.
There is a lot to consider here, as the college-educated class may be relegated to lower paying service jobs. A new definition of intelligence will be mandated. Who cares if you have access to a fund of information and can answer Jeopardy questions when the libraries of information plundered by LLMs grant you easy access? The hope would be that we would be able to use that information for purposes other than capital accumulation, but that requires self-awareness, compassion, and probably sacrifice on behalf of others.
Many things about this are conflicting—and true. The threat isn’t AI itself. It’s the economic paradigm that deploys it. AI is a vibe amplifier, a productivity multiplier, and in the hands of capital, a tool for workforce extraction on an unprecedented scale.
The displacement has already begun. Junior software hiring is drying up. ShadowIT and ShadowAI emerged because individuals inside orgs are using these tools quietly, without permission, just to keep pace or make things easier. Management didn’t lead this. They’re just now realizing they’re behind.
By the end of the year, the workforce will be in an obvious state of collapse. Not everywhere at once, but enough to break denial. Management is only incentivized to optimize their silo, not protect people. And the middle class has long functioned as a semi-meritocratic pseudo UBI—a buffer. That buffer is being erased.
This is the shift from Software-as-a-Service to Employee-as-a-Service. Add one or two zeroes to a SaaS invoice, and a platform replaces a person. It’s already happening. And it will accelerate.
This isn’t a defense. It’s a description. Even if one company resists, another won’t. Even if one region regulates, another won’t. The tech moves freely. It will hollow out regional payrolls and reroute capital into subscription services owned by the same platforms that trained the models.
The capitalists will win. And, as the vibes have been warning for generations—that means everyone else loses.
Well-crafted argument, Scarlet! I've recently been reading Capital for the first time and that chapter has stuck in my mind for weeks. I've been gathering sources to try and understand how AI may intersect with the Arts and Entertainment sector, but indeed it will impact far more than that. I worry that the battle already may be lost. I know of no examples of "productive" technology being permanently suppressed under capitalism. That the US and China are locked into an de facto AI arms race makes me even more pessimistic. Eventually the sheen of snake oil and techno-religious sophistry will come off and capitalists will realize what they have might be the industrialization of knowledge itself. While I don't think AI can be stopped, what we can do is organize those replaced by it both to ensure their own well-being, and provide a united front against Capital as it once more leverages scientific advancement to remake the world in its image.
I've really been appreciating your essays here on Substack, and I'm always up for a reappraisal of the Luddites. The world certainly needs more smart thinkers like you to help transition us all to better social arrangements together.
It's in that spirit that I am pulling out this sentence for critique:
"While AI has some negative implications for the environment, those could be negated under socialism by the use of clean energy."
My own work here on Substack is centered on environmental issues. In the past, I wrote about a wider array of political and social issues, but I have chosen to specialize this way because the more-than-human world doesn't get enough attention in my opinion.
I recently wrote about the negative environmental implications of AI here:
The tldr is that building, maintaining and running AI infrastructure requires environmental destruction, and not just what comes from fossil fuels. Water is needed for cooling, rare earth minerals for the chips, copper for the electricity, etc. At every step, habitat is destroyed and toxic pollution is produced. These are impacts that are unavoidable, regardless of who owns and operates the infrastructure.
One of the two topics I focus on here is the environmental impacts of "clean" or "green" energy. All my posts tagged that way are collected here:
Again, these technologies all have an environmentally destructive side that is not trivial. In some cases, natural places that had so far been spared the worst of industrial development because they didn't have fossil fuel resources are now under the gun. For example, lithium mining in Nevada and Oregon is tearing up high desert landscapes that are home to innumerable plants and animals, some of them endangered. Another example: bulldozing desert habitat for solar arrays and wind farms. Again, these places are home for innumerable creatures.
I have personally been to these places and seen the destruction with my own eyes. In some cases, the before and after, which is both heartbreaking and enraging. I have and will continue to oppose these projects in whatever way I can, in defense of the earth, no matter the politics of the people pushing for them.
Rather than trying to sustain our current levels of consumption by switching out energy sources, I advocate for reducing our overall energy use:
“No” to a “Green Energy Transition”—“Yes” to an “Energy Reduction Transition”
I hope you will read some of my work here, and give consideration to the points I raise. I do think that socialism could help create a more environmentally sustainable society, if only because it would be more democratic than we have now. But atm I am troubled by what seems like a blind spot to the more-than-human world.
Well said! There is no example of any technological advancement which increased the efficiency and resulted in reduced total energy usage. Even if we were to successfully move into a post-capitalist era the urge to maximize energy consumption, which lies at the heart of life itself, will remain. The question becomes … is it possible to educate the populace regarding our inherent integration within ecosystem dynamics, to the point we will collectively chose to reduce consumption in a last ditch effort to spare ourselves an uncontrollable negative feedback loop, or will we raise the stakes further by feverishly implementing another energy-intensive round intended to delay this inevitable course correction?
Came for the nostalgic movie made newly relevant, stayed for the smooth and clear writing.
One note.
"The goodness or badness of any tool lies in how it is wielded...
"Capitalism has long past exceeded its utility, and only under a new system, a socialist system, can we truly thrive and provide for everyone."
This particular moment in human history begs for a Great Remembering: Remember that government, economic & financial systems, and technology are all — or should be — used as tools. Tools that serve us, and which we choose how to wield. Capitalism unbridled, and LLMs & AI unbridled, cannot, by their very natures, serve us. I hold out diminishing but real hope that we, as a society, will remember that we must master these man-made creations or be mastered by them.
"We must continue to rage against the machines to expose the contradictions of this system..." Always.
The other example are gig apps, presented as a disruptor technology but really they just bypassed labor agreements and regulations. AI is also presented as technology but is actually just regurgitated plagiarism and bypassing intellectual property.
Typewriters and bicycles, both powered somatically, are the peak of human technology. Everything else is alien technology, spawn of the black sludge, and should be destroyed.
Whenever a capitalist unveils some grand new "innovation", I always find myself asking: "who benefits from this innovation the most?" It's hard not to have that thought when capitalists claim that "innovation" under this system benefits us all equally, when in reality it's all for the purpose of yielding greater profits for a chosen few. The core belief that guided the Luddites was anything but unfounded and I really appreciate that you highlighted them when discussing the necessity right now to rage against the machine. I couldn't agree more when you say: "The goodness or badness of any tool lies in how it is wielded." AI doesn't have to be as revolting as it currently is - a lot of that negative perception stems from those who currently wield, develop, and deploy it with such certainty that 'this' will be what "saves" humanity.
I saw someone share a new study from the Harvard Business Review yesterday that detailed what the primary uses of generative AI are amongst everyday people and the top 3 uses for 2025 so far are:
1. Therapy/companionship
2. Organizing my life
3. Finding purpose
Something tells me that if these are the top 3 most common uses for generative AI right now, then maybe we need to revisit the original question of: "who benefits from this innovation the most?" As you expertly say as well: "every time there is an advancement in technology, that advancement will be wielded by the capitalist to strip the worker of his or her ability to command a high wage, and if possible will replace the worker altogether." Add in the fact that companies like Microsoft are selling AI technology to genocidal apartheid states like Israel, tech companies teaming up with the U.S. government to enhance greater mass surveillance, and you've got a capitalist machine using a tool like AI for all the wrong reasons: greater belligerent profit seeking and control at the expense of everyday people and the planet itself.
The vision of "Dark Enlightenment" is already here and growing by the day, so the time has never been better to pull ourselves out of this capitalist deathtrap and build a better world that doesn't collapse into a techno-feudalist "gov-corp". We shouldn't have to live in fear of the tools we develop and as long as the capitalist structure remains in tact, that'll be battle we have no choice but to fight.
Capitalism develops the productive forces, but it reaches a point where the productive forces cannot be developed further because they're being held back by the relations of production.
When this contradiction reaches a boiling point, there's either revolution, reform, or reaction.
Revolution is when the productive forces win by destroying the relations of production, beginning a proletarian relation of production, allowing for the forces of production to continue to develop.
Reform manages to buy time.
Reaction is when the forces of production are destroyed, and the relations of production win, setting back time: War, fascism, genocide, destruction.
"Raging against the machine" is by definition reactionary, it's trying to destroy the forces of production rather than the relations of production. The contradiction between abudance of food and starvation, or in this case of losing your job and not being paid even though the same product and action is being performed, are contradictions caused by the relations of production, and can only be done by destroying them.
Would any of the actions in the article be any better had they been done by humans? If ICE identified people manually rather than through bots, would it be okay? How is this any different from arguing that we should stop having electricity because electricity allows USA mercenaries to kill people easier?
What is to be done is seizing the forces of production, but their development themselves aren't the issue. This is why you see so Many vote blues who hate AI.
"The goodness or badness of any tool lies in how it is wielded" is the proverbial hitting of nail on head. This is one of the better socialist critiques of tech advancement under capitalism that l've come across recently. Only one thing I take issue with: maybe it's because I come from that later Gen-X/early Millennial cusp (as do many in my inner circle of friends), but not all of us object to being luddites! I'd say some are even loud and proud about it. :)
BTW, I'm also still a little skeptical of LLMs' overall effect on software engineering so far. Automation and even auto code generation have been staples in Cl/CD pipelines for years, and rather than displace a bunch of engineers, we saw a rise in specialization around those tools—a skill set shift rather than a purge. I'm out of the industry for 2.5 years now, but I keep in contact with many who are still in it, and there doesn't seem to be quite the wave of machines replacing humans as I'd expect, given OpenAl's reported aggressiveness in enterprise sales. I don't doubt that day is coming, I'm just not yet fully sold on generative Al as the lightning bolt it's sold as. You've got more recent first-hand experience with this so l'd be curious what you've been observing on the ground. My hunch is that the vast majority of code reviews are still done by humans, even if it was initially generated by Al or by engineered systems, but that's just a hunch based on vibes and not real world data. Thanks for another great read!
For me, a gentleman in his later work life, I can recall a time when my work product was written longhand, faxed to an assistant, then typed and returned to me for final edits, which I then returned for a final version for distribution. That seemed needlessly laborious even in the early 90s, and I suggested that we work on laptops with email. One was purchased for me, and I learned DOS because no one else knew what to do. I started to produce at a much greater pace, and I was encouraged to train co-workers so they could do likewise. Initially, there was some bump in pay, though not commensurate with our production. Eventually, our pay stagnated, though our production was not expected to diminish. This is a not uncommon experience.
Now come the LLMs, which do in fact make some elements of my job easier. I’m no longer in the avant-garde of these innovations, though my production is greater. My younger colleagues, OTOH, have used these innovations to create systems of production that are much faster than we have used in the past. Production is up and that initial market efficiency has resulted in a modest pay increase for me, and a larger one for my digital native colleagues who have developed these tools. The market efficiency is gradually diminishing, but the demand for production is unrelenting. I have cautioned my co-workers that we may be planning our own obsolescence once the market adjusts to the production expectations and the profit level can no longer exploit this temporary market efficiency. But the lure of immediate pay gains trumps any desire to wrest control of the means of production. This is America, and we will gladly pay Tuesday for a hamburger today.
I hate hate to reference Malcolm Gladwell, but I can usually tell when work product has relied on GPT based on a “Blink” test. There are many tells (rule of threes, “While x may be true, one must consider y”, etc.), and sometimes users will deliberately include typos for authenticity. But the Blink test can discern whether the voice is authentic, though that doesn’t help with calculations or coding.
There is a lot to consider here, as the college-educated class may be relegated to lower paying service jobs. A new definition of intelligence will be mandated. Who cares if you have access to a fund of information and can answer Jeopardy questions when the libraries of information plundered by LLMs grant you easy access? The hope would be that we would be able to use that information for purposes other than capital accumulation, but that requires self-awareness, compassion, and probably sacrifice on behalf of others.
Many things about this are conflicting—and true. The threat isn’t AI itself. It’s the economic paradigm that deploys it. AI is a vibe amplifier, a productivity multiplier, and in the hands of capital, a tool for workforce extraction on an unprecedented scale.
The displacement has already begun. Junior software hiring is drying up. ShadowIT and ShadowAI emerged because individuals inside orgs are using these tools quietly, without permission, just to keep pace or make things easier. Management didn’t lead this. They’re just now realizing they’re behind.
By the end of the year, the workforce will be in an obvious state of collapse. Not everywhere at once, but enough to break denial. Management is only incentivized to optimize their silo, not protect people. And the middle class has long functioned as a semi-meritocratic pseudo UBI—a buffer. That buffer is being erased.
This is the shift from Software-as-a-Service to Employee-as-a-Service. Add one or two zeroes to a SaaS invoice, and a platform replaces a person. It’s already happening. And it will accelerate.
This isn’t a defense. It’s a description. Even if one company resists, another won’t. Even if one region regulates, another won’t. The tech moves freely. It will hollow out regional payrolls and reroute capital into subscription services owned by the same platforms that trained the models.
The capitalists will win. And, as the vibes have been warning for generations—that means everyone else loses.
Well-crafted argument, Scarlet! I've recently been reading Capital for the first time and that chapter has stuck in my mind for weeks. I've been gathering sources to try and understand how AI may intersect with the Arts and Entertainment sector, but indeed it will impact far more than that. I worry that the battle already may be lost. I know of no examples of "productive" technology being permanently suppressed under capitalism. That the US and China are locked into an de facto AI arms race makes me even more pessimistic. Eventually the sheen of snake oil and techno-religious sophistry will come off and capitalists will realize what they have might be the industrialization of knowledge itself. While I don't think AI can be stopped, what we can do is organize those replaced by it both to ensure their own well-being, and provide a united front against Capital as it once more leverages scientific advancement to remake the world in its image.
I've really been appreciating your essays here on Substack, and I'm always up for a reappraisal of the Luddites. The world certainly needs more smart thinkers like you to help transition us all to better social arrangements together.
It's in that spirit that I am pulling out this sentence for critique:
"While AI has some negative implications for the environment, those could be negated under socialism by the use of clean energy."
My own work here on Substack is centered on environmental issues. In the past, I wrote about a wider array of political and social issues, but I have chosen to specialize this way because the more-than-human world doesn't get enough attention in my opinion.
I recently wrote about the negative environmental implications of AI here:
https://kollibri.substack.com/p/the-planet-cant-afford-ai
The tldr is that building, maintaining and running AI infrastructure requires environmental destruction, and not just what comes from fossil fuels. Water is needed for cooling, rare earth minerals for the chips, copper for the electricity, etc. At every step, habitat is destroyed and toxic pollution is produced. These are impacts that are unavoidable, regardless of who owns and operates the infrastructure.
One of the two topics I focus on here is the environmental impacts of "clean" or "green" energy. All my posts tagged that way are collected here:
https://kollibri.substack.com/t/green-energy-ed9
Again, these technologies all have an environmentally destructive side that is not trivial. In some cases, natural places that had so far been spared the worst of industrial development because they didn't have fossil fuel resources are now under the gun. For example, lithium mining in Nevada and Oregon is tearing up high desert landscapes that are home to innumerable plants and animals, some of them endangered. Another example: bulldozing desert habitat for solar arrays and wind farms. Again, these places are home for innumerable creatures.
I have personally been to these places and seen the destruction with my own eyes. In some cases, the before and after, which is both heartbreaking and enraging. I have and will continue to oppose these projects in whatever way I can, in defense of the earth, no matter the politics of the people pushing for them.
Rather than trying to sustain our current levels of consumption by switching out energy sources, I advocate for reducing our overall energy use:
“No” to a “Green Energy Transition”—“Yes” to an “Energy Reduction Transition”
https://kollibri.substack.com/p/no-to-a-green-energy-transitionyes
I hope you will read some of my work here, and give consideration to the points I raise. I do think that socialism could help create a more environmentally sustainable society, if only because it would be more democratic than we have now. But atm I am troubled by what seems like a blind spot to the more-than-human world.
Thank you for your consideration!
Well said! There is no example of any technological advancement which increased the efficiency and resulted in reduced total energy usage. Even if we were to successfully move into a post-capitalist era the urge to maximize energy consumption, which lies at the heart of life itself, will remain. The question becomes … is it possible to educate the populace regarding our inherent integration within ecosystem dynamics, to the point we will collectively chose to reduce consumption in a last ditch effort to spare ourselves an uncontrollable negative feedback loop, or will we raise the stakes further by feverishly implementing another energy-intensive round intended to delay this inevitable course correction?
Great write up but it's sad that I had to learn who Curtis Yarvin is.
Came for the nostalgic movie made newly relevant, stayed for the smooth and clear writing.
One note.
"The goodness or badness of any tool lies in how it is wielded...
"Capitalism has long past exceeded its utility, and only under a new system, a socialist system, can we truly thrive and provide for everyone."
This particular moment in human history begs for a Great Remembering: Remember that government, economic & financial systems, and technology are all — or should be — used as tools. Tools that serve us, and which we choose how to wield. Capitalism unbridled, and LLMs & AI unbridled, cannot, by their very natures, serve us. I hold out diminishing but real hope that we, as a society, will remember that we must master these man-made creations or be mastered by them.
"We must continue to rage against the machines to expose the contradictions of this system..." Always.
The other example are gig apps, presented as a disruptor technology but really they just bypassed labor agreements and regulations. AI is also presented as technology but is actually just regurgitated plagiarism and bypassing intellectual property.
Buy robots and send them to work in your place before they do! 😜 #theendoftheworldiscancelled
Keep Mao, but otherwise, awesome work comrade! 😂
Typewriters and bicycles, both powered somatically, are the peak of human technology. Everything else is alien technology, spawn of the black sludge, and should be destroyed.
Bicycles really are good cut-off point
Whenever a capitalist unveils some grand new "innovation", I always find myself asking: "who benefits from this innovation the most?" It's hard not to have that thought when capitalists claim that "innovation" under this system benefits us all equally, when in reality it's all for the purpose of yielding greater profits for a chosen few. The core belief that guided the Luddites was anything but unfounded and I really appreciate that you highlighted them when discussing the necessity right now to rage against the machine. I couldn't agree more when you say: "The goodness or badness of any tool lies in how it is wielded." AI doesn't have to be as revolting as it currently is - a lot of that negative perception stems from those who currently wield, develop, and deploy it with such certainty that 'this' will be what "saves" humanity.
I saw someone share a new study from the Harvard Business Review yesterday that detailed what the primary uses of generative AI are amongst everyday people and the top 3 uses for 2025 so far are:
1. Therapy/companionship
2. Organizing my life
3. Finding purpose
Something tells me that if these are the top 3 most common uses for generative AI right now, then maybe we need to revisit the original question of: "who benefits from this innovation the most?" As you expertly say as well: "every time there is an advancement in technology, that advancement will be wielded by the capitalist to strip the worker of his or her ability to command a high wage, and if possible will replace the worker altogether." Add in the fact that companies like Microsoft are selling AI technology to genocidal apartheid states like Israel, tech companies teaming up with the U.S. government to enhance greater mass surveillance, and you've got a capitalist machine using a tool like AI for all the wrong reasons: greater belligerent profit seeking and control at the expense of everyday people and the planet itself.
The vision of "Dark Enlightenment" is already here and growing by the day, so the time has never been better to pull ourselves out of this capitalist deathtrap and build a better world that doesn't collapse into a techno-feudalist "gov-corp". We shouldn't have to live in fear of the tools we develop and as long as the capitalist structure remains in tact, that'll be battle we have no choice but to fight.
Great piece once again Scarlet!
Capitalism develops the productive forces, but it reaches a point where the productive forces cannot be developed further because they're being held back by the relations of production.
When this contradiction reaches a boiling point, there's either revolution, reform, or reaction.
Revolution is when the productive forces win by destroying the relations of production, beginning a proletarian relation of production, allowing for the forces of production to continue to develop.
Reform manages to buy time.
Reaction is when the forces of production are destroyed, and the relations of production win, setting back time: War, fascism, genocide, destruction.
"Raging against the machine" is by definition reactionary, it's trying to destroy the forces of production rather than the relations of production. The contradiction between abudance of food and starvation, or in this case of losing your job and not being paid even though the same product and action is being performed, are contradictions caused by the relations of production, and can only be done by destroying them.
Would any of the actions in the article be any better had they been done by humans? If ICE identified people manually rather than through bots, would it be okay? How is this any different from arguing that we should stop having electricity because electricity allows USA mercenaries to kill people easier?
What is to be done is seizing the forces of production, but their development themselves aren't the issue. This is why you see so Many vote blues who hate AI.
"The goodness or badness of any tool lies in how it is wielded" is the proverbial hitting of nail on head. This is one of the better socialist critiques of tech advancement under capitalism that l've come across recently. Only one thing I take issue with: maybe it's because I come from that later Gen-X/early Millennial cusp (as do many in my inner circle of friends), but not all of us object to being luddites! I'd say some are even loud and proud about it. :)
BTW, I'm also still a little skeptical of LLMs' overall effect on software engineering so far. Automation and even auto code generation have been staples in Cl/CD pipelines for years, and rather than displace a bunch of engineers, we saw a rise in specialization around those tools—a skill set shift rather than a purge. I'm out of the industry for 2.5 years now, but I keep in contact with many who are still in it, and there doesn't seem to be quite the wave of machines replacing humans as I'd expect, given OpenAl's reported aggressiveness in enterprise sales. I don't doubt that day is coming, I'm just not yet fully sold on generative Al as the lightning bolt it's sold as. You've got more recent first-hand experience with this so l'd be curious what you've been observing on the ground. My hunch is that the vast majority of code reviews are still done by humans, even if it was initially generated by Al or by engineered systems, but that's just a hunch based on vibes and not real world data. Thanks for another great read!
Thank you for keeping us updated on this.
Solution is same pay, shorter work week.
Problem solved.