The comparison between today’s AI backlash and earlier reactions to the printing press, mechanised looms, newspapers, radio or television is rhetorically appealing, but conceptually misleading. Those earlier technologies expanded the reach of human cognition or the scale of human labour. They did not substitute for the underlying cognitive competencies that defined skilled work.
A scribe displaced by print retained his intellectual agency; a textile worker displaced by machinery retained productive potential within the industrial order. What changed was the mode of production, not the nature of human capital. AI is different. Contemporary models encroach directly on the cognitive functions such as analysis, synthesis, abstraction, and linguistic generation, which historically formed the basis of high-skill labour markets. This represents a structural, not cyclical, shift: the automation of cognition itself rather than its amplification.
Moreover, the standard “compensation effect” invoked in historical analogies rests on two conditions that no longer reliably hold: the emergence of new labour-absorbing sectors and broadly shared gains. Both are weakened today. The economic sectors that once absorbed displaced workers are now automatable, and productivity gains increasingly accrue to capital rather than labour. Engels’ Pause may not recur as a temporary adjustment phase but as a systemic feature.
The political economy context is also distinct. Those now exposed to displacement, whether lawyers, academics, designers, or medical professionals, constitute the professional-managerial class that shapes policy, regulation, and media narratives. Their capacity to condition technological diffusion is far greater than that of Luddites or medieval scribes.
Furthermore, the recurrence of familiar moral or epistemic anxieties does not imply that the underlying technological effects are equivalent. Plato was wrong about writing; critics of algorithmic social media were essentially correct. Not every panic is unfounded simply because it resembles an older one.
AI is not merely another stage in the historical saga of labour-saving technologies. It challenges the scarcity of human cognitive labour itself, making historical analogy a poor guide to the scale and nature of the transformation.
I call this the Ad Normalis fallacy: the recurring notion that we shouldn’t worry about innovation X because people in the past were concerned about innovation Y and it turned out fine. The problem is that we accept innovation Y because it is normal to us, not because humanity collectively decided it was a good idea. We were born after its advent, or else we were young enough for it to be exciting, so it poses no threat to us.
Since radical innovation has previously been slow enough for us to become generally accustomed to a constant level of technology for an entire lifetime, this has usually been framed as an “old people mad” sort of thing.
But now that the window is shifting to decades (or even years) between radical innovations, it is untenable to deploy Ad Normalis. For one, it seems to me that we have drifted very, very far from anything close to normal. The post-war 20th century view that technology is a tool we use to effect stability and progress was distorted by a lens that could only see but one major quake of innovation, followed by its calm and reliable GDP-boosting aftershocks. And now that we are facing exponential innovation, it is clearer that it is a destabilizing process that uses us as its medium -- not the other way around.
The monk wary of the press and the statesman worried about newspapers were both, in a sense, correct. The printing press and its further iterations did fundamentally change their ways of life and alter their livelihoods.
If we take each example independently, it would seem that these are the complaints of the losers: how absurd that anyone would seek to ban something as vital as the printing press, or the loom!
But if we zoom out, it becomes clear that this is not a few bygones fighting progress: it is a few isolated examples of humans from unique vantage points witnessing a process that destroys normality.
I think the views on cultural resistance to change from a historic perspective is interesting. Whilst there is fear mongering with AI, there is also the over selling of the use case, inflating it’s actual value (as in the case of several ‘new’ ideas) and it’s hard to work out if those investments are real use cases.
I think its less about backlash and more about controlling it so that it minimizes the harm. Right now, the harm it appears to be much much more extensive and the coalitions being formed to contain it, thus, are accordingly much larger and serious.
The comparison between today’s AI backlash and earlier reactions to the printing press, mechanised looms, newspapers, radio or television is rhetorically appealing, but conceptually misleading. Those earlier technologies expanded the reach of human cognition or the scale of human labour. They did not substitute for the underlying cognitive competencies that defined skilled work.
A scribe displaced by print retained his intellectual agency; a textile worker displaced by machinery retained productive potential within the industrial order. What changed was the mode of production, not the nature of human capital. AI is different. Contemporary models encroach directly on the cognitive functions such as analysis, synthesis, abstraction, and linguistic generation, which historically formed the basis of high-skill labour markets. This represents a structural, not cyclical, shift: the automation of cognition itself rather than its amplification.
Moreover, the standard “compensation effect” invoked in historical analogies rests on two conditions that no longer reliably hold: the emergence of new labour-absorbing sectors and broadly shared gains. Both are weakened today. The economic sectors that once absorbed displaced workers are now automatable, and productivity gains increasingly accrue to capital rather than labour. Engels’ Pause may not recur as a temporary adjustment phase but as a systemic feature.
The political economy context is also distinct. Those now exposed to displacement, whether lawyers, academics, designers, or medical professionals, constitute the professional-managerial class that shapes policy, regulation, and media narratives. Their capacity to condition technological diffusion is far greater than that of Luddites or medieval scribes.
Furthermore, the recurrence of familiar moral or epistemic anxieties does not imply that the underlying technological effects are equivalent. Plato was wrong about writing; critics of algorithmic social media were essentially correct. Not every panic is unfounded simply because it resembles an older one.
AI is not merely another stage in the historical saga of labour-saving technologies. It challenges the scarcity of human cognitive labour itself, making historical analogy a poor guide to the scale and nature of the transformation.
Interesting piece!
Ah, the famous poet and statesman, Lord "Bryon" ;)
https://historyofinformation.com/detail.php?id=4654
More importantly -- a great, insightful article!
I call this the Ad Normalis fallacy: the recurring notion that we shouldn’t worry about innovation X because people in the past were concerned about innovation Y and it turned out fine. The problem is that we accept innovation Y because it is normal to us, not because humanity collectively decided it was a good idea. We were born after its advent, or else we were young enough for it to be exciting, so it poses no threat to us.
Since radical innovation has previously been slow enough for us to become generally accustomed to a constant level of technology for an entire lifetime, this has usually been framed as an “old people mad” sort of thing.
But now that the window is shifting to decades (or even years) between radical innovations, it is untenable to deploy Ad Normalis. For one, it seems to me that we have drifted very, very far from anything close to normal. The post-war 20th century view that technology is a tool we use to effect stability and progress was distorted by a lens that could only see but one major quake of innovation, followed by its calm and reliable GDP-boosting aftershocks. And now that we are facing exponential innovation, it is clearer that it is a destabilizing process that uses us as its medium -- not the other way around.
The monk wary of the press and the statesman worried about newspapers were both, in a sense, correct. The printing press and its further iterations did fundamentally change their ways of life and alter their livelihoods.
If we take each example independently, it would seem that these are the complaints of the losers: how absurd that anyone would seek to ban something as vital as the printing press, or the loom!
But if we zoom out, it becomes clear that this is not a few bygones fighting progress: it is a few isolated examples of humans from unique vantage points witnessing a process that destroys normality.
I think the views on cultural resistance to change from a historic perspective is interesting. Whilst there is fear mongering with AI, there is also the over selling of the use case, inflating it’s actual value (as in the case of several ‘new’ ideas) and it’s hard to work out if those investments are real use cases.
I think its less about backlash and more about controlling it so that it minimizes the harm. Right now, the harm it appears to be much much more extensive and the coalitions being formed to contain it, thus, are accordingly much larger and serious.