A Multitudes study recently cited in Scientific American showed exactly this.
AI led to not only longer hours overall, but also a shift from development to bug fixing and a 19.6% increase in out-of-hour commits. So longer hours, less interesting tasks, and more weekend work.
This is also my experience. At [my company name], we have a coding agent integrated with Slack that allows people to simply @bot to complete no brainer tasks. And the result is there is no more easy task, because easy task is now automated, and everything left are mentally exhausting task.
Yeah, it’s very good for productivity & focus to have some easy tasks to give you a quick win & dopamine boost. Spending all day reviewing AI output is NOT that.
I suppose some say this is an argument in favor of non-human workers, but the whole point of this endeavor should be to improve human society. (Isn’t that, allegedly, what tech companies are all about? :p)
Indeed, since you can't relax by doing the mindless boilerplate that AI can do for you. You instead spend your time on more engaged, high leverage activities.
its pretty clear that especially because AI have become worker-level tools that the negative externalities are passed to the consumer (worker). This means that we will be expected to keep up with an increasing pace of work because johnson in the cubicle next door is pumping out tons.
and better yet for the corporate overlord, the worker becomes more replaceable as well. their knowledge is less valuable and their overall skill is also less. they are in a sort of perpetual race to the bottom where they are just the meat monkey in the loop that is only valuable for keeping their agent(s) on task.
profits go up but the company knows in their darkest nights that they are also easily replaced. all of capitalism is just a house of cards waiting for the next model to pull the final card out from beneath us and then there will be truly nothing left except a handful of frontier ai companies who are fully integrated with government powers for all 'lawful purposes.'
the riots in the streets are easily quelled without a human in the loop.
This is exactly my experience. Because the barrier to doing things is lower, you just do more things. But it doesn't actually expand one's capabilities or make learning happen any faster. That needs time (mostly sleep).
Compare playing Counter Strike with having a game of cricket.
Cricket takes a while to organise and set up. You have to contact your pals, physically travel to the field and set up the wickets. Then expect to commit several hours to the game.
Meanwhile, Counter Strike takes no effort. You move your arm to the mouse and click. A game is like 5 minutes of commitment.
The trouble is, before you know it you've spent 5 hours on Counter Strike. You could have easily played cricket in that time, caught up with your pals and exercised your body to boot.
Able to improve at CS because the time between rounds is so short (even instant almost in death match etc). Wonder if any relation to how fast time can fly
Yeah, when I used to play CS it was incredible how much time you could "accidentally" spend on it. The commitment per round is so low you just end up having "one more game" over and over again. It's what I've found with coding agents except it's "one more feature". There's less thought going into what is worth doing and what isn't, so you the up doing more in the long run.
Care to explain why? I think it’s not a bad analogy: contrasting two things with the same goal, one fast paced, saturates your cognitive load , and one that leaves more space for reflecting on what is actually happening, inherently less exhausting. You may disagree with the analysis, but it works in this framework.
Creating is enjoyable. Reviewing the creation output of AI as a career would certainly be "harder" in the sense that it would be unenjoyable and make work feel like a chore when it doesn't have to.
A Multitudes study recently cited in Scientific American showed exactly this.
AI led to not only longer hours overall, but also a shift from development to bug fixing and a 19.6% increase in out-of-hour commits. So longer hours, less interesting tasks, and more weekend work.
https://www.scientificamerican.com/article/why-developers-us...
This is also my experience. At [my company name], we have a coding agent integrated with Slack that allows people to simply @bot to complete no brainer tasks. And the result is there is no more easy task, because easy task is now automated, and everything left are mentally exhausting task.
Yeah, it’s very good for productivity & focus to have some easy tasks to give you a quick win & dopamine boost. Spending all day reviewing AI output is NOT that.
I suppose some say this is an argument in favor of non-human workers, but the whole point of this endeavor should be to improve human society. (Isn’t that, allegedly, what tech companies are all about? :p)
https://archive.is/2026.03.06-155838/https://www.ft.com/cont...
Indeed, since you can't relax by doing the mindless boilerplate that AI can do for you. You instead spend your time on more engaged, high leverage activities.
its pretty clear that especially because AI have become worker-level tools that the negative externalities are passed to the consumer (worker). This means that we will be expected to keep up with an increasing pace of work because johnson in the cubicle next door is pumping out tons.
and better yet for the corporate overlord, the worker becomes more replaceable as well. their knowledge is less valuable and their overall skill is also less. they are in a sort of perpetual race to the bottom where they are just the meat monkey in the loop that is only valuable for keeping their agent(s) on task.
profits go up but the company knows in their darkest nights that they are also easily replaced. all of capitalism is just a house of cards waiting for the next model to pull the final card out from beneath us and then there will be truly nothing left except a handful of frontier ai companies who are fully integrated with government powers for all 'lawful purposes.'
the riots in the streets are easily quelled without a human in the loop.
This is exactly my experience. Because the barrier to doing things is lower, you just do more things. But it doesn't actually expand one's capabilities or make learning happen any faster. That needs time (mostly sleep).
Compare playing Counter Strike with having a game of cricket.
Cricket takes a while to organise and set up. You have to contact your pals, physically travel to the field and set up the wickets. Then expect to commit several hours to the game.
Meanwhile, Counter Strike takes no effort. You move your arm to the mouse and click. A game is like 5 minutes of commitment.
The trouble is, before you know it you've spent 5 hours on Counter Strike. You could have easily played cricket in that time, caught up with your pals and exercised your body to boot.
Able to improve at CS because the time between rounds is so short (even instant almost in death match etc). Wonder if any relation to how fast time can fly
Yeah, when I used to play CS it was incredible how much time you could "accidentally" spend on it. The commitment per round is so low you just end up having "one more game" over and over again. It's what I've found with coding agents except it's "one more feature". There's less thought going into what is worth doing and what isn't, so you the up doing more in the long run.
Another addition to the list of really bad analogies on here.
Care to explain why? I think it’s not a bad analogy: contrasting two things with the same goal, one fast paced, saturates your cognitive load , and one that leaves more space for reflecting on what is actually happening, inherently less exhausting. You may disagree with the analysis, but it works in this framework.
Creating is enjoyable. Reviewing the creation output of AI as a career would certainly be "harder" in the sense that it would be unenjoyable and make work feel like a chore when it doesn't have to.