Did you see the recent job numbers? Men in the heart of their prime working years are dropping out of the labor force at an alarming rate, a trend that seems to be accelerating since the advent of generative AI.
Curious. One hypothesis: more likely to drop out with belief they can make something with AI. But another is that it has nothing to do with AI. Scott Galloway discusses this a bit: young men today are increasingly not in training, education, or a job and are more and more often dependent on parents, spouses, or friends in young adulthood. 40% of them in certain age ranges. Gambling and prediction markets and perhaps even AI delusion (I've got the next big idea!) targets the risk-taking behaviors of their high-testosterone years and overwhelms the less exciting impulses of staying present in the workforce and family.
Just curious ... could you look back for the firms adopting AI and see what their hiring patterns for two years prior to substantial AI investment might have been?
"A firm is classified as adopting AI in the first month of its earliest three-consecutive-month spell with at least $100 of AI vendor spend in each month. High-intensity adopters generally spend much more. This screens out one-off purchases and captures sustained paid AI use."
So, if you spend $100 a moonth for 3 months you are "adopting AI"? And if you hire 10% over two years (5% headcount growth in first year and 4% the next), you are in this cateogory you would consider who are seeing gains from AI. Is that correct?
That’s our minimum definition of treatment but the vast majority of firms spend much more. Especially the high-intensity adopters, which are defined as the top third of business per employee spend on AI. Note, we only see positive gains for the high-intensity firms.
The implication of "Total headcount rises after AI adoption" is that it is sustainable growth for these companies that would keep these jobs over time. It is implied, or at least that is what one would surmise by reading the paper.
If companies like Uber and Microsoft cannot make the economics work, shouldn't we focus on economics that is sustainable before we surmise that "headcount rises after AI adoption?"
Great work on this! We need to shine more perspectives on the outcome of AI. People are tough to persuade here, for obvious reasons, but this helps immensely with progress in this regard
Is there any reason this couldn't be explained as "firms which adopted AI see competitive advantage, leading to growth which includes growth in human employment due to incomplete adoption of AI"?
A situation which would cease to exist as adoption of AI broadens and deepens.
I think that’s possible, and to some extent likely: firms that use AI will take labor share from firms that don’t. Note that’s still a novel finding for people whose prior is that firms that use AI will hire less.
I'm sure you know this. How do we deal with selection effects? Could it be that better firms do more AI? (Maybe you have fought this in your paper, but in the blog article you are only saying reduced form averages).
I would think firms are just having senior staff train AI agents on their specific workflows. AI handles 80% of the baseline, juniors do the first pass to make sure it doesn't sound like AI slop, and seniors give the final nod.
Which totally flips what the expectation was that AI would get rid of entry-level. If your top talent can suddenly output 10x more work, you don't need more top talent you just need a layer of cheaper, tech-native eyes to check the AI from obvious hallucinations or errors before production.
Did you see the recent job numbers? Men in the heart of their prime working years are dropping out of the labor force at an alarming rate, a trend that seems to be accelerating since the advent of generative AI.
Why would AI have a disproportionate impact on men?
Curious. One hypothesis: more likely to drop out with belief they can make something with AI. But another is that it has nothing to do with AI. Scott Galloway discusses this a bit: young men today are increasingly not in training, education, or a job and are more and more often dependent on parents, spouses, or friends in young adulthood. 40% of them in certain age ranges. Gambling and prediction markets and perhaps even AI delusion (I've got the next big idea!) targets the risk-taking behaviors of their high-testosterone years and overwhelms the less exciting impulses of staying present in the workforce and family.
Just curious ... could you look back for the firms adopting AI and see what their hiring patterns for two years prior to substantial AI investment might have been?
"A firm is classified as adopting AI in the first month of its earliest three-consecutive-month spell with at least $100 of AI vendor spend in each month. High-intensity adopters generally spend much more. This screens out one-off purchases and captures sustained paid AI use."
So, if you spend $100 a moonth for 3 months you are "adopting AI"? And if you hire 10% over two years (5% headcount growth in first year and 4% the next), you are in this cateogory you would consider who are seeing gains from AI. Is that correct?
That’s our minimum definition of treatment but the vast majority of firms spend much more. Especially the high-intensity adopters, which are defined as the top third of business per employee spend on AI. Note, we only see positive gains for the high-intensity firms.
Thanks Ara. And why do you think this trend is sustainable?
I didn't make that assertion
The implication of "Total headcount rises after AI adoption" is that it is sustainable growth for these companies that would keep these jobs over time. It is implied, or at least that is what one would surmise by reading the paper.
If companies like Uber and Microsoft cannot make the economics work, shouldn't we focus on economics that is sustainable before we surmise that "headcount rises after AI adoption?"
Great work on this! We need to shine more perspectives on the outcome of AI. People are tough to persuade here, for obvious reasons, but this helps immensely with progress in this regard
Is there any reason this couldn't be explained as "firms which adopted AI see competitive advantage, leading to growth which includes growth in human employment due to incomplete adoption of AI"?
A situation which would cease to exist as adoption of AI broadens and deepens.
I think that’s possible, and to some extent likely: firms that use AI will take labor share from firms that don’t. Note that’s still a novel finding for people whose prior is that firms that use AI will hire less.
I'm sure you know this. How do we deal with selection effects? Could it be that better firms do more AI? (Maybe you have fought this in your paper, but in the blog article you are only saying reduced form averages).
I would think firms are just having senior staff train AI agents on their specific workflows. AI handles 80% of the baseline, juniors do the first pass to make sure it doesn't sound like AI slop, and seniors give the final nod.
Which totally flips what the expectation was that AI would get rid of entry-level. If your top talent can suddenly output 10x more work, you don't need more top talent you just need a layer of cheaper, tech-native eyes to check the AI from obvious hallucinations or errors before production.