The economist managing $700 billion just said AI's job math will look like China's WTO entry—but he's betting you remember only the second half of that story.
The Summary
- Torsten Slok, chief economist at Apollo Global Management, argues AI productivity gains will create more jobs than they destroy, mirroring the net-positive employment outcome after China joined the WTO
- The comparison is selective: he's invoking the "best aspects" of the China Shock while sidestepping the decade of manufacturing displacement that came first
- For workers watching AI automate their roles right now, this is a promise that the pain comes before the payoff
The Signal
Slok's framing is careful. He's drawing a parallel to China's 2001 WTO accession, an event that ultimately expanded global employment but also hollowed out America's industrial Midwest for over a decade. The "China Shock" displaced roughly 2.4 million U.S. manufacturing jobs between 1999 and 2011. New jobs came later, in different sectors, in different places, for different people.
The key word in Slok's thesis is "probably." He's confident AI will be net-job-positive, but the timeline and distribution matter more than the aggregate. If AI follows the China Shock playbook, we're looking at a brutal transition period where certain job categories vanish faster than new ones materialize.
"Productivity gains will probably result in more jobs being created than are lost—but probably doesn't pay this month's rent."
Apollo manages $733 billion in assets. Slok's view isn't academic. It's the intellectual scaffolding for how one of the world's largest alternative asset managers is positioning capital. If they believe AI drives long-term job growth, they're funding the companies automating work today with the expectation that demand expansion and new job categories will absorb displaced workers tomorrow.
The China parallel cuts both ways:
- Yes, global trade ultimately created more jobs than it destroyed
- But those jobs required geographic mobility, retraining, and a decade-plus adjustment window
- The workers who lost manufacturing jobs in 2003 didn't become software engineers in 2004
The Implication
If Slok is right, the question isn't whether AI creates jobs—it's whether the transition happens fast enough and equitably enough to avoid a lost generation of workers. The China Shock took a decade to resolve into net job growth. AI is moving faster. That compression could mean quicker recovery or deeper dislocation.
For anyone whose job touches information work, the message is clear: the productivity dividend is coming, but you might not be the one collecting it unless you're building the tools or positioning yourself in the roles AI can't replicate yet. Slok is optimistic about the macro. He's quiet about the micro.