AGI won't automate most jobs–because they're not worth the trouble

9 hours ago

A new economics paper argues most human work won't be automated by AGI not because AI lacks capability, but because it's not important enough to replace.
The paper distinguishes between 'bottleneck' work (essential for growth like energy, science, security) which will be automated, and 'supplementary' work (arts, hospitality, research) which AI may ignore.
In a post-AGI economy, wages become decoupled from GDP, and labor's share of GDP converges to zero, with most income going to owners of computing resources.
The distribution of compute ownership becomes the defining political challenge, with wealth concentration risks highlighted by figures like BlackRock's Larry Fink.
The paper identifies two automation transition modes: a gradual 'compute-binding' scenario and a more disruptive 'algorithm-binding' scenario resembling current leaps in AI capability.
While AGI won't make workers collectively poorer, gains may be concentrated at the top, exacerbating inequality without structural interventions like universal income or treating compute as a public resource.
The paper's title 'We Won't Be Missed' reflects that in an AGI world, work may lose its societal recognition, as human labor becomes economically marginal.

Hasty Briefsbeta