If there’s a human prompting and judging the AI, then why not pay $10k in Philippines rather than $100k in the US for near identical output? And if no human is needed at all, perhaps there’ll be a jobs crisis everywhere, not just in LMICs.
LMIC workers being priced out but rich country workers being fine would seem to require that AI performance saturate in a narrow band of intermediate performance/autonomy. Plausible, but an unusual coincidence.
Good question. This is similar to the objection I address in the piece. A few points:
- I don't think this requires AI to saturate in a narrow band. The work LMICs export is lower on the skill chain and already segmented away from the rest of companies' operations meaning there are fewer interdependencies with other departments, so it's likely to be automated first. Being hit harder and sooner will matter even if other countries are also impacted.
- On paying $10k in the Philippines: much of this work is already there, so there aren't so many $100k US jobs to get rid of. Maybe some human supervision jobs remain in those countries, but probably very few jobs are created by that and it still looks like a headcount collapse.
- There may be a serious jobs problem in rich countries too, but it's a different kind of problem, especially in the US. Where jobs are automated in the US, the displacement is correlated with wealth creation from the productivity gains (since the AI firms doing the automating are largely US-based), so the question is more about whether we can do redistribution well.
AI's impact is more uncertain because it's applications and capabilities are uncertain.
Why not focus more on the impact of the green transition and emergence of lab grown proteins on developing countries?
My assumptions is that the green transition and lab grown meats will make Latin America, Middle East, West Africa poorer but will make most of Asia and East Africa richer.
Great piece!
Thanks!
If there’s a human prompting and judging the AI, then why not pay $10k in Philippines rather than $100k in the US for near identical output? And if no human is needed at all, perhaps there’ll be a jobs crisis everywhere, not just in LMICs.
LMIC workers being priced out but rich country workers being fine would seem to require that AI performance saturate in a narrow band of intermediate performance/autonomy. Plausible, but an unusual coincidence.
Good question. This is similar to the objection I address in the piece. A few points:
- I don't think this requires AI to saturate in a narrow band. The work LMICs export is lower on the skill chain and already segmented away from the rest of companies' operations meaning there are fewer interdependencies with other departments, so it's likely to be automated first. Being hit harder and sooner will matter even if other countries are also impacted.
- On paying $10k in the Philippines: much of this work is already there, so there aren't so many $100k US jobs to get rid of. Maybe some human supervision jobs remain in those countries, but probably very few jobs are created by that and it still looks like a headcount collapse.
- There may be a serious jobs problem in rich countries too, but it's a different kind of problem, especially in the US. Where jobs are automated in the US, the displacement is correlated with wealth creation from the productivity gains (since the AI firms doing the automating are largely US-based), so the question is more about whether we can do redistribution well.
AI's impact is more uncertain because it's applications and capabilities are uncertain.
Why not focus more on the impact of the green transition and emergence of lab grown proteins on developing countries?
My assumptions is that the green transition and lab grown meats will make Latin America, Middle East, West Africa poorer but will make most of Asia and East Africa richer.