#anthropic latest economic paper, "Labor market impacts of AI".
Figure 2 in the main paper suggests we might be able to finally fulfill #JoniMitchell's vision, and get back to the garden.
https://www.anthropic.com/research/labor-market-impacts
[1/n]
Key findings:
1. "Occupations with higher observed exposure are projected by the BLS to grow less" (p.2)
* "for every 10% increase in coverage, the BLS's growth projection drops by 0.6%" (p.9);
* @aaronsojourner.org thoughts on how BLS growth projections work, how well they might take into account 'AI' impact?
[2/n]
2. "Workers in the most exposed professions are more likely to be older, female, more educated, and higher-paid"
* is this demographic relatively well placed to adapt?
* could we build a more continuous measure of tasks (e.g. in latent space rather than O*NET), and transition matrices between them?
Overall, a bit of a hacky feel (e.g. rough discretisation in beta, fudge factors in alpha), but the best shot I've seen so far.
[3/n]
Colin Rowat