Cody Cook

Kevin Schaul and Shira Ovide have an excellent write up this morning in the The Washington Post looking at recent work Sam Manning and Tomás Aguirre did on #AI exposure that my crew at The Brookings Institution wrote up as a brief. Kevin and Shira do a really nice job detailing the novel part of the analysis which focuses on which workers may be most and least able to adapt to AI, given various socioeconomic traits. They capture exactly our suggestion that while many will be "exposed" to AIs fewer (often clerical and administrative workers; often women) will struggle to shift into other good-paying work if AI destroys their jobs. Yet what I like best about Kevin and Shira's piece is how thoughtfully it contextualizes our particular labor-market analysis amid the broader AI prediction business. Kevin and Shira provide really good perspective on the fact that much AI-related work--say on labor market trends--amounts to "useful but fallible best guesses," even when the methodologies are hard core and rigorous. And the Post reporters are not wrong about the result of so much well intentioned but often contradictory work. As they write: "A flood of sometimes conflicting analyses shows the yawning gap between what is known about how AI is changing work and everyone's understandable hunger for certainty. The divide lets Americans, business leaders, and policymakers cheery-pick their preferred narratives." That's a genuine problem. Kevin and Shira are right to reference researcher Jed Kolko for declaring AI labor-market research inconclusive with key questions "unanswered." And yet, having observed the field for a while, and participated in it from time to time, would say that while the #AI predictions industry has some faults, it's also provided more clarity than many acknowledge. Good research and reporting over time is how insights come into focus, and as for AI's influence on the labor market, much of the debate is directionally suggestive and really valuable. Anyway: Here's Kevin's and Shira's story: https://lnkd.in/eSmPQsSh And here's our original @BrookingsMetro brief: https://lnkd.in/exUcEZSu Brookings Metro GovAI Jed Kolko Nathan Goldschlag Peterson Institute for International Economics Shriya Methkupally Justin Heck Papia Debroy Courtney Haynes Joe Parilla James Feigenbaum Robert Puentes Elham Tabassi Melina Kennedy Nathan Ringham Molly Kinder Xavier de Souza Briggs Nicol Turner Lee Alan Berube Gad Levanon Sanjay Patnaik

87

8 Comments

New CES Working Paper with Hang Kim and Martin Rotemberg. We show that the methods the Census Bureau uses to edit and impute data in the Annual Survey of Manufactures dramatically affect estimates of productivity dispersion, allocative efficiency, and aggregate productivity growth. Using a Bayesian approach for editing and imputation, we model the joint distributions of all variables needed to estimate these measures, and we quantify the degree of uncertainty in the estimates due to imputations for faulty or missing data. https://lnkd.in/e7CQi3rG

26