My Collection of Empirical Econ Papers on AI
Lots of new empirical papers on AI have been released. To help keep everything straight, I’ve been keeping an updated table of everything I’ve found.
Friends,
A few months ago, I told you that I was restarting Exformation and working on a bunch of new things for this Substack and at AEI. Part of the reason I’ve been so quiet, with only a few posts in the last couple months, is that I spent a good amount of time this summer in founder mode, thinking about projects, working on papers, and executing ideas. I am excited for all of the new work that I am going to be rolling out soon and into next year.
I will be honest. I haven’t been the best about sending my AEI work through this Substack. So if you want to get caught up on my AI work, I wrote an extended summary for my website and I updated my publications list. Otherwise, here are a couple recent highlights:
My latest is titled “Procedural Rituals Over Governance Results” and is a reflection on Jen Pahlka’s talk at the Roots of Progress conference. It builds on Nicholas Bagley’s concept of procedure fetish, ultimately coming to the conclusion that: “Our procedural culture doesn’t just make it harder for institutions to function: It also shapes the public’s instinct to say no. The tragedy is that the impulse behind participatory democracy—the desire for inclusion and fairness—has translated into bureaucratic systems that confuse involvement with improvement. Until we rediscover forms of governance that reward results instead of rituals, we’ll keep mistaking the appearance of responsiveness for the real thing.”
In “Senator Sanders’ AI Report Ignores the Data on AI and Inequality,” I did some digging into Senator Bernie Sanders’ new report which claims that 100 million jobs will be lost in the next ten years due to AI. Beyond the issues with how the AI job loss model was constructed (it simply asked ChatGPT), my biggest concern with the report is that it “reviews some key papers on automation and income inequality, but nowhere does it review the current literature showing that new AI tools are reducing inequality. In Brynjolfsson et al. (2023); Caplin et al. (2024); Choi et al. (2023); Hoffmann et al. (2024); Noy & Zhang (2023); and Hauser & Doshi (2024), advanced AI tools were found to be skill equalizers, raising the performance of those at the bottom in customer support, legal work, and software development, among others. If Sanders was truly concerned with worker inequality, he should be optimistic about AI tools and engaging with the empirical work on this subject.”
In “The AI Revolution in Property Tax Assessment,” I dove into one of the lesser known applications of AI tools in property tax assessment. As I wrote, “traditional assessment methods face a litany of problems. Valuations can often be inconsistent and municipalities are typically understaffed and resource-constrained.” This is where AI comes in because models can be trained on property characteristics, sales data, and market trends “to address the core challenges of traditional property tax assessment methods.”
Please don’t hesitate to send me ideas for future work and if you don’t mind sharing these emails with colleagues that will help me grow this site. Now on to the good stuff…
Empirical economists have been busy during the last couple of months trying to understand how LLMs are impacting companies and labor markets. To help keep everything straight, I spent some time in April compiling everything I could find into a table. Since then, I have been updating it with high quality papers. It excludes surveys, dataset, and indices because I am working to collect those for another project. But as I find material, the table below will be updated, so please send me what you’ve got!
This article comes at the perfect time, as confusing proces with progress seems like a major design flaw in our social algorithms, and I'm eager to read more.