@jennahgosciak.bsky.social showed how LLM assistance can improve government caseworker accuracy in the context of SNAP eligibility questions. It was really cool to see Jennah + team get ahead of the ever-shifting technical capacity of LLMs by varying chatbot accuracy.

I also looked compared historical vegetation coverage to the present-day, estimated via LiDAR imagery -- illustrating the (obvious) point that the city looks quite different today in terms of 🌳 green space!

My takeaway: historical wetlands and the Sandy inundation zone look almost identical; the relationship to stormwater flooding and flood vulnerability is less clear.

In particular, I was interested in examining how historical wetlands in the city correspond to modern-day
⛈ flooding risk (inspired by several articles that came out shortly after Hurricane Ida, and my own experience with neighborhood flooding in Hurricane Sandy).

We also encourage everyone to check out research on pipeline aware fairness (🔗 arxiv.org/pdf/2309.17337), demographic data collection (🔗 arxiv.org/abs/2011.02282), and broader challenges with disparity assessments (🔗 dl.acm.org/doi/10.1145/3593013.3594015), which influenced our work!
(12/n)

Huge thanks to Stanford PHS for their support with using American Family Cohort data that enabled this study! Also thank you to members of the RegLab for providing helpful feedback from different perspectives!
(11/n)