Kylie Zhang retweeted
We’re thrilled to introduce the inaugural class of Crosby Intelligence Fellows: Teddi Worledge and Irena Saracay from Stanford, and Kylie Zhang from Princeton.
When we launched the fellowship this summer, we were looking for researchers drawn to problems where ground truth is messy and the right move depends on who's across the table.
In order for the field to deliver on its potential, we believe legal AI has to solve two problems,
1. How do you measure quality rigorously when two senior attorneys can redline the same clause differently and both be right?
2. How do you build AI that sharpens a lawyer's judgment rather than replacing it?
Teddi, Irena, and Kylie’s fellowship projects approach these questions from different directions: verifying what AI systems produce, adapting review to the priorities of the person doing it, and learning the strategic choices that shape a negotiation.
Very excited to have @TeddiWorledge @irena_gao and @skyienz as fellows and to help push the frontier of legal AI research forward!
crosby.ai/blog/introducing-t…
Kylie Zhang retweeted
Researchers are actively improving memory for LLMs/VLMs. But are we measuring it right?
A system can score 100% by rereading the input for each question. If a human did that, we’d call it terrible memory!
We test beyond accuracy with ECC: Efficiency, Compression, Calibration. 🧵
Kylie Zhang retweeted
Love this. Makes you take a step back and appreciate how far we've come—but also how much more can be achieved.
ordinaryabundance.com/
