Twitxy Saved

Neel Guha

Verified account This account is verified. Learn more

@NeelGuha

Associate Professor @ColumbiaLaw | JD @StanfordLaw PhD @StanfordAILab | tech, law, data
neelguha.com Joined November 2018
903 Following 1.9K Followers 615 Posts

Latest posts 5

Pinned post

Neel Guha Verified account @NeelGuha
We released LegalBench in August 2023. In the three years since, it's become one of the more widely used benchmarks for legal AI. In the spirit of critical self-reflection, I wrote some thoughts about (1) what went well, (2) what I wish I'd done differently, and (3) what this all means for the future of legal AI research.

I wanted to highlight a few specific reflections.
9 replies 27 reposts 153 likes 19K views View on X
Neel Guha Verified account @NeelGuha
Just to add to this: I think public evals also build communities, and that's so important for emerging areas/applications/domains.
Nathan Lambert Verified account @natolambert 1d
I don’t agree with these statements. many public evals shift the Overton window to interesting new things. Then they are goodhart’d and private ones take over. Where you are in the cycle is important.
View quoted post on X
1 replies 2 reposts 42 likes 1.5K views View on X