What it does
LegalForecastBench provides an alpha benchmark and evaluation workflow for LegalForecast-MTD.
Benchmarks & research
LegalForecastBench
Bookmark: LegalForecastBench
Keep this in your collection.
Leave a noteWhat would you try with this?
: LegalForecastBench
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
How you can use it
You can test how well different AI models predict federal court rulings. Before testing, choose which models to compare on motions to dismiss, which are requests to throw out a lawsuit early. A developer can install the tool and request access to the project's separate, protected dataset of court filings.
The software runs models without web access and scores their predictions against actual decisions. Keep in mind that published training dates do not guarantee an AI has never seen a case. Also, cloud AI services use their own settings, so asking the same AI twice can produce slightly different answers.