# Harvey — How We Use ELO Scores to Build Better Legal AI

- Company: Harvey (harvey.ai)
- Announced: 2026-02-26T16:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.harvey.ai/blog/using-elo-scores-to-build-better-legal-ai
- Record: https://forck.live/items/17606-how-we-use-elo-scores-to-build-better-legal-ai
- Subject: Harvey

Harvey describes how it uses ELO scores, derived from lawyer preference evaluations, to measure and guide improvements to its AI systems. The post explains that an ELO gap of 150–200 points reflects a clear preference (about 70% preference rate), while a 400-point gap marks a step-change. Examples include building an Australia knowledge source and rebuilding the EDGAR system, where human-written answers set a benchmark for improvement.

## Evidence

Verbatim from https://www.harvey.ai/blog/using-elo-scores-to-build-better-legal-ai:

> ELO is a rating system for competitors originally used in chess and later adopted to measure skill in many competitive games. It converts a large number of head-to-head wins and losses into a single, comparative measure of how good a competitor (e.g., an AI system) is.

---

Record: https://forck.live/items/17606-how-we-use-elo-scores-to-build-better-legal-ai
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md