Human evaluation means people, not software, judge how good an AI model's answers are.
In plain words
Many AI tasks have no single right answer: a translation or a summary can be good in several ways. For tasks like these, people judge the outputs, either rating them or picking the better of two. Large public comparisons of chatbots collect such votes from many users, and those votes can agree well with expert ratings.
A quick example
To compare two translation models, bilingual reviewers read both translations of the same sentences, without knowing which model wrote which, and pick the better one each time. The model that wins more often is judged the better translator.
Where you'll meet it
- Comparing chatbots and translation models.
- Judging tasks that have no single right answer.
Related
Sources
- Docshuman evaluation, Google Machine Learning Glossary · read 29 Sept 2026
- PaperChatbot Arena: An Open Platform for Evaluating LLMs by Human Preference, Chiang et al. (arXiv) · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 29 Sept 2026. Spotted an error? Report it.