The work
You will review responses from conversational AI tools and judge how well they match a user's preferences. You will make careful comparisons and explain your decisions in clear Japanese.
This is independent, asynchronous work that you can complete remotely alongside other work.
- Compare and rank alternative AI-generated responses, including responses with subtle differences.
- Explain which response better fits a user's preferences and why.
- Review tone, warmth, writing style, emoji use, and personalization.
- Complete AI evaluation tasks using your experience with ChatGPT or a similar AI assistant.
- Provide thoughtful written feedback in Japanese.
What it pays and takes
The role is designed for an individual contributor who can make nuanced judgments about conversational AI and explain those judgments clearly.
- Pay: Up to $30 per hour.
- Time: 20+ hours per week.
- Location: Open to candidates in Japan; the work is remote and asynchronous.
- Language: Fluent written and spoken Japanese, plus working English for platform instructions, training materials, and task guidance, ideally at B2 level.
- Experience: Hands-on experience using ChatGPT or a similar AI chat assistant.
- Requirement: Ability to form, explain, and defend nuanced preferences between AI responses.
- Requirement: Comfort sharing eligible ChatGPT conversation history for research and product improvement, with at least 41 conversations remaining after personal content is removed.
- Work style: Ability to work independently in a remote setting.
How it works
Apply on OpenTrain with your resume, then complete the application on the hiring site.