AI Response Evaluator (Indonesian)

The work

You will test an AI personalization feature that uses conversation history and connected personal data. Your work will help assess whether responses are relevant, helpful, natural, and supported by the available context.

• Create and run realistic multi-turn prompts, usually lasting one to five turns, using personal experiences and context.
• Check whether responses follow the original prompt and use personalization appropriately.
• Find unsupported claims, wrong inferences, hallucinations, forced connections, and unnatural overexplaining.
• Compare two responses for helpfulness, ease of use, naturalness, and overall quality.
• Write clear comparison reasons that point to specific parts of the conversation.
• Extract and check debug information to confirm whether summaries and personal data sources were used correctly.
• Delete evaluation conversations to keep personal chat history and data clean.

What it pays and takes

This is a part-time contractor role for someone who can review nuanced AI responses carefully and explain their judgment in clear written Indonesian.

• Pay: $15 USD per hour.
• Time: 20 or more hours per week.
• Location: Worldwide and remote.
• Language: High proficiency reading and writing Indonesian.
• Level: Entry level.
• Equipment: A desktop or laptop and a reliable internet connection.
• Work style: Ability to work independently.
• Required strengths: Analytical thinking, creative multi-turn prompt writing, attention to detail, and concise written communication.
• Helpful background: A bachelor's degree or equivalent experience in policy, law, ethics, linguistics, journalism, computer science, or a related analytical field.
• Preferred experience: Data annotation, AI quality evaluation, content moderation, or related work.

How it works

Apply on OpenTrain with your resume and then complete the application on the hiring site.

About AI training work

AI training work is the human work behind systems that learn from examples, including reviewing model responses and rating their quality. People with strong language, analytical, or subject knowledge help make those systems more accurate and useful.