← Back to jobs

AI training & evaluation

Cultural Data Annotator - English (Singapore)

OneForma

Join the MARS Cultural Annotation Project and help shape the future of culturally aligned artificial intelligence.

Work arrangement & location
Singapore
Engagement
Fixed Rate Per Hour
  • Annotation
  • Data Annotation

Employer description and requirements

About This Project

Join the MARS Cultural Annotation Project and help shape the future of culturally aligned artificial intelligence. This large-scale initiative focuses on collecting high-quality cultural data that will be used to train and align Large Language Models (LLMs). As a Cultural Data Annotator, you will provide authentic insights into cultural values and perspectives, contributing directly to the development of more culturally aware AI systems.

Why This Work Matters

The MARS Project aims to collect and validate culturally relevant textual data that reflects the values, beliefs, and social norms of your country. This data will be used to improve the cultural understanding and alignment of AI models such as ChatGPT, Gemini, Claude, and Sea-Lion.

Requirements

Native-level fluency in English and strong familiarity with your country’s culture and societal values.

Must currently reside in the target countries.

Ability to distinguish between:

Cultural Perspective: What most people in your country would consider appropriate or acceptable.

Personal Perspective: Your own individual beliefs and opinions.

Complete validated survey questionnaires and annotate AI-generated responses.

Ensure annotations reflect authentic cultural understanding and human judgment.

Strong attention to detail and ability to follow detailed guidelines.

Reliable internet connection and computer access.

No prior annotation experience required.

Project Details

Location: Remote

Minimum productivity target: 10 completed annotation samples within 3.5 hours.

Annotation work includes a balanced mix of Create, Select, and Refine tasks.

Continuous quality assurance reviews and random audits will be conducted.

Annotations must reflect current cultural norms and avoid stereotypes or outdated generalizations.

Non-compliant submissions may require rework and may impact continued participation.