Source-to-Answer Consistency Specialist

Globik AI; United States; Remote; USD 17,000 - 25,000; Full time

Job summary

Globik AI is seeking Source-to-Answer Consistency Specialists to evaluate whether AI-generated answers stay faithful to their supporting sources. You will compare passages and generated responses to identify changed meanings, contradictions, omitted conditions, or unsupported additions.

Roles include documenting evidence, maintaining consistent standards across large batches, and following project rubrics. A reliable internet connection is required; some candidates may complete a practical RAG

Qualifications

• Experience with RAG search information retrieval, research content QA, or AI evaluation is a plus.

Responsibilities

• Compare source passages with AI-generated answers.
• Check whether the answer preserves the meaning of the source.
• Identify contradictions, changed facts, and unsupported additions.
• Identify important conditions or qualifications that were omitted.
• Assess whether summaries remain faithful to source material.
• Document source evidence for evaluation decisions.
• Maintain consistent standards across large batches.
• Follow project-specific RAG evaluation rubrics precisely.

Skills

RAG search experience

QA experience

Job description

Globik AI is seeking Source-to-Answer Consistency Specialists to evaluate whether AI-generated answers remain consistent with the information contained in their supporting sources.

You will compare source passages and generated responses to identify changed meanings, contradictions, omitted conditions, and unsupported additions.

Responsibilities

• Compare source passages with AI-generated answers.
• Check whether the answer preserves the meaning of the source.
• Identify contradictions, changed facts, and unsupported additions.
• Identify important conditions or qualifications that were omitted.
• Assess whether summaries remain faithful to source material.
• Categorize consistency errors.
• Document source evidence for evaluation decisions.
• Maintain consistent standards across large batches.

Qualifications

• Experience with RAG search information retrieval research content QA or AI evaluation is a plus.

• Maintain high accuracy and consistency across assigned evaluation tasks.
• Follow project-specific RAG evaluation rubrics precisely.
• Clearly document difficult or ambiguous cases.
• Participate in calibration exercises when required.
• Regular quality reviews may be conducted throughout the project.
• Selected candidates may be required to complete a practical RAG assessment before onboarding.
• A reliable internet connection and computer are required.

Pay

$12–$18/task