The work

Each task presents two blinded AI-generated images created from the same prompt. You will compare them side by side, choose a preferred image or mark both good or both bad, and write a short rationale grounded in visible evidence.

Evaluation criteria

Judgments should account for prompt alignment, overall visual quality, anatomy and structure, rendered text, and generation artifacts. The work rewards careful, repeatable evaluation rather than fast decisions, particularly when one image is aesthetically stronger but less faithful to the prompt.

Screening and logistics

The platform begins with a paid onboarding quiz, which must be passed before project tasks become available. Screening is likely to test whether you can notice specific image defects, apply criteria consistently and explain close calls in fluent written English. No specialist degree is required.

Pay band: $30/hr

The listing does not specify whether work is remote or asynchronous, nor does it state typical weekly hours or a fixed schedule. Candidates should confirm task availability, scheduling expectations and technical requirements directly with the platform.