Help create visual challenges to evaluate how AI systems research and understand real-world images. You'll combine local cultural knowledge, web research, and basic image editing to create clear, evidence-backed evaluation tasks.
What you'll do
- Source authentic images of real-world places, products, and media from your assigned locale. Use images you have permission to use, and document their source and usage permissions.
- Investigate image subjects using Google Search, Google Lens, reverse image search, and other reliable sources.
- Crop, mask, or obscure identifying details while preserving realistic visual clues.
- Write clear, time-bound prompts in English without revealing the target identity or answer.
- Create concise, objective grading criteria and solution notes that explain the correct answer, supporting evidence, and research process.
- Organize your work in shared Google Drive folders using the required file names and submission structure.
What you'll bring
- 1+ years of experience.
- Native or highly fluent French, with strong working English.
- Strong knowledge of French-speaking cultures, including less-touristy landmarks, universities, municipal buildings, and domestic film and media.
- Confidence researching from partial visual clues and checking information across sources.
- Basic image-editing skills, such as cropping and masking, and the ability to check and document image usage permissions.
- Attention to detail and comfort following structured guidelines.
- A bachelor's degree or equivalent practical experience in any field.
Experience in AI evaluation, data annotation, content review, quality assurance, or similar analytical work is helpful but not required.
Project details
Freelance project for contributors based in Belgium, Switzerland, Luxembourg, Benin, Madagascar, Algeria. Availability of 20 hours per week part-time or 40 hours per week full-time is required, including at least four hours overlapping Pacific business hours.