AI Trainer – GNU Image Manipulation Program Users (Remote)

About this role

AI Trainer – GNU Image Manipulation Program Users (Remote) The role We are seeking professional Graphic and Visual Designers with experience using GNU Image Manipulation Program (GIMP) to act as Domain Experts for a high-level AI evaluation project. AI models are evolving beyond simple image generation and now tackle complex design tasks such as layout, typography, image editing, composition, and information density. We need experts who can look beyond “pretty pictures” and assess the professional utility, readability, technical quality, and structural integrity of design outputs. Your experience working with GIMP — including layers, masks, paths, selections, filters, retouching, typography, and export-ready assets — will help evaluate whether AI-generated designs meet real-world creative and production standards. This is a flexible, ad hoc project where your professional eye directly influences how AI understands visual design principles and practical design workflows. Researchers looking for your skills tend to pay up to 30 USD . You must be prepared to complete paid tasks that require one hour of uninterrupted work , though many are shorter. What you’ll bring GIMP expertise and technical literacy: You must be confident using GNU Image Manipulation Program (GIMP) in a professional or advanced design context. This includes familiarity with layers, masks, paths, selections, filters, retouching, typography, composition, file formats, and export settings. A critical eye: You can distinguish between “consumer-grade” and “professional-ready” visual content, ensuring outputs are polished, readable, well-structured, and suitable for real-world use.. What you’ll be doing in the role Side-by-side preferences: Evaluate two candidate designs (e.g., Google Slides, posters, or advertisements) on the same topic and determine which better meets professional standards. Expert analysis: Judge candidates based on specific design dimensions, including readability, text density, infor