This entry-level, worldwide freelance role involves evaluating how computer-using AI agents understand instructions, reason through problems, and complete tasks. Contractors will review model responses, fact-check claims online, solve logic puzzles, and write quality annotations for 20 or more hours per week. Candidates must have strong English reading comprehension and analytical skills.