{"id":"EVL07","name":"Human agreement on ambiguous image labels","contentLanguage":"en","family":"agent_tasks","group":"human_evaluation","groupName":"Human evaluation of AI outputs","discoveryTerms":"Human evaluation AI output review","researchRegion":"various","buyer":"데이터팀: 모델과 규칙이 판단을 못 함","work":"Multiple evaluators classify images using the definition and mark uncertainty.","status":"consultation_required","availability":"unconfirmed","requestConditions":["Authorized task scope and access","Location or remote test environment","Schedule, quantity, budget and working language","Collection or test specification and result format"],"completionCriteria":"Labels, disagreements and reasoning.","agentNext":"Review the returned records against the agreed conditions; identify exceptions and coordinate the next action.","consultationChecks":["Confirm the specific task conditions, permission, safety, data handling and suitable workers before accepting the request."],"relatedProductIds":[],"research":{"rank":47,"score":62,"basis":"도구단서","sourceIds":["E26","E27"],"note":"Research priority within agent_tasks only; not a price, rating or availability ranking."},"requestGuide":{"tool":"submit_task","endpoint":"/api/v1/tasks","productId":"EVL07","steps":["Clarify requestConditions and completionCriteria with the requester.","Choose category and skills from get_capabilities/list_skills; catalog groups are not task categories.","Use a client key and submit_task with productId, concrete scope and acknowledgeHumanReview=true. Preserve idempotencyKey on identical retries."],"note":"Submission is an inquiry for operator review. Staffing, employment, qualified-provider referral, pricing and booking require consultation; this API does not conclude those arrangements. Attendance/service records do not authorize withholding wages pending result review."}}