Data Labeling Experience – AI Training & Evaluation (Outlier) Type of data labeled: Text data — including AI-generated
Data Labeling Experience – AI Training & Evaluation (Outlier) Type of data labeled: Text data — including AI-generated responses, prompts, and written content Tasks performed: Preference labeling — comparing two or more AI responses and labeling which one is better based on criteria like accuracy, helpfulness, and clarity Quality annotation — rating individual responses on scales (e.g. factual correctness, coherence, tone) Prompt creation — generating labeled input examples used to train and test AI models Instruction-following evaluation — labeling whether a model's output correctly followed a given instruction Creative content labeling — writing or reviewing text samples tagged by style, format, or quality Code evaluation — annotating AI-generated code for correctness, efficiency, and adherence to requirements