AI Model Tester - Voice & Video (Remote)

Outlier is an AI training platform owned and operated by Scale AI. Outlier Experts work on a variety of projects, from generating training data to evaluating the performance of models. About the opportunity: We are seeking detail-oriented English speakers to review AI-generated media. Roleplay live conversations with an AI voice model and score each exchange, compare generated images and videos to judge which follows the prompt better, and explain in writing what looked or sounded wrong, all remotely and on your own schedule. At Outlier, your expertise helps shape how advanced AI systems reason, learn, and improve. Through Outlier's evaluation and feedback loop, your work helps turn frontier research into AI that performs better in the real world. Responsibilities: * Holding a live, scenario-based conversation in English with AI voice models, then ranking how natural, useful, and clear each one sounded * Comparing pairs of AI-generated images and videos and judging which one better follows the prompt * Inspecting outputs closely, including zooming in and replaying, to spot problems like distorted anatomy, flickering or unnatural motion, text errors, and audio that doesn't match the video * Rating each output across dimensions like adherence, quality, and realism, then writing a specific explanation in English that points to what you saw or heard Common tasks: * Image evaluations: Compare two AI-generated images for the same text prompt and pick the better one, judging how well each follows the instructions, visual quality, and visible AI artifacts. Some tasks rank edited images against the original. * Video evaluations: Rank AI-generated videos against a text prompt, or edited videos against their source clip. * Video caption labeling: Label first-person videos of people or robots doing hands-on tasks. Contributors split each clip into individual actions and write precise captions, starting from an AI draft and correcting it. * Audio and voice evaluations: Rate AI-generated speech on naturalness, pronunciation, and how closely it matches a target voice. Some tasks require recording a short voice explanation of the choice. Ideal qualifications: * Fluent English speaker with strong, clear written communication and comfortable with natural, everyday conversational speech * Strong attention to detail, with the patience to look and listen closely, and to replay or zoom in when something seems off * Comfortable thinking on your feet and talking through realistic situations in a natural, conversational way * Experience writing short, evidence-based explanations or reviews * Comfortable making a clear decision when both options have flaws * Nice to have: Pursuing or holding a Bachelor's degree or above What you'll need to sign up: * A valid ID and mobile phone from your country of residence * An up-to-date resume that highlights your expertise * A LinkedIn profile showing your educational background and work experience