Remote AI Voice & Media Quality Evaluator

Scale AI; Richmond Hill; Remote; CAD 24,000 - 34,000; Full time

Benefits offered by this job

Flexible schedule

Free access to leading LLMs

Global community

Job summary

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 model performance.

We are seeking detail-oriented English speakers to review AI-generated media, roleplay live conversations with an AI voice model, and score exchanges on accuracy and naturalness. You’ll compare AI-produced images and videos, inspect outputs for issues like distortions or audio mismatches, and write concise explanations in

Qualifications

• Fluent English with strong written communication.
• Excellent attention to detail and ability to replay or zoom to spot issues.
• Experience writing concise, evidence-based explanations or reviews.
• Ability to make clear decisions when options have flaws.

Responsibilities

• Holding live, scenario-based conversations in English with AI voice models and ranking naturalness, usefulness, and clarity.
• Comparing pairs of AI-generated images or videos and judging which follows the prompt better.
• Inspecting outputs closely to spot issues like distorted anatomy, flickering, text errors, or audio-video mismatches.
• Writing a concise explanation in English that points to what was observed or heard.

Skills

Fluent English

Attention to detail

Clear written communication

Evidence-based writing

Decision making

Education

Bachelor’s degree or higher

Job description

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 model performance.

We are seeking detail-oriented English speakers to review AI-generated media, roleplay live conversations with an AI voice model, and score exchanges on accuracy and naturalness. You’ll compare AI-produced images and videos, inspect outputs for issues like distortions or audio mismatches, and write concise explanations in