This talk is part of a two-presentation session running from 1:30 PM - 2:20 PM. The session will feature two presentations back-to-back, with Q&A after each presentation.

Videos are easily assimilated, causing a surge in volume on social media platforms in the past few years. Content moderators may be subjected to disturbing content for hours, and have to watch the entire video to identify, if any, the few policy violating frames. Under tight time constraints, this can lead to high cognitive load, leaving room for error and bias when making non quantifiable decisions. In this presentation we propose a reusable way of extending AI processors to assist manual video reviews. Content labels are suggested based on the images, audio and text in the video. The spam/low-quality score, confidence score and AI model features are displayed on the review tool, allowing reviewers to trust and question the AI scores, thereby improving the human-AI performance. Aware of AI presence, reviewers become cognizant of any biased judgment and strive to mitigate them by taking the information perspective.