Emotional Trauma and Machine Learning
summary
Work combining IBM Watson chatbot design, five years on Wikimedia’s anti-harassment team and interviews with over 40 Gamergate victims leads to a concrete toolset: a scraped corpus of alt‑right subreddits and a hate speech database used by ProPublica, experiments with Google’s Perspective API showing ~30% relevance on coded phrases, and a deployed design called ARMS that pairs moderators with human-guided supervised ML to learn intent early and let communities set user-generated safety definitions.
AI-written summary · about
chapters (7)
- 0:00Start
- 8:17The Living Room
- 10:45Keyword Blocking
- 13:34Gamergate
- 16:54Alt-Right Subreddit
- 17:07White Nationalism
- 19:07Hate Speech Database
related talks
46mAI That Reads Emotions: What It Means for Data Privacy, Consent, Your Teams
45mEthical use of AI: seeking, understanding, and countering the risks
16mFighting Harmful Content with AI: Real-World Safety
11mMachines of Love & Grace – Building Truly Inclusive AI
23mUser Experience Now Includes Machine Experience
36mCan AI Help Moderate In-game Chat?
28mUser Research GenAI Better Together
35mAI ethics for tech writers