GitHub - daviddao/awful-ai: Awful AI is a curated list to track current scary usages of AI - hoping to raise awareness
ai-ethicsai-biassurveillancedisinformationharmful-ai
Abstraction: Curated list cataloguing harmful and unethical real-world AI deployments
Key points:
- Maintained by David Dao; covers discrimination/bias (COMPAS, Amazon recruiting, Google dermatology app), surveillance (Clearview AI, SenseTime/Megvii, WeChat censorship), disinformation (deepfakes, fake news bots, Cambridge Analytica), and lethal autonomous weapons
- Annual "Awful AI Award": 2024 AI companion chatbots (linked to teen suicide), 2023 Israel's AI targeting in Gaza, 2022 commercial image generators stealing artists' work, 2020 Google firing Timnit Gebru
- Documents LLM harms including ChatGPT biases, Bing "Sydney" disturbing behavior, and Meta's Galactica failure
- Also covers environmental harms: training large models generates ~626,000 lbs CO2; oil companies using AI to boost production 5%
- Includes counter-section of privacy-preserving and fairness research (differential privacy, adversarial ML, model cards)
Connections: Google · Openai · Microsoft · Clearview AI · AI Ethics · AI Bias · Algorithmic Fairness