Is ChatGPT (and Other AI) the Enemy of Diversity?
Talkat Open Source Summit Europe 2023 · on Tuesday, 2023-09-19· at 12:20· Room 3G (3JAUREGIA - Floor 3), Abandoibarra Etorb., 4, 48011 Bilbao, Bizkaia, Spain·
summary
Claims and evidence center on real numbers and real failures: large language models hit 100 million users in two months, GPT‑4 shows an 82% reduction in disallowed outputs after tougher RLHF rules, the industry is 22% women, Kenyan contractors are paid cents to label trauma content, robotic police dogs and a Detroit misidentification case show deployment harms, and examples that work include lung‑cancer detection, flood prediction and Be My Eyes; reinforcement learning from human feedback and data curation are the concrete levers proposed.
AI-written summary · about
chapters (15)
- 0:00Welcome
- 3:49Why this talk matters
- 7:00AI 101 basics
- 11:12Bias from history to AI
- 14:41Bias in AI training
- 17:27AI ethics gone wrong
- 21:23Global AI threats
- 23:07AI empowering communities
- 27:19AI for sustainability and health
- 30:48AI regulations and governance
- 33:39Ethical AI proposals
- 36:08Poem on AI and ethics
- 39:44Q&A
- 43:55Q&A older generations
- 47:49Q&A removing personal data
related talks
01h 05mIs ChatGPT (and Other AI) the Enemy of Diversity?
20mHuman-Centered AI for Sustainable Development
41mThe AI boom: what and why?
29mThe Rise of AI: A Reality Check on Generative AI
29mUnintended consequences of technology: Enabling NextGen discrimination
52mSome pitfalls in AI
50mPanel: Humanity in the Loop: Balancing AI, Data Science, and Human Ingenuity in the Cybersecurity..
41mGhost in the Machine: The Unexpected Results of Not Understanding Your Users