interviewed at the Artificial Intelligence Conference, London 2018 – Susan Etlinger
summary
Argues that current ML practice creates real customer harms unless fixed: supervised learning promotes the status quo illustrated by hiring models that train on tenure, promotions and pay and therefore hiring model amplifies historical bias; geospatial optimization for same‑day delivery produces same-day delivery "delivery deserts" tied to ZIP codes; medical image sets lack darker‑skin examples so skin‑cancer detectors underperform; practical mitigations include algorithmic impact assessments and checklists, bias‑testing toolkits and cross‑functional governance.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 2:20Susans tweet thread
- 4:34Use cases
- 6:08The supply chain
- 8:19Supervised machine learning
- 10:17Ethics
- 15:40Principles
- 19:51Fullstack
- 21:40Risk
- 25:14Adoption
related talks
26mBuilding ethical AI: Transparency, bias mitigation, and trust
05mToward Ethical AI
19mMoral Outsourcing: Humanity in the age of AI
30mInside AI: What Every Professional Needs to Know to Stay Relevant, Protected, and Ahead
06mInterview with Luis Seco
09mInterview with Vladan Joler
39mKeynote Panel State of AI in the Enterprise
43mUX and the Changing Relationship Between Mind and Machine