(Un)Ethical Machines? Why AI Bias Is a Problem and What to Do About It
summary
Examples where models deny women access to credit and label dark‑skinned people as gorillas anchor the claim that bias comes from training data and historical injustice, not from sentient machines. Concrete causes include skewed datasets (celebrity images overrepresented), legacy hiring patterns (resume filters reproducing male‑dominated roles), low diversity in AI teams (22% women in AI), and regulatory/privacy tensions that prevent checking protected groups (sensitive attributes blocked by privacy law). Recommended fixes are audits, education, data diversification, and new rules such as the EU AI Act amid ~130 recent AI laws.
AI-written summary · about
chapters (15)
- 0:00AI bias introduction
- 2:48Why AI bias attention now
- 5:57AI bias human creation
- 7:01Bias sources in AI systems
- 9:33Historic bias in AI teams
- 12:44Fairness is subjective and requires diverse perspectives
- 14:32AI regulations limits
- 16:40Responsible AI beyond tech
- 18:25Ethical AI training
- 20:13Open data for inclusive AI
- 22:40Data diversity for inclusion
- 26:36Who fixes AI bias
- 28:41Human behavior shapes AI
- 32:52Fix human bias first
- 33:38Q&A
related talks
56mAn AI with an Agenda: How Our Cognitive Biases Leak Into Machine Learning
21mFairness and bias in AI
47mEthics & AI: Identifying & preventing bias in predictive models
05mToward Ethical AI
35mZero Bias AI
23mMachine Learning Ethics
38mSocial Implications of Bias in Machine Learning
11mMachines of Love & Grace – Building Truly Inclusive AI