TED Talks Daily How to keep human biases out of AI | Kriti Sharma
Mar 15, 2019
Kriti Sharma, a technologist dedicated to ethical AI, sheds light on the pressing issue of human biases embedded in algorithms. She discusses the far-reaching impacts of bias on decision-making, especially regarding gender and race. Kriti emphasizes the crucial role of diversity in tech teams, sharing her experiences to illustrate how varied perspectives can lead to fairer solutions. She also proposes actionable strategies to create more equitable AI systems, aiming to harness technology for social good and better serve marginalized communities.
AI Snips
Chapters
Transcript
Episode notes
AI's Real Danger
- The biggest risk with AI is not robots taking over.
- The real danger is AI making biased decisions based on gender, race, and background.
Targeted Ads Reveal Bias
- AI's biases are evident in targeted advertising, like pregnancy test ads or fertility clinic ads.
- Kriti Sharma even received ads for Indian marriage bureaus due to AI assumptions.
Examples of AI Bias
- AI has made biased decisions, such as deeming a black person less likely to repay a loan.
- It has also concluded that a person named John is a better programmer than a person named Mary.