Uploaded February 2021 | Updated September 2026, 1 week ago
Biometrics are a critical tool in information security and of particular interest to cybercriminals but the biggest issues remain with how we train machine learning model and how we get that data.
A fascinating study by Raji & Fried points out that the way these data sets have been gathered has become increasingly problematic. Gone are the days of explicit consent to participate as data is simple vacuumed up from the internet.
The consent issue is huge but so are the bias and discrimination issues introduced with these methods.
Coverage of the research and inspiration for this short via Karen Hao writing for the MIT Technology Review, technologyreview.com/2021/02/05/1017388/ai-deep-learning-facial-recognition-data-history
Index:
- The question [0:00]
- Raji & Fried's study [0:08]
- Bias & discrimination in data [0:23]
- Call to action [0:35]
#shorts
Biometrics are a critical tool in information security and of particular interest to cybercriminals but the biggest issues remain with how we train machine learning model and how we get that data.
A fascinating study by Raji & Fried points out that the way these data sets have been gathered has become increasingly problematic. Gone are the days of explicit consent to participate as data is simple vacuumed up from the internet.
The consent issue is huge but so are the bias and discrimination issues introduced with these methods.
Coverage of the research and inspiration for this short via Karen Hao writing for the MIT Technology Review, technologyreview.com/2021/02/05/1017388/ai-deep-learning-facial-recognition-data-history
Index:
- The question [0:00]
- Raji & Fried's study [0:08]
- Bias & discrimination in data [0:23]
- Call to action [0:35]
#shorts
