old and recovered reading list! somewhat updated:2021-06
An odd but worthy reading list
I have been following Cathy O’Neil since around 2015 when I was introduced to her blog by a colleague at the time.
Her talk at the Women in Analytics Conference was really inspiring. I about this in twitter but here are my highlights.
- People trust AI because people trust math - she wrote Weapons of Math Destruction because she thought: “what would it look like to put the science into data science?”
- Math, AI and algorithms is not something we need to intimidate people with
- In models, we have to define what success looks like - people who are in power get to define it, we have white guys hiding behind the authority of mathematics, algorithms of math destruction are widespread, mysterious, destructive
- “These algorithms don’t just fail for the people they harm, they fail their underlying mission”
- Algorithms cause damage when they are used to punish - they could be used as a way to empower people, to help people, but they are instead being used negatively– use them to help with poverty, with health…
- Just because the algorithm is less accurate does not mean its less fair - fairness has to do with the way it is used.
- The people who are affected want to know: “are my civil rights being violated?”
A more thorough and formal blog about my takeaways is here -> Ethics in algorithms and other key take aways from the 2019 Women in Analytics Conference USA
