I was kindly asked by the PyData Amsterdam organizers to keynote the conference. As a passionate fan of ethical machine learning and the great research being done by data scientists and academics around the world -- I am very enthused to present the topic to the conference.
My slides are currently available as a jupyter notebook via GitHub and I will be posting them in an easy way to key through them soon. I will be adding the video as well as several extra posts regarding the research and findings here.
Talk and Slide References
- Minority Areas Pay Higher Car Insurance than White Areas with the Same Risk by ProPublica
- Ethics Can't be a Side Hustle by Mike Monteiro
- Stereotyping and Bias in the Flickr30 Dataset by Emiel van Miltenburg
- Why Facebook is giving out free Wi-Fi for check-ins by CNet
- Do not untick this box if you do not want to receive updates by Formismo
- Say hi to your new boss: How algorithms might soon control our lives and GitHub Repo by Andreas Dewes
- Semantics derived automatically from language
corpora necessarily contain human biases by Aylin Caliskan-Islam, Joanna J. Bryson, and Arvind Narayanan (Related 33c3 talk by Aylin Caliskan-Islam: Story of discrimination and unfairness)
- Machine Bias by ProPublica
- Ethics for powerful algorithms by Abe Gong
- Certifying and removing disparate impact by Michael Feldman, Sorelle A. Friedler, John Moeller, Carlos Scheidegger, Suresh Venkatasubramanian
- FairTest: Discovering Unwarranted Associations in Data-Driven Applications and Github Repository by Florian Tramèr, Vaggelis Atlidakis, Roxana Geambasu, Daniel Hsu, Jean-Pierre Hubaux, Mathias Humbert, Ari Juels, Huang Lin
- When Recommendation Systems Go Bad by Evan Estola
- Equality of Opportunity in Supervised Learning with interactive data visualization with generated loan data by Moritz Hardt, Eric Price, Nathan Srebro (interactive by Google BigPicture)
- Ideas on Interpreting Machine Learning by Patrick Hall, Wen Phan and SriSatish Ambati
- Financial Modeler's Manifesto by Emanuel Derman and Paul Wilmott
Recommended Reading & Related Work
In addition to the papers I was able to reference in the slides, I have appended here some recommended reading on the topic of Ethics in Machine Learning. Expect this list to expand over time :)
Blogs and Publications
- UnBias: Emancipating Users Against Algorithmic Biases for a Trusted Digital Economy
- Joanna J Bryson's Blog (and entire CV). You can also [follow her on Twitter].(https://twitter.com/j2bryson)
- Hal Daumé III's NLP Blog
- Algorithmic Fairness by Suresh Venkat
- Arvind Narayanan's work
- I also wrote a post about embedded racism and sexism in word vectors.
- A Guide to Solving Social Problems with Machine Learning
- The Ethics of Artificial Intelligence
- Predicting Recidivism Risk: New Tool in Philadelphia Shows Great Promise
News on Ethical Machine Learning and Models Gone Bad
- Michigan unemployment agency made 20,000 false fraud accusations
- Facebook, Twitter, and Instagram surveillance tool was used to arrest Baltimore protesters
- Many Cars Tone Deaf To Women's Voices
- Uber seems to offer better service in areas with more white people. That raises some tough questions.
- A beauty contest was judged by AI and the robots didn't like dark skin
- Amazon Doesn’t Consider the Race of Its Customers. Should It?
- On The Ethical Use Of Data Vs. The Internet Of Things
- How Subtle Class Cues can Backfire on Your Resumé
- Can Artificial Intelligence Wipe Unconscious Bias From Your Workday?
- On Computational Ethics
- AI learned to betray others. Here's why that's okay
- Errant code? Not just a bug.
- Artificial Intelligence Will Be as Biased and Prejudiced as Its Human Creators
- If We Don’t Want AI to Be Evil, We Should Teach It to Read
- How LinkedIn’s search engine may reflect a gender bias by Matt Day, Seattle Times (and follow up by Samanta Cooney: LinkedIn Tweaks Search Algorithm After Report Suggests Gender Bias)