Coded Bias, AI, and the Future of Civil Rights

Written By Dylan & Jess

All Tech is Human Series #6: with Shalini Kantayya, Meredith Broussard, and Timnit Gebru

Details from the original livestream event put on by All Tech is Human

Details from the original livestream event put on by All Tech is Human

Action Items & Takeaways

Considerations on activism in this space:

  1. Check out the resources below mentioned by the guests! Get informed and educated!

  2. Demand that algorithmic ethics classes are required for anyone who is getting a CS degree

    1. These demands can come from faculty and also from undergraduate or graduate students!

  3. People assume that including “ethics” means adding philosophy to these classes.. Which can be harmful

    1. Don’t just include units on white male philosophy

    2. Instead focus on disability studies, feminism studies, critical race theory, gender studies, etc…

  4. Sign the declaration of data rights on the coded bias website (link below)

Considerations for influencing legislation on these topics:

  1. Start with local legislation, then it can help tip the federal legislation

    1. E.g., SF banning facial recognition, now it is going for the state ban in some places, maybe eventually federal ban

    2. Local legislation can be harder, but you can actually achieve something and actually measure what you’ve achieved

  2. Legislation creates culture shift, if you can just start to pass local legislation and bans, even at the university level (e.g., banning facial recognition on campuses), it sets a new norm and a precedent for the culture of that place

  3. Stop allowing tech companies to regulate themselves (Meredith)

    1. People who are not in position to make billions of dollars from the company should write the regulation for that company, not those who work on the inside already

    2. Motivate these companies to be more ethical through business / PR pressure: e.g., if the company doesn’t make a general purpose facial recognition or gender classification software, they won’t have to redact it and waste their money like other organizations have done already

Responsible tech needs bridge builders, consider becoming one:

  1. People who can translate the technical jargon to activists who can utilize it to make legislative change

  2. People who can translate the social sciences to the technical in industry

  3. People who can speak to both sides in their jargon, language, and cultural context / norms to make it happen!

Combat technochauvinism (the idea that tech is always the best solution):

  1. Instead ask what is the right tool for the task? It might be computational, it might not be. 

  2. Recognize that when technological tools are created for solutions, they are often not created by the people who actually need the solution and are not made equally

  3. Make more diverse teams creating technology, give more seats at the table to those who have historically not been welcomed

    1. There needs to be a structural shift (Timnit). It’s not just about inviting diverse teams to create products (Women in power can still be sexist, people of color in power can still be racist), instead we have to create structural and cultural shifts to fix these problems as a community, rather than relying on “diversity” to fix it for us.

  4. Talk to the stakeholders (not just consumers, but also the people who are evaluated by automated systems) before designing the technology to get their input!

Do you have any more action items or takeaways that you’d like to share that are related to this topic? Any resources you’d like to include in this list? We’d love to hear from you! Feel free to leave your thoughts in a comment below (you can post your comment by hovering over the bottom-right corner of the comment box).

Dylan & Jess
Previous

The Business Case for AI Ethics
Next

Social Media's Role in the US Election