Artificial Intelligence & Race: Does AI have issues with racial bias?

My lecture/case study; was based on me sending the wrong man to prison for a crime he didn't commit, using Facial Recognition software.

I began to conduct research and submitted questionnaires and interviews throughout the technology community. My findings lead me to this presentation.


Outline/Structure of the Talk

The proposal/lecture/case study; was based on me sending the wrong man to prison for a crime he did not commit, using Facial Recognition software. AI facial recognition, some of the most prominent being these programs fail to recognize dark-skinned faces and dark skin in general, over fifty percent of the time. 50% is unacceptable and will only serve to further systematic oppression and racism, which already plagues people of color around the world.

From having our speed and traffic violations recorded at every stoplight via a computer-generated AI with facial recognition software to systems that immediately analyze a criminal’s fingerprints, it impacts us in many ways we do not even notice. Artificial Intelligence is on our streets, in our grocery stores, and even in our homes. It is an issue we need to address for a variety of reasons; one of the most primary being some AI does not work if an individual is black.

We are not yet at a point wherein artificially intelligent robots are after human blood; this is good but does not mean it will ever occur. While these AI processes may streamline many procedures and increase efficiency regarding certain aspects of modem society the fact remains if it is used to make life-changing decisions for us a disparity between appropriately recognize the color of the subject’s skin is paramount and akin to further propagating the already vast racial disparities experienced around the world.

•My research indicates that the latest facial recognition software is 99% effective if the subject is white, but 80% or less effective when recognizing black faces, specifically black women. My research found issues rooted within databases used to provide a basis for facial recognition.

•Until such a time that Artificial Intelligence is amended to include and recognize all races, it will remain a facet of institutionalized racism, which continues to threaten the livelihood of the global majority.

Learning Outcome

Key Takeaways:

**Implement more training for developers to reduce overt acts of bias.

**The future of facial recognition and ethnic partiality.

**The Consequences of inaccurate data.

**Public data vs. real-life data.

**Adaptation for Facial Recognition to lessen the unintentional acts of preconceptions.

**Distinctiveness intelligence to comprehend & resolve.

Target Audience

Anyone with an interest in AI Ethics & Integrity.

Prerequisites for Attendees

Have a general understanding of the issues surrounding race and ethics concerning AI


schedule Submitted 1 year ago

Public Feedback