https://www.nytimes.com/2020/06/24/technology/facial-recognition-arrest.html
In 2020, Robert Julian-Borchak Williams was the first American to be wrongfully arrested based
on a flawed match from a facial recognition algorithm. Facial recognition technology has been
used by law enforcement to identify suspects for nearly two decades. The technology works well
for white men, but not very well for other demographics. Williams is not white.
Read the newspaper article carefully.
Your task (5-6 page paper, 12-point font, double-spaced):
-Descriptively Diagnose. What has likely happened? Why is this software so bad at
generating accurate matches for minorities? Where could the bias have crept into this
facial recognition software? Where might bias be perpetuated? Use the machine learning
loop to do this analysis. Explain to an audience that has no background in machine
learning, data science, or computer science.
-Normatively Diagnose (a). Identify and explain what kind of discrimination is at play
here. There can be more than one. You are welcome to just pick one and go into depth or
offer multiple. Either way, be sure to justify your diagnosis and explain in detail.
-Normatively Diagnose (b). What makes this algorithm morally wrongful or a concern
from the perspective of justice? Provide careful and thorough reasoning for why someone
might think the use of this algorithm (used by law enforcement to identify and arrest
suspects) is wrongful. Here you are building a case that the use of the algorithm is
wrongful (even if you ultimately disagree or have concerns).
-Implications. Building on or responding to the analysis above, explain what you think
the upshot should be. Make it clear how your normative diagnosis leads you here or how
your thoughts are a response to your analysis.
-Counterargument AND Response. What might someone say in response to your
position? Respond. This need not come at the end of the paper, but within the paper, you
need to need to entertain and respond to a dissenting view.
Include an introductory paragraph (or two). In your introduction: set up, motivate, and
clearly state the problem; state your thesis; tell the reader what you plan to do in the
paper.
• Explain all your terms when you first introduce them (e.g. indirect discrimination).
• Provide reasoning for any claims or analysis.
• Be explicit.
• Be concise and clear. Only include information that is important for the flow of the
argument (setting the stage, motivating the problem, explaining the problem, articulating
your thesis, showing your thesis resolves or clarifies the problem, etc.)
• Include a concluding paragraph that summarizes your analysis.
