Module 4 -> Lesson 1
Active listening, collaborative discussion, learn through writing
Anchor Text(s) for this Lesson
Supporting Text(s)/ Resources for this Lesson
Sample slide deck: Module 4 --> Lesson 1
Lesson Overview
In this lesson, students learn about other forms of bias that might shape the impacts of facial recognition technology, with a particular focus on its application in policing. Students learn about confirmation bias--through an experiential learning hook--and learn about automation and authority biases. Students closely watch a video that is an oral history of Robert Williams' experience after FRT deployed by Detroit police falsely identified him as a suspect. Students consider how various forms of bias were potentially at play in this scenario and continue to weigh the risks and benefits of this technology in the context of policing.
Nota Bene
The activity in today's lesson includes an 8-minute video with an activity-guide that asks students to evaluate information presented in the video, so they can consider the influence of different forms of bias on the reliability of FRT and analyze the impacts of FRT used in policing. Depending on your setting, you might want to collectively watch the video and pause strategically or invite students to watch in pairs or independently with headphones. Too, you will likely need to adjust the lesson to suit the pace that your particular setting allows.
Objectives
Students will be able to...
analyze the impacts of FRT used in policing by evaluating events described in an oral history
consider the influence of different forms of bias on the reliability of FRT.
Suggested Duration
45 minutes (adjust according to your students' needs)
NYS Next Generation ELA Standards
W1c: Use precise language and content-specific vocabulary to express the appropriate complexity of the topic.
W1:Write arguments to support claims that analyze substantive topics or texts, using valid reasoning and relevant and sufficient evidence.
RH9: Compare and contrast treatments of the same topic in several primary and secondary sources.
RH7: Integrate and evaluate visual and technical information (e.g., in research data, charts, graphs, photographs, videos or maps) with other information in print and digital texts.
NYS Computer Science & Digital Fluency Standards
9-12.IC.5 Describe ways that complex computer systems can be designed for inclusivity and to mitigate unintended consequences.
9-12.IC.3 Debate issues of ethics related to real world computing technologies.
9-12.IC.1 Evaluate the impact of computing technologies on equity, access, and influence in a global society.
Vocabulary
algorithmic bias: systematic and repeatable errors in a computer system that create unfair outcomes, such as privileging one arbitrary group of users over others; example: facial recognition algorithms trained with a data set that does not reflect human diversity and therefore results in different reliability based on gender and race.
cognitive bias: a thought process caused by the tendency of the human brain to simplify information processing through a filter of personal preferences and opinions; examples: confirmation bias; automation bias; authority bias; gender bias; group attribution error
automation bias: to trust information provided by a machine/ computer and ignore information that contradicts it; automation bias is one type of cognitive bias. example: trusting the output of FRT when it produces a “match” without verifying that match by other means.
confirmation bias: the conscious or subconscious tendency for people to seek out information that confirms their pre-existing viewpoints, and to ignore information that goes against them (regardless of whether that information is true or false); example: only paying attention to news stories that confirm your opinion and ignoring or dismissing information that challenges your position.
authority bias: tendency to trust / follow influence of a leader/ person in position of authority
false positive: when the face recognition system does match a person’s face to an image in a database, but that match is actually incorrect
false negative: when the face recognition system fails to match a person’s face to an image that is, in fact, contained in a database
Hook
Choose (or create your own) experiential learning activity to demonstrate confirmation bias for students.
OPTION 1: Ask the class to weigh in on their perspective on astrology. Who 'believes in' astrology? Who does NOT believe in astrology? Then hand deliver a 'secret horoscope' to each student. Let students know that you found their horoscopes according to their birthday data in the school's register. [All 'horoscopes' will be the same: here's one you might use.]
Give students the following prompt:
Read your secret horoscope! Make sure you do not show it to anyone. Do you think this horoscope is accurate? Give examples from your life and your personality to support your answer.
Ask one student to share their responses to the questions; ask a few more volunteers to share their answers--be sure to elicit answers from both groups of students: those who identify as believers and non-believers in astrology. Then take a step back and ask the first student to read aloud their horoscope. If necessary, ask another student to read theirs--until the class notices that they each have identical horoscopes. Then present the definition of confirmation bias and help students connect to other examples of confirmation bias they may have experienced/ witnessed in their lives.
OPTION 2: Use the "Experience Confirmation Bias" strategy included in the Facing History and Ourselves lesson demonstrating the Wason Rule Discovery Test. You can do this with the Can You Solve This video or using an interactive version of the test on the NYTimes. Find complete directions on the Facing History and Ourselves website; scroll down to the Extension Activities (link)
Mini-Lesson
Present the definition of cognitive bias and let students know that confirmation bias is one example of bias that falls into this category. Briefly review the definition of algorithmic bias and elicit from students examples they learned about in the last module and/ or examples they have witnessed in their lives. Then present the definition of automation bias and ask students to think about how algorithmic bias and automation bias might interact.
Activity
Inform students that they are going to watch/ listen to a first-person oral history of Robert Williams experience in Detroit where police used FRT to search for a suspect in an armed robbery. As they closely watch and listen to the video, they will consider the ways in which different forms of bias influenced the reliability of the technology. Provide students with the activity guide that includes the vocabulary listed above (much of which they've seen/ discussed during the hook and mini-lesson) and the following video-based questions:
How, if at all, did algorithmic bias play a part in Robert Williams arrest in Detroit?
Do you think automation bias influenced the events leading to Robert Williams arrest? Explain your reasoning.
Do you see any evidence of confirmation bias in this scenario? Explain your reasoning.
Did Detroit police department’s FRT report a “false positive” or a “false negative” when it identified Robert Williams as the person recorded by the surveillance camera? Explain your thinking?
As a society, would you say all people face the same potential risks and benefits when FRT is used in policing? Explain your reasoning.
EXTENSION: Create a program that would provide police officers with training that would aim to reduce the influence of different forms of bias on their application of FRT.
Wrap Up
Check in with students to amplify their analysis of the impacts of FRT in policing in the context depicted in the video. Elicit their ideas about how biases influenced the reliability of the technology. Ask students if scenarios like this are overshadowed by those in which serious crimes are solved or prevented. Let them know that in the next lesson, they will evaluate the rules the NYPD has outlined for its use of FRT alongside another first-person oral history.
Last updated