AI summary by TwinMind
How racial bias works -- and how to disrupt it | Jennifer L. Eberhardt
TED
14m 18s
Personal Anecdote: 5-Year-Old Son's Racial Bias
- Airplane incident with 5-year-old son
- Son identified Black man as looking like daddy (husband)
- Man looked nothing like husband - was only Black person on plane
- Son expressed fear man would "rob the plane"
- Child couldn't explain why he thought this
- Demonstrates how racial stratification affects even young children
- Association between blackness and crime penetrates minds without explicit hatred
Research Findings: Bias Effects on Perception and Behavior
- Lab experiments exposing people to faces
- Exposure to Black faces → increased clarity/speed seeing blurry gun images
- Bias controls what we see and where we look
- Violent crime prompts → direct eyes to Black faces, away from white faces
- Police officers thinking about capturing/shooting/arresting → eyes settle on Black faces
Criminal Justice System Bias
- Large dataset of death-eligible defendants
- "Looking more Black" more than doubled chances of death sentence
- Effect significant when victims were white
- Controlled for crime severity and defendant attractiveness
- Black people punished proportional to blackness of physical features
- "More black = more death worthy"
Educational System Bias
- Teacher discipline studies with colleagues
- Teachers desire to discipline Black middle school students more harshly than white students for same repeated infractions
- Recent study findings:
- Teachers treat Black students as group, white students as individuals
- If one Black student misbehaves, then different Black student misbehaves days later
- Teacher responds to second student as if he misbehaved twice
- "Sins of one child get piled onto the other"
Categorization and Bias Mechanisms
- Brain creates categories for world comprehension
- Categorization allows quick, efficient judgments
- Relies on predictable patterns
- Categories help quick decisions but reinforce bias
- Same mechanisms that help us see world can blind us
- Render choices effortless/friction-free but exact heavy toll
Nextdoor Case Study: Adding Friction to Reduce Bias
- Tech company purpose: stronger, healthier, safer neighborhoods
- Online platform for neighbor information sharing
- Problem: racial profiling
- People see Black man in white neighborhood → assume criminal activity
- No evidence of wrongdoing required
- Online behavior reflects real-world behavior
- Risk: amplifying bias instead of dismantling disparities
Nextdoor Solution: Three-Item Checklist
- Co-founder consulted speaker and two others
- Decision to add friction (slow people down)
- Checklist items:
- "What was this person doing that made him suspicious?" - "Black man" not grounds for suspicion
- Describe physical features, not just race/gender
- Provided racial profiling definition - strictly prohibited
- Modified "If you see something, say something" → "If you see something suspicious, say something specific"
- Result: 75% reduction in racial profiling
Oakland Police Department Intervention
- Collaboration with speaker and colleagues
- Goal: reduce stops of people not committing serious crimes
- Added question to stop forms: "Is this stop intelligence-led? Yes/No"
- "Do I have prior information to tie this person to specific crime?"
- Forces officers to pause, think before stopping
Oakland Results
- 2017 (before intervention): ~32,000 stops citywide
- Following year (with question): 19,000 stops
- African-American stops fell 43%
- Crime rate continued falling - city became safer for everyone
- Stopping fewer Black people didn't increase danger
Technology Solutions: Body-Worn Cameras
- George Floyd death recorded by cell phone cameras
- Police departments required to wear body cameras
- Recordings capture extreme encounters and everyday interactions
- Stanford interdisciplinary team using machine learning
- Analyzed large numbers of traffic stop encounters
Stanford Findings
- Even professionally behaving officers speak less respectfully to Black drivers than white drivers
- Could predict driver race from officer word choice alone
- Problem: vast majority of footage unused by departments
- Not used for understanding street interactions or officer training
Personal Story: Eldest Son's Elevator Experience
- Son age 16 discovered white people feel fear when looking at him
- Elevators worst - trapped in tiny space with someone associated with danger
- Son's coping mechanism: smile to put others at ease
- When he speaks: bodies relax, breathe easier, enjoy his cadence/diction/word choice
- "Sounds like one of them"
- Initially thought son was natural extrovert like father
- Realized smile wasn't connection desire - was protective talisman
- Survival skill honed over thousands of elevator rides
- Learning to accommodate tension his skin color generated
- Puts his own life at risk
Call to Action: Interrupting Bias
- Brain wired for bias
- Interrupt bias by pausing, reflecting on evidence of assumptions
- Questions to ask:
- What assumptions do we bring to elevators/airplanes?
- How do we become aware of unconscious bias?
- Who do assumptions keep safe?
- Who do they put at risk?
- Must insist schools, courts, police departments, every institution ask same questions
- Without this: bias continues to blind us
- "If we do none of us are truly safe"
