ML in society presentations

Misconduct in police force

• Connections:
  • Clare Baldwin’s work in the Philippines: using ML to anaylze large public-ish datasets
  • Consent decree research in Oakland (voice recordings)
• Theoretical framing: What’s the problem? (New Jim Crow?)

Discouraging people from supporting racism

• In the “High Level Problem” slide, I like the theorizing I’m seeing, but want to encourage more inquiry here. For example, it’s contradictory to say that racism is an excuse for economic exploitation, but then also to say that people hold racist ideologies for the greater good.
  • Jane Hill “The Everyday Language of White Supremacy”
• Also need thorization of how social media interacts with race.
  • Ruha Benjamin, Race After Technology.
  • Andre Brock, Jr. Distributed Blackness
• I like the connection between the problem and approaches such as ML, Social Network Analysis.

Disparities in education

• Read Anna Blatto’s history of segregation in Buffalo.
  • Make sure you are not positioning segregation as “traditional” or “natural” in an ahistorical way.
  • It might make sense to zoom in on a specific city where you can engage with the specific history and context of school segretation.
• Data. Make sure it means what you think it means.

White supremacy?

• High-level problem is too broad in scope
• Raciolinguistic ideologies, theories of organizational behavior (or other research from business schools) could provide some theory with explanatory power for how people end up in decision-making roles.
• Once you have committed to a conceptual model of how inequity is produced, you may be able to make clever use of existing data sets.

The othering

• “The promise of access” Daniel Greene
• “Learning in schools is omitting a diverse history” Center for Black Racial Literacy
• Economics of education perspectives needed–you won’t understand funding without policy.

How do textbooks enforce white supremacy

• LaGarrett King’s analysis of textbooks
• Erasure
• Modeling latent racial ideologies
  • Embeddings?
  • Topic modeling?
  • Autoencoder?
• Diachronic word embeddings
• AJ Alvaro

Judicial system

• Data availability will be a key question here

Ending modern radicalization online

• How does radicalization happen? As an individual phenomenon and as a social phenomenon?
• Model the meaning of language
  • Jane Hill “The Everyday Language of White Supremacy”
  • Langauge ideology, indexicality
  • Speech genres
• Model identity: Is it a story we tell ourselves? To what extent is identity stable across context and to what extent is it
• Time series modeling of trajectories has a lot of potential

Higher education attainment

• First-generation college students: Talk to folks in DEE.
• Pay attention to how ML reifies categories.

Race and caste in India

• In general, very strong theoretical framing
  • Postcolonial theory needed?
  • White gaze
• Talk with Sameer Honwad