Data Ethics, Part II

Data Literacy

Neutrality Thesis

  • Why does the neutrality thesis continue?
  • Subject vs Object.

Technological Mediation

  • Technology mediates our decisions.
  • The form Shapes the behavior.

Delegation

  • Technology that delegates decision-making.
  • Example: speed bump.
  • Data Structures.

Rubric

  • PFAT:
    • Privacy
    • Fairness
    • Accountability
    • Transparency

Privacy

  • Consent
  • Personally Identifiable Information (PII)
  • Anonymization
  • Data Minimization
  • We don’t know what we collect today will be used for tomorrow.
  • General definition: “not being observed or disturbed by other people.”
  • Freedom from arbitrary disturbance. (consent is key).
  • Information not shared without consent.
  • Anonymizing data.
  • Using data in unauthorized ways.
  • Knowledge of how data will be used.
  • Privacy is connected to free will or autonomy.
  • Why?
    • Possible harm.
    • Loss of Trust.
    • Undermines autonomy.
      • Freedom can get co-opted into acting in ways against own best interest.

Fairness

  • Definition
  • No arbitrary discrimination.
  • No disparate impact.
  • Why?
    • Libertarian Answer.
    • Egalitarian Answer.
  • General definition: “treating people equally without favoritism or discrimination.”
    • Merit-based definition.
  • You ensure that you don’t discriminate against already marginalized groups.
    • Most want to push the standard higher.
    • This pushes standard higher.
  • Why?
    • Could aggravate the problem.
    • Don’t aggravate existing social inequalities.
    • Undermines faith in the system.
  • Fairness
    • no arbitrary discrimination.
      • Why?
      • Libertarian: free market answer. We should be equal in marketplace (makes market less free).
      • Egalitarian: social justice, groups have a history of being discriminated against.

Transparency

  • What are we collecting?
  • How are we using it?
  • With whom do we share it?
  • roles: analyst, manager, model subject (theoretical person we are collecting on).
  • Transparency at every level.
  • Clarity in data generation collection process.
  • Explain to the subject in a way they can understand.
  • GDPR says model subject has a right to explanation.

Accountability

  • Must justify actions or decisions.
  • Aligning theory & practice.
  • Oversight
  • Plan in case of failure.
  • General definition: responsibility, “required to justify actions or decisions.”
  • Obligation to implement ethical principles, and demonstrate legal compliance, and recognize potential negative outcomes.
    • Largely role of manager.
    • responsibility, integrity, respect.
    • Comply with implementing measures that help ensure that fairness and transparency are met.
  • Some oversight committee used to maintain accountability.
  • Why?
    • If not, no justice for the people harmed.
    • Socially destabilizing.
  • Accountability
    • Responsibility, integrity, respect.
    • Comply with implementing measures that help ensure that fairness and transparency are met.
    • Ability to demonstrate compliance when someone asks. (shareholder, customer, etc.)
    • Recognize and respond to possible negative consequences.