Roles for Computing in Social Change

Lead Scribe: Damon Coffey

Roles for Computing in Social Change -concerns about fairness, bias and accountability in the field

Introduction:

• high stakes decision making algorithms have potential to predict outcomes more accurately

• cs has generally failed to target the correct point of intervention

• ex: intervention at the selection phase in an employment context could prevent a hostile work place

Computing as a Diagnostic

• computing can help us measure social problems and diagnose how they manifest in tech systems

• computing cannot solve issues on its own

• Diagnostics work can be valuable

  • highlight tech dimensions of social problems

• misinformation can negatively affect marginalized populations more ex: search engines displaying low quality health information

• not presented as solutions, rather as tools to document practices

  • not to confuse diagnostics with treatment

  • computing is not unique in helping diagnose social problems

    • sociology, etc..

  • certain tools can be treated as certainity for every situation, which is not the case

Computing as a Formalizer

• computing requires explicit specification of inputs and goals

• these inputs and goals can be affected by transperency, accountability and stake holder participation

  • need to be precise ex: risk assessment: debate over how to formalize pretrial risk, if and how to use these instruments

• not all data is easy to quantify

• may press people to rely on measures that are incorrect

Computing as Rebuttal

• computing can clarify the limits of technical interventions and of policies promised on them

• limits of computing can drive people to reject computational approaches

• ex: using an algorithm to determine an immigrant’s societal worth, not good. Should seek a different method rather than forcing a technological one

• need to understand what algorithms are actually capable of, instead of forcing it on everything

  • need to show what an algorithm CANT do (prove limits)

• prediction algorithms for risk assessment

• computational research on fairness is built on discrimination law

• Risks

  • proclomations of what a computational tool is incapable of may focus on improving tool even if it is not possible

Computing as a Synecdoche

• computing can foreground long standing social problems in a new way

• Eubank’s core concern: computing is just one mechanism through which longstanding poverty policy is manifested

• Automated systems can divert poor people from the resources they need

• computing can help bring attention to old problems, however

• synecdochal focus on computing must walk a pragmatic line between over emphasis on tech aspects and recognition of the work tech actually does

• need to find a balance between the two and develop better systems with more emphasis on social issues