Selected quotes
“Algorithmic accountability is the process of assigning responsibility for harm when algorithmic decision-making results in discriminatory and inequitable outcomes”
“Among many applications, algorithms are used to:
• Sort résumés for job applications;
• Allocate social services;
• Decide who sees advertisements for open positions, housing, and products;
• Decide who should be promoted or fired;
• Estimate a person’s risk of committing crimes or the length of a prison term;
• Assess and allocate insurance and benefits;
• Obtain and determine credit; and
• Rank and curate news and information in search engines.”
“Algorithmic systems present a special challenge to assessors, because the harms of these systems are unevenly distributed, emerge only after they are integrated into society, or are often only visible in the aggregate”
“What our research indicates is that the risk of self-regulation lies not so much in a corrupted reporting and assessment process, but in the capacity of industry to define the methods and metrics used to measure the impact of proposed systems”
How Are Algorithms Used to Make Decisions?
Example: Racial Bias in Algorithms of Incarceration
Complications with Algorithmic Systems
• Fairness and Bias
• Opacity and Transparency
• Repurposing Data and
Repurposing Algorithms
• Lack of Standards for Auditing
• Power and Control
• Trust and Expertise
What is Algorithmic Accountability?
• Auditing by Journalists
• Enforcement and Regulation
.
In summary: The Algorithmic Impact Assessment is a new concept for regulating algorithmic systems and protecting the public interest. Assembling Accountability: Algorithmic Impact Assessment for the Public Interest is a report that maps the challenges of constructing algorithmic impact assessments (AIAs) and provides a framework for evaluating the effectiveness of current and proposed AIA regimes. This framework is a practical tool for regulators, advocates, public-interest technologists, technology companies, and critical scholars who are identifying, assessing, and acting upon algorithmic harms.
First, report authors Emanuel Moss, Elizabeth Anne Watkins, Ranjit Singh, Madeleine Clare Elish, and Jacob Metcalf analyze the use of impact assessment in other domains, including finance, the environment, human rights, and privacy. Building on this comparative analysis, they then identify common components of existing impact assessment practices in order to provide a framework for evaluating current and proposed AIA regimes. The authors find that a singular, generalized model for AIAs would not be effective due to the variances of governing bodies, specific systems being evaluated, and the range of impacted communities.
After illustrating the novel decision points required for the development of effective AIAs, the report specifies ten necessary components that constitute robust impact assessment regimes.
.
What is an Impact?
What is Accountability?
What is Impact Assessment?
THE CONSTITUTIVE COMPONENTS OF IMPACT ASSESSMENT
Sources of Legitimacy
Actors and Forum
Catalyzing Event
Time Frame
Public Access
Public Consultation
Method
Assessors
Impacts
Harms and Redress
TOWARD ALGORITHMIC IMPACT ASSESSMENTS
Existing and Proposed AIA Regulations
Algorithmic Audits
External (Third and Second Party) Audits
Internal (First-Party) Technical Audits and
Governance Mechanisms
Sociotechnical Expertise
CONCLUSION:
GOVERNING WITH AIAs
ACKNOWLEDGMENTS
- Databite No. 145: Algorithmic Governance and the State of Impact Assessment in the EU, US, and Canada. (n.d.). Data & Society. Retrieved 18 December 2021,
-
Moss, E. (2021, June 29). Assembling Accountability, from the Ground Up. Medium.
