4 Questions about the Algorithmic Accountability Act 2022: What it Means for Organizations
Decisions that impact individuals’ lives are increasingly being made by Artificial Intelligence and Machine Learning algorithms. When AI and ML algorithms are used for making decisions, it is called algorithmic decision-making. These decisions can include loan applications, job applications, housing, and sentencing length, to name a few.
These decisions are often efficient and effective, but they can also be biased, negatively impacting lives. These biases are usually based on protected information like an individual’s race, gender, and income.
Since these decisions can be consequential in people’s lives, organizations that use algorithms have been asked to share how the decisions are made. Citing that the algorithmic decision-making systems are “black boxes,” organizations have provided little or no transparency in how the systems generate the decisions.
Essentially, these organizations become non-accountable when their systems negatively affect or cause harm to people.
For many years, researchers have discussed the need for transparency and accountability of artificial intelligence especially given the issues that have come to light regarding how Artificial Intelligence and Machine Learning (AI/ML) reproduce social and political biases that affect individuals’ lives and livelihoods.
These issues are known to disproportionately affect minoritized individuals and communities. Researchers have been calling for legislation that addresses accountability, transparency, and fairness on organizations’ use of AI/ML.
Therefore, the introduction of the Algorithmic Accountability Act, 2022, seems timely.
Before celebrating the timeliness of the bill, let’s note that this is only an introduction. There’s no telling if or when it will pass.
What is the Algorithmic Accountability Act of 2022?
The Algorithmic Accountability Act of 2022 is a bill introduced in February 2022 by U.S. Senator Ron Wyden, D-Ore., with Senator Cory Booker, D-N.J., and Representative Yvette Clarke, D-N.Y. The act highlights the need for
“transparency and oversight of software, algorithms, and other automated systems that are used to make critical decisions about nearly every aspect of Americans’ lives.”
The preamble from the bill is that:
“When algorithms determine who goes to college, who gets healthcare, who gets a home, and even who goes to prison, algorithmic discrimination must be treated as the highly significant issue that it is. These large and impactful decisions, which have become increasingly void of human input, are forming the foundation of our American society that generations to come will build upon.
And yet, they are subject to a wide range of flaws from programming bias to faulty datasets that can reinforce broader societal discrimination, particularly against women and people of color. It is long past time Congress act to hold companies and software developers accountable for their discrimination by automation,” said Rep. Clarke.
“With our renewed Algorithmic Accountability Act, large companies will no longer be able to turn a blind eye towards the deleterious impact of their automated systems, intended or not. We must ensure that our 21st Century technologies become tools of empowerment, rather than marginalization and seclusion.” – Wyden, Booker, and Clarke, 2022
“The bill requires companies to conduct impact assessments for bias, effectiveness, and other factors when using automated decision systems to make critical decisions.” – Wyden, Booker, and Clarke, 2022
In case you’re still wondering what algorithmic accountability really is, I will define it in the next section.
What is Algorithmic Accountability?
Algorithmic accountability is the assessment of bias and prejudices that are built into or result from algorithms. Algorithmic accountability looks at the process of assigning responsibility and blame for harm when algorithmic decision-making leads to discrimination and inequity.
The term algorithmic accountability comes from combining two terms, ‘algorithm’ and ‘accountability.’ An algorithm is a set of instructions or procedures presented to a computer system that considers the social, cultural, and technical contexts in which the instruction is deployed.
Accountability is the obligation to explain or justify behavior to another entity, where the entity has the power to ask questions, pass judgment or impose punishment on the explainer (Bovens p. 447). The explainer could be an individual or an organization.
Although the combined term and its application to the context of AI/ML is new, the notion of holding someone or something accountable for the results of computer systems is not new. An example of an older term that captures a similar goal is ‘computational accountability’.
What is an Algorithmic Decision?
The bill defines “decision” as judgments that impact individuals’ lives legally or materially in terms of education, employment, utilities, family planning, financial services, healthcare, housing, and legal services, to name a few (Wyden, Booker, and Clarke, 2022).
What Types of Organizations will be affected by the Algorithmic Accountability Act?
In essence, what types of organizations will be held responsible if algorithmic decision-making results in inequity? The bill notes that organizations making at least $50,000,000 in revenue that use AI/ML for decision-making that impacts individuals’ lives fall under this act
Who will Enforce the Algorithmic Accountability Act 2022?
The bill directs the Federal Trade Commission as the governing body in charge of enforcing the act. It also allows the FTC to hire more staff to enforce the law.
Conclusion
The introduction of the Algorithmic Accountability Act of 2022 is a good start. Let’s wait and see whether and when the bill will be deemed important enough to pass.
References
Mark Bovens. 2007. Analysing and Assessing Accountability: A Conceptual Framework. European Law Journal 13, 4 (2007), 447–468. https://doi.org/10. 1111/j.1468-0386.2007.00378.x arXiv:1468-0386
https://datasociety.net/library/algorithmic-accountability-a-primer/
Copyright 2024 © Indigomark - All Rights Reserved.
