An Algorithm Is Targeting Black Families for Child Welfare Investigations

Algorithms have become increasingly influential in various aspects of our lives, now extending their reach into child welfare services, where they are used to identify families for neglect investigations. However, evidence suggests these tools are not only flawed but also exhibit racial biases. This is not the first instance of algorithms designed to assist inadvertently causing harm. From creating echo chambers during the 2016 elections to influencing targeted advertising on social media, algorithms play a significant role in shaping the information we encounter, which in turn can reinforce our biases.

Disproportionate Impact on Black Families

Recent findings from Carnegie Mellon University, shared exclusively by the Associated Press, reveal that a predictive algorithm employed by child welfare services in Allegheny County, Pennsylvania, disproportionately flags Black children for mandatory neglect investigations compared to their white peers. Alarmingly, social workers often disagree with the algorithm’s risk assessments—about one-third of the time. This suggests the algorithm, known as the Allegheny Family Screening Tool (AFST), would receive a grade of D+ if evaluated, with a mere 67% accuracy.

The Challenge of Identifying Bias

Identifying the specific issues with the algorithm is challenging. As noted by Rebecca Heiliweil from Vox, understanding the root causes of an algorithm’s bias is nearly impossible since users typically only see the end results. In the case of the Allegheny algorithm, there is no clarity regarding which factors are prioritized in assessing neglect, leaving room for interpretation and bias. The algorithm considers various factors, such as housing conditions and hygiene standards, which can be vague and subjective.

Moreover, it relies on extensive personal data collected from birth, including Medicaid records and substance abuse histories—data that are inherently influenced by systemic biases, particularly from institutions rooted in racial inequality. The programmers behind these algorithms are human and carry their own biases, leading to concerns that machine learning can perpetuate and amplify existing social and racial injustices. When algorithms like AFST operate without human oversight, they risk making hasty, detrimental decisions that can significantly affect the lives of many families.

Consequences of Algorithmic Bias

Public Citizen, a consumer advocacy organization, highlights that algorithmic bias has tangible consequences for people of color across various sectors. For instance, communities of color may face higher car insurance rates compared to white communities with similar accident histories due to predictive algorithms. Social media platforms such as TikTok and Instagram have faced backlash from Black creators whose content is often incorrectly moderated by algorithmic systems.

These “black box” algorithms can be likened to the scene in Fantasia where Mickey Mouse, as the sorcerer’s apprentice, creates chaos by programming a broom for efficiency. Left unchecked, such systems can escalate problems rather than solve them. The issues observed in Allegheny County may very well exist in similar systems across the country, posing risks that could outweigh any potential benefits.

Search Queries

Conclusion

In summary, the use of algorithms in child welfare services raises significant concerns about racial bias and accuracy. The Allegheny Family Screening Tool’s flawed assessments highlight the dangers of relying on technology without proper oversight. As these systems become more prevalent, it’s crucial to scrutinize their impact on marginalized communities.

Keyphrase: Algorithmic Bias in Child Welfare Investigations

Tags: [“home insemination kit”, “home insemination syringe”, “self insemination”]