Samuel O. Carter and John G. Dale. 2025. “Social Bias in AI: Re-Coding Innovation through Algorithmic Political Capitalism.” AI & Society (2025) 40:6. https://doi.org/10.1007/s00146-025-02540-2

This research examines the social dynamics underpinning algorithmic bias, proposing a framework for addressing these issues through the lens of algorithmic political capitalism. We explore how socio-technical-ecological relations of power often reproduce harmful algorithmic effects, including social bias, data exploitation in the knowledge economy, prejudiced predictions, and unexamined user biases that obscure power asymmetries and harm society. Building on complexity theory, particularly Morçöl’s definition of public policy as a dynamic system with co-evolving relationships between actors and systems, we analyze the challenges and opportunities to mitigate these harms within a multilayered framework. Our framework extends Keller and Block’s concept of ‘technology-dependent political capitalism’, incorporating mechanisms to ensure government assistance is conditional, allowing bicameral governance in supported corporations, and empowering local and state authorities to hold organizations accountable. Finally, we highlight the crucial roles of transparency, accountability, and democratization in fostering meaningful innovation, and argue that addressing algorithmic bias and the inequities of the knowledge economy requires a nuanced understanding of the interplay between public policy, technological systems, and societal structures. Our proposals aim to reshape the socio-technical-ecological landscape, creating conditions for algorithmic innovation that align with democratic values and equitable societal progress, while mitigating systemic violence.

Haner, Murat, Melissa M. Sloan, Justin T. Pickett, and Francis T. Cullen. 2025. “Outgroups and ingroups: how support for torture and aggressive counterterrorism policies varies by extremist type.” Social Forces. soaf190, https://doi.org/10.1093/sf/soaf190

As domestic terrorism has become central to US national security, the American public has shown divided reactions to political violence. In the current context of increasing political polarization and racial tension, we draw on social identity theory to compare responses to Islamist, left-wing, and right-wing terrorism and identify moderators of those responses. Analyses of data from a 2022 national survey experiment (n = 1,300) reveal that Americans’ responses to terrorism depend heavily on who is doing the terrorizing. Whereas Americans are equally outraged by Islamist, right-wing, and left-wing terrorism, support for controversial policies varies by terrorist type, with greater support for the use of torture on Islamist terrorists. Our findings also point toward the importance of Republicanism and white nationalist sentiment. Compared to Democrats, Republicans were more supportive of policy and the use of torture targeting Islamist terrorists and less supportive of policy targeting right-wing extremists. In addition, white nationalist sentiment corresponded to increased support for aggressive counterterrorism policy and the use of torture when applied to left-wing and Islamist terrorists. As public opinion is key to the development of government policies, it is critical that policymakers recognize the role of outgroup animosity in public support of counterterrorism measures.

Eiermann, Martin, Maria Fitzpatrick, Katharine Sadowski, and Christopher Wildeman. “How Do (Human) Child Welfare Workers Respond to Machine-Generated Risk Scores?.” Sociological Science 13 (2026): 1-21.

Algorithmic risk scoring tools have been widely incorporated into governmental decision making, yet little is known about how human decision makers interact with machine-generated risk scores at the street level. We examined such human–machine interactions in the child welfare system, a high-stakes setting where caseworkers ascertain whether government interventions in family life are warranted. Using novel data—verbatim transcripts of caseworker discussions—we found that decision makers: (1) disregarded scores in the middle of the distribution while paying attention to extremely high or low risk scores and (2) rationalized divergences between human decisions and machine-generated scores by highlighting the algorithm’s overemphasis on historical data and specific risk factors and its lack of contextual knowledge. This meant that caseworkers were unlikely to modify their decisions so that they aligned with risk scores. However, we did not find evidence of principled resistance to algorithmic tools. Our findings advance research on such tools by specifying how human perceptions of the utility and limitations of novel technologies shape discretionary decision making by state officials; and they help to explain their uneven and potentially modest impact on the bureaucratic management of social vulnerability.