How Formerly Incarcerated People Envision Technologies for Prison Parole

Technical Summary: How Formerly Incarcerated People Envision Technologies for Prison Parole

Problem Statement
The integration of AI-driven algorithms and automated tools into the parole system has increasingly shaped eligibility, release decisions, and post-release surveillance. While these systems are often framed as objective solutions to inefficiency and bias, they are rarely designed with input from justice-impacted individuals. Consequently, computational systems often encode institutional goals and assumptions that fail to reflect the lived realities of those navigating parole. Current literature predominantly frames parole technology through three lenses: prediction/risk management, surveillance/compliance, and fairness/accountability critiques. However, there is a significant gap in understanding how justice-impacted individuals themselves imagine technology as a resource for navigating these power structures. Without their input, these tools risk expanding state control and reinforcing historical disparities rather than supporting reintegration.

Methodology
To address this gap, the authors conducted a qualitative study grounded in the principle of "nothing about us without us." The research team, which included a co-author with over 25 years of incarceration experience and two parole board hearings, designed a trauma-informed, asynchronous mail-in survey.

• Participants: 31 formerly incarcerated individuals (all male, primarily Hispanic, with incarceration lengths ranging from 10 to 25+ years) who had experienced the parole process within the California Department of Corrections and Rehabilitation (CDCR).
• Data Collection: An asynchronous, paper-based survey was chosen to allow participants to respond at their own pace, reducing emotional distress associated with recounting sensitive parole experiences. The survey included 10 open-ended questions focusing on parole preparation challenges, collaboration with family, cultural factors, and visions for future computational tools (including AI-enhanced systems).
• Analysis: Researchers employed qualitative thematic analysis, involving independent open coding, iterative codebook refinement, and collaborative discussion to identify salient themes and tensions. The analysis was guided by the lived experience of the research team to ensure interpretive validity.

Key Results
Participants did not envision technology as a tool to dismantle the prison system or solely as a mechanism for surveillance. Instead, they imagined tools designed to build "strategic agency"—the capacity to interpret, anticipate, and effectively respond to institutional power to pursue freedom. Three primary themes emerged:

1. Navigating Cultural Stereotyping: Participants described a "double bind" where parole board commissioners held cultural stereotypes about how a "rehabilitated" person should speak and behave.

   • Challenge: If individuals conformed to stereotypes, they were seen as unchanged; if they exceeded expectations (e.g., through education or articulate speech), they were often viewed as manipulative or inauthentic.
   • Vision: Participants imagined tools to track and quantify pro-social activities to create "board-legible" evidence of transformation, reducing reliance on subjective cultural impressions. They also envisioned culturally aware tools that translate abstract parole concepts (e.g., "insight") into culturally familiar terms and help individuals anticipate how their narratives might be interpreted through stereotyped lenses.

2. Informal Information Networks: Due to a lack of official guidance, incarcerated individuals rely on informal networks (peers, transcripts, shared experiences) to understand commissioner expectations and "profile" board members.

   • Challenge: Those without external support networks face significant disadvantages.
   • Vision: Participants proposed tools to stabilize and extend these networks, such as accessible databases of felon-friendly employers, multimodal guidance (e.g., etiquette videos), and platforms for mock hearings and sharing board experiences.

3. Family Support Networks: Parole preparation is often a collective effort involving unpaid, invisible labor by family members who manage documentation, research housing, and navigate complex systems.

   • Challenge: Families often struggle with the specialized language of the parole system and face barriers such as language differences (e.g., Spanish-speaking families) and fear of institutional repercussions.
   • Vision: Participants envisioned family-centered tools, including multilingual support (specifically Spanish), document generation aids, and resources that break down complex concepts into lay terms. These tools would help families understand where they can intervene effectively and support the incarcerated individual's strategic preparation.

Key Contributions

• Empirical Contribution: The study provides the first detailed account of how formerly incarcerated individuals envision computational tools for parole, shifting the focus from institutional needs to the navigational strategies of justice-impacted people.
• Design Contribution: The paper reframes the design of parole technology away from risk assessment and surveillance. Instead, it proposes a design direction focused on strategic agency, where tools help individuals and their families build the capacity to navigate existing power structures.
• Theoretical Contribution: The authors introduce "strategic agency" as a distinct form of response to algorithmic harm under conditions of extreme power asymmetry. Unlike resistance or collective action, strategic agency involves learning to operate within institutional constraints to achieve specific outcomes (freedom).

Significance and Design Implications
The paper argues that computational tools for parole should not aim to automate decisions or simply expose bias, but rather to scaffold learning, translation, and preparation. The authors propose that these tools should be designed for family members and support networks rather than directly for incarcerated individuals, as prison approval for software is often restrictive and risky.

• Proposed Implementation: The authors suggest "alternative social platforms" such as web browser plugins that operate within the information ecosystems families already use. These tools could guide families in helping incarcerated loved ones develop the strategic agency needed for parole hearings.
• Broader Impact: The concept of strategic agency is presented as a lens for understanding engagement in other high-stakes bureaucratic systems (e.g., immigration, welfare) where outcomes depend on discretionary evaluation and credibility judgments.

Limitations
The authors acknowledge limitations, including the reliance on an asynchronous survey which limited real-time clarification, the demographic homogeneity of the sample (all male, primarily Hispanic, California-based), and the potential bias in recruitment through organizations that may favor those with stronger support networks. The findings are presented as a theoretical contribution and an analytic lens rather than a statistically generalizable claim.