The Simulation Experience
When I first stepped into the role of a county judge in the Detain/Release simulation, I must admit that I did not initially internalize the profound weight of the judicial responsibility. Lacking a clear set of decision-making criteria, my eyes were immediately drawn to the Risk Assessment Tool, primarily because of its vibrant, salient visual design. I then prioritized the prosecutor’s recommendation, while the defendant’s statement and photo felt like secondary factors in my initial rulings.
Early on, my decisions were inconsistent; I fluctuated between following the prosecution and making unstructured decision-making. Consequently, I was voted out for being risk-tolerant and releasing too many defendants. I thought this failure highlighted a lack of systemic logic. In my subsequent attempts, I established a rigid judgements: I would only detain if both ‘Fail to Appear’ and ‘Commit a Crime’ indicators were at or above the medium threshold. When I reapplied this rule consistently (I ignored all other qualitative factors) the simulation concluded without hitting the ‘Fear’ or ‘Jail Capacity limits. However, while this mathematical consistency won the game, it left me with a profound sense of uncertainty: would this same logic remain just if applied to a hundred or a thousand real-life cases?
Reflection
A significant takeaway from this experience was the difficulty of making a just decision when overwhelmed by information whose origin remains a black box. I found myself questioning the foundations of the tools provided: How is the Risk Assessment score calculated? What exactly constitutes a high risk versus a medium one? What data informs the prosecutor’s stance?Without transparency regarding these algorithms, the labels feel arbitrary yet authoritative.
Furthermore, I realized how the simulation’s mechanics of visible meters for the jail capacity and the public fear pressured me to treat a human life as a binary Yes/No choice. Despite having studied the ethical implications of these technologies in our course modules, the moment-to-moment pressure of the simulation forced me to rely heavily on the risk assessment scores. This created a ‘Mathematical Shield’ as described by Dr. Cathy O’Neil mentioned. (Talks at Google, 2016). Because I was following a score, I felt a misguided sense of security that my decisions were correct and objective, even though I was merely following a pre-calculated probability.
This reliance echoes Dr. O’Neil’s warning that algorithms, while appearing objective, are often weapons of math destruction embedded with the biases and definitions of success held by their creators. As Dr. Shannon Vallor (2018) notes, AI acts as a mirror to our society, it does not reflect an ideal version of justice, but rather a distorted and amplified version of our existing human prejudices and historical data. When we use these tools to make life-altering decisions, we risk automating and accelerating social inequality under the guise of efficiency.
Though this was a simulation, it felt like a real-world reminder. Algorithms now permeate nearly every facet of our lives. This suggests that both human-led and algorithm-aided decision-making ultimately require robust governance frameworks, such as structured discussion and rigorous auditing. While algorithms excel at accelerating the selection process, achieving a just and better decision necessitates a human-centric approach to ensure ethical accountability and contextual depth. Rather than viewing these tools as neutral instruments, this experience invites us to approach them with critical digital literacy. Ultimately, it is vital to maintain a sharp, critical eye, especially in situations requiring human empathy, to ensure that the ‘human’ is not entirely erased from the pursuit of justice.
Gemini (Google, 2026) was used to refine the grammar, spelling, and overall clarity of this post.
References
O’Neil, C. (2017, July 16). How can we stop algorithms telling lies?Links to an external site. The Observer.
Santa Clara University. (2018, November 6). *Lessons from the AI Mirror Shannon VallorLinks to an external site. [Video]. YouTube.
Talks at Google. (2016, November 2). Weapons of math destruction | Cathy O’Neil | Talks at GoogleLinks to an external site. [Video]. YouTube.
Mars, R. (Host). (2017, September 5). The Age of the Algorithm (no. 274)Links to an external site. [Audio podcast episode]. In 99 Percent Invisible.
Porcaro, K. (2019). Detain/Release [web simulation]. Berkman Klein Center.
Porcaro, K. (2019, January 8). *Detain/Release: simulating algorithmic risk assessments at pretrial.*Links to an external site. Medium.
