IP#3 - Algorithms

Recidivism Risk is the tendency to relapse into a previous condition or mode of behaviour (Merriam-Webster, 2023). Cathy O’Neil, in her book Weapons of Math Destruction, criticised these algorithms for their potential biases and negative impacts on individuals and communities. …Whether we’ve eliminated human bias or simply camouflaged it with technology. The new recidivism models are complicated and mathematical. But embedded within these models are a host of assumptions, some of them prejudicial (O'Neil, 2017, p.28). Cathy O'Neil, throughout her book speaks to the fact that recidivism can perpetuate injustice and undermine the principles of fairness and rehabilitation in the criminal justice system. She emphasises the need for transparency, accountability, and careful evaluation of these algorithms to ensure that they do not contribute to further harm.

In my past schools I have noticed many different instances that I am now reflecting upon in a different light having now read Cathy O’Neil’s book. Although we don’t have (that I’m aware of) a program that indicates whether a child is at a high or low risk of recidivism we do have files that follow students and can impact our perception of that student. Teachers are asked, before the school year begins, to take time to review and look at who we have in our classroom and what notes come along with them. I personally haven’t done this since my second year of teaching. My first year teaching I read notes on how a student was physical, their incident reports, their academics/lack of interest, etc. Beyond that, I also had previous teachers coming and ‘warning me’ about said student, how much time they spent in the office and their outbursts. These notes and comments caused me to have a bias towards this particular student and my brain automatically was filled with the year I may face with them. The reason I stopped reading student notes prior to school was because I was then able to allow students to begin their school year with a fresh start. Removing these biases has aided me in creating a classroom where students can flourish and be who they want to be, not who they were or who I expect them to be. I am then able to reference their files’ notes when and if I believe there is an issue arising (new or old).

In my experience (elementary is my specialty), I have also seen inconsistent discipline/punishment in my schools based on who the student, in question, is. I have found over my past six years of teaching a multitude of differing ways to discipline/punish students in the schools. You may send a student to one of the admin as they have been the one to deal with them their whole school career. You may send a student to one admin as they are more severe and demanding of respect. You may not send one to the admin or the office as they don’t respond to that form of discipline. You may send one to admin as there are issues with the parents and they are the ones that need to broach the conversation between home and school. I could go on and on with instance upon instance of different reasons you may choose a particular solution over another but the reason I bring these to light is that I find these different ways of dealing with the matter at hand can cause biases towards certain students, mixed messages about how things at school are dealt with, or cause confusion for children in knowing expectations. I also find, at times the rules put in place by the board aren’t upheld because of who they are dealing with or the students history.  

When thinking of a  behavioural risk algorithm for an elementary school, it could be designed to monitor student behaviour and identify potential risk factors that may lead to disruptive or harmful behaviour. This algorithm could use data from a variety of sources, including student records, attendance data, disciplinary records, and information gathered from teachers and staff. Once potential risk factors are identified, the algorithm could trigger an alert or notify relevant school personnel, such as the school counsellor or administrator, who could then take appropriate steps to intervene and provide support to the student. This could include targeted counselling, increased supervision, or other interventions aimed at addressing the underlying issues that may be contributing to the student's behaviour. 

When reflecting on my past 6 years of teaching and comparing to the recidivism discussed by Cathy O’Neil much like she outlined with recidivism risk algorithms, there are potential risks and biases associated with using such algorithms in the school context. For example, the data used to create the algorithm may reflect systemic biases such as racial or socioeconomic disparities in discipline practices. This could lead to the algorithm unfairly targeting certain groups of students and exacerbating existing inequities. Additionally, using a behavioural risk algorithm could lead to a focus on punishment rather than prevention and support, as students who are identified as high-risk may be subject to increased surveillance and discipline. This could create a negative and punitive school climate, which may not be conducive to learning and growth. I believe it is important to note that I do believe there are significant differences between recidivism risk algorithms and behavioural risk algorithms in terms of their intended use and potential impact.

In terms of implications for the use of big data in education, it is important to approach the use of algorithms with caution and consideration for potential biases and unintended consequences. Much like Cathy O’Neil discussed these algorithms can be tweaked and used for many different reasons. Like she said, Compare this attitude to the one found at Amazon.com. The giant retailer, like the criminal justice system, is highly focused on a form of recidivism. But Amazon’s goal is the opposite. It wants people to come back again and again to buy. Its software system targets recidivism and encourages it (O’Neil, 2017, p.87). I believe when it comes to using these algorithms in either schools or judicially, there should be transparency and accountability in the development and use of such algorithms, as well as ongoing evaluation to ensure that they are not perpetuating inequities or harming students or inmates.