So, it looks like I was a bit too lenient with the drug dealers in my caseload! ????♀️
I think that I was trying so hard to not be biased that my decisions may have been impacted. The piece of data that I relied most on was the risk of violence. This simulation was a great example of how important it is to reflect on the many implications and consequences of relying on AI-informed decision making.
As the Gimlet Media Reply All podcast (2022) reveals, the more data that is collected in an area (high crime neighbourhoods) the more details will be uncovered about the crimes and the more arrests will likely be made. The data then turns into a self-affirming cycle.
Does this happen in education as well? My school has begun to try and crack down on a rise in students not completing their courses. We are now collecting more official data on students’ attendance by recording how many weeks behind students are in their asynchronous, online courses in our Student Information System (SIS). Every two weeks we need to record if our students are two or more weeks behind. The office then takes that data and uses it to make decisions about how to proceed. Usually, we start with support measures to see what is holding the student back from participating, but after several cycles of nothing changing on the student’s part, we move to more disciplinary actions such as reaching out to the provincial attendance board.
I think that my school is doing a great job at using the initial data from the teachers, to inform the SIS. Over time I am sure that the built-in AI in our SIS will help the administration to find correlations between the types of registrations that tend to enroll and not complete courses and use that data to inform administrators to make enrolment decisions in the future. Thankfully, we are in the business of people (students); meaning I think there is less risk in allowing algorithms to take over our decision making. While I am not naive enough to completely overlook the fact that there are risks of being too data and budget driven, I do still believe that fundamentally our education system does still hold student success as a core value.
There are many other areas of life that are vulnerable to the use of AI. For example, the lesson mentions “For many years, scientists and linguists have been in pursuit of replicating human speech and reasoning patterns in ways convincing enough that would make human and AI speech indistinguishable (Pena, 2023).” I have noticed in the news (see article referenced below) and on Tik Tok recently, there appears to be a rise in the number of people being duped by AI generated voices of their teenage or adult children in trouble begging for money during phone scams. I must admit this does make a little more conscious and cautious of the photos and videos that I post to social media.
References
Gimlet Media. (2022, September 23). #127 The Crime Machine, Part I. Reply All [Audio Podcast]. https://gimletmedia.com/shows/reply-all
Pena, E. (2023, November 17). [11.3] A new breakout? Canvas. https://canvas.ubc.ca/courses/124246/pages/11-dot-3-a-new-breakout?module_item_id=5919144
Phone scammers are using artificial intelligence to mimic voices. (2023, July 11). CBS News – Breaking news, 24/7 live streaming news & top stories. https://www.cbsnews.com/news/artificial-intelligence-phone-scam-fake-voice/
