Jordon Lovig ETEC 540

For the final exercise of the course, I’ve chosen to present two scenarios related to the topic of my final assignment, which describes text analysis technologies as a way to measure student academic success and mental wellness. In both scenarios, we have a UBC course instructor… let’s say his name is Jordon. Jordon teaches an online English Literature course for first year undergraduate students. As a new(ish) professor not yet on tenure track, Jordon is very interested in the health and wellbeing of his students, and believes that if problems are identifies early on, he can intervene before they grow out of hand. To this end, Jordon utilizes text analysis technology to moderate discussion posts of the students in his class. He’s keen to identify students that may be in danger of straying off the right path. Below are two scenarios as to how this could play out.

Scenario 1- Utopian

As part of the course material for his online English Literature course, Jordon posts a discussion prompt in the form of a question in the discussion forums. Students are expected to apply core concepts from that weeks’ reading material to answer the question, as well as post their own questions to generate further discussion.

Over the course of the first few weeks, Jordon employs LIWC-22 text analysis tools to measure various dimensions of each student’s text, including word count, analytic vs authentic word usage and words per sentence. Early in the course, Jordon notices that there is one student whose textual data skews differently from the rest of the class. The student- James- relies heavily on authentic words as opposed to analytical. James’ writing style is dynamic in that it uses pronouns, conjunctions and negations in much higher frequency than his classmates. Upon closer inspection of James’ actual text submissions, it’s determined that he seems to be writing more about himself than the course material. More than that, James seems to be going through a rough time- he references loneliness, isolation and a recent breakup on more than one occasion. In some instances James doesn’t seem to be interacting with the course material at all, just writing stream of consciousness style responses based on his personal life.

Jordon Reaches out to James to see if how the course is going and if he needs any help with the required readings. James is happy to hear from Jordon- in fact, he mentions that he hasn’t really had direct contact from anyone over the past several months, since his partner left him very suddenly, and he’s been feeling very alone and despondent. Jordon listens to the student, and lets him know that there is support available for him. He passes on the contact information for UBC counselling services and shares some of his own experiences with similar situation from his personal past. James is receptive to the message, and books his first appointment with counselling services the following day. As the days and weeks go by, James’ work in the course starts to improve. He’s engaging with the course material more and making valuable contributions to the discussion forum. James admits he was in a bad situation but didn’t feel he had anyone to reach out to, so many of his issues were coming out in his forum postings. He’s thankful for the early intervention, and his academics and mental health are on their way to getting back on track.

Scenario 2- Dystopian

Same as Scenario 2, Jordon is a new instructor for an online English Literature course, and he intends to use textual analysis technology to monitor the performance and mental health of his students by analyzing their forum posts on a weekly basis.

Once again, a few weeks into the course Jordon’s analysis reveals that a student, James, is contributing a bit differently than his classmates. James’ writing style is dynamic and his word choice is authentic, not analytical. With the best of intentions, Jordon decides that an intervention is warranted, so he arranges for a Zoom chat with James. Things don’t quite go as expected. James is totally taken aback by Jordon’s suggestion that he may not be in a great state of mind. James had no idea that his forum postings were being monitored so closely. He asks why he wasn’t informed that his data would be collected at the start of the course, and questions why UBC doesn’t have a clear policy with regards to use of student data. James reveals that he is from a cultural background with a proud tradition of oral-based knowledge transfer. Based on his prior learning, it makes perfect sense that he would engage with course material in a way that may seem “dynamic” and “personal” compared to his classmates. Additionally, James is currently on a personal journey exploring his own personal gender identity, which is why the use of personal pronouns are more pronounced in his work. If Jordon has bothered to actually read James’s posts closely and consider the inherit biases in the LIWC algorithm, perhaps he would have seen that this was the case.

James is now discouraged from engaging in the course material. He feels he is “under a microscope” and that relating the course reading to his own personal experiences is not a valid approach to learning. James withdraws from the course prior to the withdrawal deadline and discontinues his studies at UBC.

Jordon may have meant well in this scenario, but in his excitement to implement a new learning technology, he totally lost track of the human being on the other side of the equation. It’s a learning experience to be sure. In the future, he vows to inform students of his practices at the beginning of the course, to refer them to relevant institution policy if they have any questions or concerns, and to always remember that technology has its own biases coded into it. Hopefully, this won’t happen again.

#1

Joseph Villella, Task 10: Attention Economy

https://blogs.ubc.ca/jvillella540/2022/07/21/task-10-attention-economy/

Joseph and I had a similar experience with this task as we both found it enjoyable and entertaining, Where our experiences differed seems to be that his focus was more so on the practical applications of manipulative design interfaces where as I was more concerned with the frustrating user experience.

I appreciate how Joseph pointed out that the exercise was a good demonstration of the dangers of blindly clicking boxes and giving permission without really thinking things through. Organizations can easily fool end users into accepting terms and conditions that are not favourable to them or even installing malicious software on their devices if we don’t pay close attention to what it is we are agreeing to.

We both highlighted the cultural norms of certain colours an layouts leading to specific responses- for example, green means yes, red means no. Joseph’s post had me thinking that if a bad faith actor chose to do so, they could exploit these norms an manipulate end users into agreeing to terms or accepting conditions that the otherwise would not have. I also appreciate how he points out that we can combat these shady practices with early intervention- educating children at an early age.

#2

Jessica Presta, Task 11: Detain/Release

https://blogs.ubc.ca/jpresta/2022/07/31/task-11-algorithms-detain-release/

I enjoyed reading Jessica’s write up on her experience with the Detain/Release program, mostly because I feel it differed from my own quite a bit. I think that Jessica approached the task with a great deal of skepticism, and it sounds like she put a great deal of thought into her decisions.

My own experience was a bit more “by the seat of my pants” so to speak, as I found myself making decisions rather hastily and allowing the AI to do most of the thinking for me, particularly towards the end.

Jessica highlights some fantastic resources in her post that really speak to the dangers of relying on AI for decision making processes such as this. We both noted that bias can be hard-coded into algorithms, but only Jessica pointed out the potential for racist profiling before completing the exercise, where as in my post I note that I didn’t really consider the possibility until after the fact.

My take away is that it’s important to approach important tasks intentionally, do your research beforehand and be aware of potential biases before making your decisions- it’s not enough to simply acknowledge them after the fact. I feel that if nothing else, Jessica would make a better lawyer than me.

#3

Melissa Santo, Task 3: Voice to Text

https://blogs.ubc.ca/melissasanto/2022/06/20/etec-540-task-3-voice-to-text/

One thing Melissa and I have in common is that we both neglected to verbally add punctuation to our audio records. However, she points out this mistake right away, mentioning she thought of it as soon as the task finished, whereas I didn’t know that was a standard practice until… well, until I read her post just now.

Both Melissa and I highlight some standard errors in our text, which I think is to be expected using this sort of technology. One difference between our assessments is that Melissa focused a bit more on the technical aspect of the exercise, where as I was more focused on whether or not the technology got the general feel and theme of what I was trying to communicated.

We also both were more or less just speaking off the top of our heads, and we both chose an emotional story to tell. Melissa and I both noticed that the technology does a poor job of conveying feeling, which is a major limitation to the technology. Reading her post caused me to reflect a bit on this point, how as technology advances it still lags behind the human experience. I think this is especially true when it comes to analyzing the written word, something that both Melissa’s post and my own make a point of conveying.

#4

Jacey Bell, Task 6: An Emoji Story

https://blogs.ubc.ca/jaceysmetcollection/2022/07/08/task-6-an-emoji-story/

I chose to link Jacey’s task for this part of the assignment because the second I saw her emoji story I knew without question which movie she was talking about (UP, 2009).

Jacey’s contribution really helped me realize how communicating with images- in this case emojis- can be a totally valid and useful mode of communication, it just all depends on context. “Reading” through her post, the title of the film was immediately obvious, and only required one emoji. The plot was fairly easy to convey as well. I think this is because the film itself is very visual and light on dialogue. Additionally, it’s a pretty well-known movie and I’m personally very familiar with it. I think she did a great job of conveying the plot through the use of images.

Contrasting Jacey’s task to my own, it’s a world of difference. The novel I chose, “White Noise” by Don DeLillo, is almost the exact opposite of the film “Up”. My novel is very dense with a great deal of dialogue and very little imagery. It also doesn’t have much plot, it’s more focused on themes and ideas. As such, I think my post makes a great foil to Jacey’s- it’s an example of a work that cannot be easily communicated through use of images.

Again, my take away is that mode of communication is highly contextual. Some modes work very well in some cases, but don’t work well at all in others.

#5

Jane Wu, Task 9: Network Analysis

https://met.for.education/?p=337

I had some technically difficulties accessing the Palladio app, but the course instructor was kind enough to take some time to share the data with me via Zoom. Although I was grateful for his assistance, I was disappointed that I wasn’t able to dig around with the data on my own. With that in mind, for this linking assignment I wanted to be sure I selected a classmate who was able to access the data to see how our experiences differed.

Jane did a great job of pulling a lot of useful information out the raw data that was provided. She came to a similar conclusion as the rest of us- that there were some consistencies in song selection across groups and individuals. However, the community she was in (1) made some different selections than what I observed when shown the data. Percussion and Night Chant were both present in her community’s top selections. My analysis was from a broader perspective, looking at all the communities, where I noticed a greater emphasis on tracks that are more contemporary and familiar in western culture- Johnny B Goode, Beethoven’s 5^th, and Melancholy Blues.

I think it just demonstrates that the same data can be manipulated in many different ways, and there are a multitude of conclusions that can be drawn. I’m not sure I agree with Jane’s observation that “the graph did not tell us anything about the people that made the choices”. From my perspective, the data gave us a cultural context to the people making the choices. I think the preference for contemporary songs performed in the English language is indicative of the cultural background of the participants. I do agree that more data is needed, and we would need to better categorize the data before making any strong conclusions.

#6

Amanda Botelho- Task 1: What’s in My Bag?

Task 1 – What’s In My Bag?

I thought Amanda’s contribution was very interesting and I chose it as my final link because it stands as a contrast to my own submission. Whereas I had very few items in my bag, Amanda’s is packed pretty full. She’s prepared for emergencies, and has tons of essentials to get her through the day. By comparison, I travel pretty light. One thing we had in common is that both of our bags included items that say something about where we’re from.

But the main thing I noticed is that while the contents of my bag are all for personal use, Amanda has items on hand for others. Blue tokens that she uses in the classroom, Kleenex for her niece and nephew, ID to access her school and classroom.

This got me thinking a bit about the nature of people and what you can tell from the contents they have on hand. I’m a pretty solitary person, but my impression from reading her submission is that Amanda is outgoing, social and always ready to help others. I also think it demonstrates how the contents of our bags are fluid- they change according to our life circumstances. For example, this week I’ll be visiting my own niece and nephew in Calgary, so I’ll be brining presents, books and yes, maybe even Kleenex.

Our original submissions were just a snapshot of a single point in time, but I think if we were to do this task every term, the contents would always tell a different story.

This was another interesting task to work through, as I surprised myself with some of the choices that I made.

As a general rule, my political beliefs hold that as a society we rely too heavily on incarceration as a rehabilitative method. I feel that the public tends to view jail as a mechanism for punishment, not rehabilitation and therefore attitudes tend to skew towards “locking them up and throwing away the key” whenever someone is deemed unfit or unsafe to participate in society. As such, I felt right from the get go that I would likely be releasing the majority of defendants and that public opinion would be very much against me.

When I actually began the exercise, I was a bit surprised at my own choices. When reading the descriptions I found myself leaning a bit more towards detaining the defendants, especially when the risk of committing a crime or violence was deemed to be high. In one case a defendant even pleaded that they would lose custody of their children if detained, but when I saw that their risk of violence was “high” I thought to myself “good! You SHOULD lose custody!” It was an odd feeling, a bit of a Stanford Prison Effect- when given just a little bit of power, I found myself wielding it with a sense of righteous authority.

But things got even stranger after that. As the exercise continued I found that I was spending less time deliberating. The three categories- Failed to Appear, Commit a Crime and Violence were color coded for each defendant with green meaning low risk, yellow medium and red high. I noticed that in all cases where Violence and Commit a Crime were red, I would default to incarceration. Soon, I wasn’t reading the contextual info at all, I was just zipping through making my judgements solely on color. This is concerning because I don’t know exactly what criteria was used to evaluate the defendants in the three categories. I just blindly trusted the information I was given. Upon finishing the exercise I did some further reflection and realized that in each case the defendants face was blurred out, yet the color of their skin was plainly visible. Could this also be a contributing factor on my decision making, even unconsciously?

As a whole, the exercise was a great opportunity to reflect on biases in decision making. Not just the inherent bias present in algorithms or hard coded into the justice system, but my own personal biases as well. I don’t believe bias can ever be fully removed from the decision making process, but acknowledging it is a good first step towards making choices in a fair and equitable manner.

I ran into an obstacle with this class right from the get go. Unfortunately the Palladio app didn’t work properly with my operating system. I wasn’t able to upload the .json file and therefore, was not able to access the data on my own. Thankfully our course instructor made himself available to assist me. We met on a Zoom call and over the course of about half an hour we worked through the data together.

Watching him manipulate the data points across the various groups of students, something became very clear almost immediately: there was a selection of three songs which were present in each group’s selections, and nearly every single student included these songs on their curated list of selections from the Golden Record.

The songs are:

1. Melancholy Blues,” performed by Louis Armstrong and his Hot Seven (Tack 14)
2. Beethoven, Fifth Symphony, First Movement, the Philharmonia Orchestra, Otto Klemperer, conductor. (Track 18)
3. “Johnny B. Goode,” written and performed by Chuck Berry. (Track 7)

While I did find it interesting that these three tracks were so ubiquitous among those selected, I can’t say I’m particularly surprised. While our class includes students from a variety of cultural backgrounds, it is a class taught in English at a prominent western institution. As such, I think it’s fairly safe to assume that most students in the class have had some exposure to western arts and culture, particularly music and film. From there, it follows that these three tracks would be selected by most students, as they represent the three pieces (out of 27 total) which are most recognizable in western culture. Johnny B Goode is the template for modern rock and role, Beethoven’s 5^th is arguably the most recognizable piece in the western canon and Melancholy Blues, while perhaps not as well known, is sung in English and in a familiar style of music.

I consider myself a pretty big music fan, and to be honest, when I first listened to the original 27 tracks on the record, I was a bit perplexed by some of the choices. There were very few songs that I recognized and some just didn’t sound very musical to my ear at all. But after looking at the data, I feel I have a better understanding of why many of these tracks were selected. Music is a universal experience, but as this exercise shows, it’s very tuned to cultural experiences. I think it was important for NASA to ensure that the record encompassed as much of the human experience as possible, and simply choosing 27 rock, jazz, country or classical songs wouldn’t have accomplished that.

As a predominately western audience, our class selected the songs that resonated most with them. But we don’t know how this data would look if it included selections from a more diverse selection of humans, such as someone who doesn’t speak English, or someone who has never heard a song on a radio. Since the intended audience for the Golden Record is definitively not human, it makes sense to include selections from as many periods and cultures as possible.