If you were hoping for a blog entry as entertaining as the first week, rest assured: this week’s blog will come nowhere near that. Not in a bad way or anything—just not as eventful.
I genuinely thought that I would be somewhat prepared. I mean, my AP Seminar class at school had already left me shaken—I had to read through about 20 fifty-page journal articles in just two weeks. And before any university students start complaining, I just want to say: this is a high school class. But maybe I should stop being overdramatic. I’ll be honest—now that I think about it, the papers I read weren’t actually all that jargon-filled. I mean, was it fun? No. Was it difficult? Not really, either.
Real research, on the other hand, was. My friend and I were tasked by our host to make a slideshow on how artificial intelligence (large language models, to be exact) could be used for electronic health record data analysis and knowledge extraction. On Monday morning, I told myself this day would be a piece of cake! The task was just to read a few papers! Oh boy, was I wrong.
I already told you about the confusing confusion matrices last week. This week, we got semantic vector quantization, schema ontology mapping, and… hypernymization (how do you even pronounce that)? So much easier. Sitting all day with a back injury from badminton didn’t help. Sleep deprivation didn’t either. This was when I truly learned how slow research could be. After many, many hours—although painful—I ended up getting through the grueling task of summarizing papers (with the help of ChatGPT to explain literally every second word in the article) for my slides.
I proudly showed my host my meticulously crafted work. She said it was excellent (yay!). I thought that was the end of it. I thought I was free. Then she instructed us to make another slideshow.
And that’s when I decided to put $25 on a Starbucks Gift Card and began buying a drink a day. The end.
Writing this throughout Thursday and Friday, I’ll admit: making these two slideshows was difficult, especially when, conceptually, you have no idea what’s going on. ChatGPT, however, does do a decent job of breaking down complex concepts. You just have to prompt it many, many times before you finally get the full picture. With this, I’ve realized I’ve indeed gained a lot of new knowledge. If you asked me to explain what I now know about large language models, ontologies, and knowledge graphs to myself three weeks ago, I’d probably also be confused. Research is definitely a slow process that requires hours and hours of work; I’ve asked peers in other placements, and they totally agree, too. Our projects aren’t even breaking the tip of the iceberg, either. I can’t imagine how difficult it is to eventually publish a paper. But I’ll be delusional—maybe if I set that as a goal, it’ll keep pushing me harder.
So, if you’re someone doing research for the first time reading this, rest assured: it is by no means an easy task. We only see the results on Google Scholar, but the time and effort behind publications is unfathomable. Those American college application decision accounts on Instagram that sometimes pop up on your feed set a standard that’s unrealistic. No one talks about the effort. But at the end of the day, you’re also learning huge amounts of information. Like, I don’t think school itself has taught me as many hard and soft skills in a semester as I’ve learned in the past three weeks. It’s also kind of a flex. As I’ve said before, the key is to pretend you’re doing something until you actually know what you’re doing. That’s how I’ve come to terms with the research process. Still, getting paid isn’t so bad.
Alexander (Alex) D., a student from Sherwood Park, is an avid sports fan and competitive badminton player—which takes up most of his time. He also enjoys playing piano and watching movies or TV, and channels his love for quantitative subjects, like math, into analyzing fantasy NBA leagues to win.