No ratings yet.

Generative AI is everywhere these days. From text to speech, from images to videos, from music to art, there is hardly any domain that has not been touched by the undeniable power of artificial intelligence. Everyone knows about this technology, or the people that are reading this blog post all know about it at least. It is likely that we all use it daily to help us with our daily tasks. Whether to plan out a trip, find a recipe, or kickstart our creative work, it is already becoming increasingly hard to imagine our daily routines without it.

But do you remember the first time you really saw what generative AI could do? This is a question that popped into my head when thinking about a topic to write about for this blog. For me, it was back in 2020, a time when I was aware of the prowess of AI, but in which I had not heard the term “Generative AI” as it is being used today.

So, in 2020, a fake Travis Scott song was “released” that was produced by an AI model. Over the course of the years, I have become rather acquainted with the musical library of the Houston rapper, so I could obviously hear that it was not him. However, if someone had told me that it was made by an upcoming rap artist, I would have probably believed it at first.

Especially the lyrics of the song by TravisBott (yes, this is what the creative minds of the project call him) stood out. In fact, they really seem to dwell in the realm of absurdity:

“She got the crew on top of my chain (It’s lit); Wasted in the street like a pain (Straight up); You see the diamonds in the light of chain; They say I f*cked the bad b*tch like I’m rain; I was the b*tch on the plane (Straight up).”

If you are a trap/rap connoisseur, these lyrics just seem a bit off to you. Most people, however, will likely not tell the difference fake and real lyrics. Because indeed, this song also mirrors the prevalent contemporary rap narrative of using swear words and talking about money and sexual desires. But if you pay attention, the sentences just do not really make sense.

So, considering that most people do not really know what “normal” rap lyrics look like, and considering that this is made by AI, I was rather astonished when I read this news and listened to the song. I was specifically amazed by how authentic and realistic the production of the song was, from the melody to the text. Apparently, Space150, the company behind the song, started out feeding real-life Travis Scott lyrics into a language model for two weeks until it began creating its own rhymes. The company mentioned that the lyrics were initially foo-obsessed, but that they smarter as time went on. Then, they used additional neural network programs to create melodies and percussion arrangements to accompany the generated lyrics. Back then, these “AI terms” were quite new to me but I did realize how clever it was.

It was the moment where I first saw an application of how far generative AI had come already, and how much more it could achieve. As I write this blog, I meander the internet to watch the video clip one more time and find the news articles that talked about this musical production. I now stumble upon multiple other AI-generated rap songs that have been created recently. So, if you all do not mind, I will give them a listen to see if lyrics make a bit more sense than the ones in the TravisBott production.

No ratings yet.

Many of this blog post’s readers have likely discovered that ChatGPT has become an indispensable companion in their educational journey over the past year (or so). As students following a course in Information Strategy and enthusiasts of emerging technologies, the allure of generative AI has naturally drawn us towards this incredible tool. ChatGPT has consistently proven its worth by assisting us in a multitude of ways.

The relationships we’ve formed with ChatGPT vary widely in nature. Some may admit to having leaned too heavily on this resource, relying on it to write assignments, essentially seeking to delegate the entire workload. Others maintain a healthier, more symbiotic rapport, using ChatGPT to extract concise summaries from voluminous texts, enabling them to better manage their study time while still engaging with course materials independently. Regardless of the approach, ChatGPT is undeniably poised to revolutionize education and will likely be embraced further as the new study companion.

Throughout the past year, which is when ChatGPT steadily started occupying student’s  laptop screens, I’ve also had the privilege of acquainting myself with its capabilities. It guided me during my thesis by displaying information on what a typical methods section should includea methods section for my thesis, providing valuable inspiration. During my semester abroad, it efficiently generated travel itineraries for trips through South Africa and Namibia during study breaks. While I recognized its potential as a lifelong productivity companion, I had yet to experience that jaw-dropping moment that would truly distinguish this tool.

However, just this week, my admiration of ChatGPT took a deeper turn, as it revealed its remarkable intelligence to me. As I grappled with a statistics assignment in a research methods course, ChatGPT stepped in to offer comprehensive assistance. Despite having some proficiency in statistics, until now I often struggled with the intricate intricacies of R-Studio, wrestling with endless errors when attempting to produce specific statistical outputs during my BSc. Quite often, I knew what an exercise wanted me to do, yet I often did not know the specific codes to run some complicated multi-layered functions.  At that time, securing personal guidance from instructors proved challenging, given the shift to online learning, which resulted in quite awkward situations where you had to share your screen to ask the teacher for help. Furthermore, these courses were often perceived to be quite complicated, so you often found yourself competing with other students to get a teacher to help you with your problem.

This is where, this week, ChatGPT proved to be an invaluable companion. The more I specified my queries and the errors I encountered, the more accurate and detailed ChatGPT’s guidance became. It not only provided solutions but also explained the underlying issues and offered step-by-step instructions with interpretations. In fact, it even displayed what such solutions would look like in your R-Studio console, and provided a button on this display to copy the code, which makes the output even more comprehensive.

I’ve come to appreciate the tremendous productivity ChatGPT brings to my learning process in statistical programming. It bridges the gap between my uncertainties and effective solutions, saving me from the frustration of dealing with endless errors and vague or irrelevant responses from traditional search engines. It is, in essence, my ideal companion for tackling complex issues and streamlining the learning process.

In conclusion, ChatGPT has undoubtedly demonstrated its value in education for everyone. Yet, my recent experience with its unparalleled ability to dissect and resolve intricate, case-specific statistical challenges left me truly amazed. ChatGPT has become more than just a study partner; it’s the companion that can decode the nuances of your code and illuminate the path to comprehending and passing my research methods course.