Introduction

My name is Sophie Kersun, and I am a senior majoring in finance with a minor in international business.

After graduation, I will be working in internal strategy at a financial services institution, but longer term, I am interested in consulting for nonprofits.

I chose to take this class out of genuine interest and as an opportunity to push myself outside of my comfort zone. I would not consider myself the most technologically savvy person, and the world of technology has always felt somewhat foreign to me.

While I have used AI extensively over the past few years, I am interested in pushing myself to learn more about: 1) how to use AI to create and produce more impactful work; 2) the technology itself and how it works; 3) how to use AI to do more, both in my day-to-day life now and in my career after graduation; and 4) how to develop habits that allow me to continuously learn and adapt as AI continues to evolve.

Reflection

This week, I used ChatGPT for several very different purposes. I learned more about how to improve its value as well as some limitations.

First, I used AI to help me with this assignment. I asked ChatGPT to recommend a platform for my digital reflection log and, based on what I wanted to create, it suggested Google Sites. I then used it to guide me through setting up and organizing the site. However, the initial instructions it provided did not accurately match the navigation options I was seeing in Google Sites. When I pointed out the discrepancy and began providing screenshots of my actual interface, its guidance became much more accurate.

This experience highlighted an important limitation of using AI for interface-based guidance. ChatGPT could explain how Google Sites should generally work, but that did not necessarily translate to the exact version of the platform in front of me. I saw, as I've seen before, how much the quality of an AI response depends on the information I give it. A screenshot provided context that my written prompts could not, allowing the model to respond to what I was actually seeing rather than what it assumed I was seeing. It also reminded me that a detailed AI response can still be incorrect.

I encountered a different limitation when I asked ChatGPT to reflect on my own AI usage by answering the question, "What has my use of ChatGPT over the past week taught me about AI and how I interact with it?" It had the ability to identify patterns across the various ways I had used the platform and draw broader conclusions from them. At the same time, it did not adhere particularly well to the timeframe in my question. Despite asking specifically about the past week, it referenced interactions from months earlier. I have noticed this in other interactions as well: AI does not always interpret references to time as precisely as I expect it to.

I also used ChatGPT in a much more practical situation this week when I accidentally got Sharpie on a white couch cushion. Instead of one question producing one answer, I used AI throughout the cleaning process. I shared photos as the stain changed, explained which cleaning products I had available, and asked follow-up questions before trying different approaches. This experience gave me a better understanding of the value of using AI as an interactive, ongoing problem-solving tool, where each new piece of information can inform and improve the next response.