Before this week of class, I had no idea what the exact purpose of an LLM (Large Language Model) was. I also didn’t know it was the main backbone for generating answers in key AI apps, like ChatGPT and Copilot. The most interesting fact I found out was how it actually operates. The input text is split into smaller units called tokens, which can be words, subwords, or characters. Then those tokens are turned into an answer that represents semantic meaning. Also, the video where it visually showed the breakdown of the process was very interesting and something I would’ve never guessed went down.

One perspective that has changed about AI is the use of environmental resources required to operate it. My main issue is the energy and carbon emissions. As we know, training and running AI models require large -scale computing, increasing greenhouse gas emissions. This is a fact that has been talked about in almost all of my classes, which means it is a pressing matter for everyone constantly using AI and those who are being affected by this.

Going off of that statement, one practical application that I plan to use in my academic work is to encourage responsible AI usage. I will limit unnecessary high-energy tasks, such as large-scale image or video generation, and promote energy-aware AI design.

3Blue1Brown (2024, April 1). Transformers, the tech behind LLMs. Dftba Records. Transformers, the tech behind LLMs | Deep Learning Chapter 5