# GAICA Project Summaries

---

**Project Name:** Generative Music Model

In this project, a GAICA developed a generative model to create musical notes for **MUSIC225**, focusing on musical forms, especially in classical music. The project aims to explore the capabilities of AI in understanding and generating musical forms. Utilizing TensorFlow and Keras, the model is designed to train on music data and generate new compositions. Originally using a database of monophonic folk songs converted to text, the GAICA scaled the project to take in midi files, allowing the model to produce polyphonic melodies.

**Technology Stack:** TensorFlow, Keras, Python, custom Python library for transformer models.

**GAICAs:** Gabe Mehra

---

**Project Name:** Overhead CV Model

In this project, an AI model was developed to process overhead videos of a table, recognizing and cataloging features such as card color and orientation using Camera Vision. This model was initially trained on recorded footage and successfully extended to handle live video streams. Future enhancements include interactive elements like triggering lights and generating sounds based on the visual input. This project serves as a prototype demonstrating how AI can be integrated into physical spaces to enhance classrooms, workshops, and lectures in real-time with novel and humanistic applications.

**Technology Stack:** OpenCV, YOLOv7 (with Pytorch and Tensorflow), LabelImg, Scikit Learn, MediaPipe, Numpy

**GAICAs:** Indra Islas-Luz



---

**Project Name:** Colab Singles

This project contributed to "**COMPLIT 200**: Computing Fantasy" by developing code blocks, termed "singles," which enable students to generate illustrated folktales. The course is experimental, focusing on the intersection of literary theory, history, and computer-assisted creation using generative AI tools for storytelling. The GAICAs’ work aimed to enable students to engage creatively with the course's objective of crafting an illustrated volume of AI-generated folktales. This aligns with the course's emphasis on the history and theory of generative storytelling, from structuralist theories to practical applications in narrative construction.

**Technology Stack:** Python, Google Colaboratory, OpenAI API, Gemini API

**GAICAs:** Gonzalo Pelenur, Josefina Biernacki, Jōsh Mysore

---

**Project Name:** Colab “Machines”

This project involved the creation of mainly two types of psuedo-machines, machines being "an apparatus using or applying mechanical power and having several parts, each with a definite function and together performing a particular task." These machines not only utilize code, but apply "rhetorical" or "intellectual" power to perform a function.

One type, a fairytale generator for **COMPLIT200**, utilized the stacking of “singles” from another porject, as well as non-ai functions and inputs (such as a new “color” for Little Red Riding Hood’s cloak) to generate tales in structured ways.

The second type of project involved a recursive machine designed for **EMR162**. This machine processes images by placing them into a recursive feedback loop. It starts with a "revised prompt" used by Dall-E to generate an image, which is then reintroduced into the loop along with a vision description of the resulting image. This process creates a sequence of opposites and extremes, uncovering how AI systems describe and create images. Such insights reveal details about the training and cognitive processes of these AI models.

**Technology Stack:** Python, Google Colaboratory, OpenAI API

**GAICAs:** Gonzalo Pelenur, Josefina Biernacki

---

**Project Name:** Tutor Assistant

This project developed a chatbot assistant for **PHIL16** using the OpenAI API. The chatbot employs a Socratic method to assist students in brainstorming and developing ideas for their papers. Initially deployed in OpenAI's playground, there are plans to integrate this tool with Slack, enhancing its accessibility and utility for students.

**Technology Stack:** JavaScript, Node.js, Slack, OpenAI API

**GAICAs:** Josefina Biernacki

---

**Project Name:** STEM Slackbot

This project developed a chatbot assistant for **PHYSICS15b** in collaboration with the course professor, using the OpenAI API. The chatbot assists students in understanding complex physics concepts and solving problems. After initial development and testing, the Slackbot was fully integrated and deployed on Slack for student use following spring break, enhancing its accessibility and practical utility in educational settings.

**Technology Stack:** JavaScript, Node.js, Slack, OpenAI API

**GAICAs:** Suvin Sundararajan

---

**Project Name:** AI Assignment Pedagogy

This [project](https://hackmd.io/@ll-23-24/rkb6gwJkC/%2FDA1U2pI6S4-mpFL8fDQdwQ) theoretically explores the integration of AI into educational assignments, particularly within the humanities. It aims to both enhance existing assignments and develop new types that mimic traditional functions, thereby improving educational outcomes across various disciplines. By formally assessing assignments (like expository essays and creative writing projects), the project quantifies educational processes and seeks to fill gaps in knowledge transfer with AI-driven solutions. These hypothetically include real-time effectiveness assessments and iterative improvements using AI tools such as automated feedback generators and tutoring systems.

**Technology Stack:** n/a

**GAICAs:** Jōsh Mysore

---

**Project Name:** Generative Web Development

This project involved the development of a website featuring generative backends applied to the **COMPLIT280X** course. After students in the class created "generators" in Google Colaboratory, a GAICA developed AI-based tools that converted text files of these poems into CSS, HTML, and PDF formats. These digital assets, along with other generated content, were then integrated into a comprehensive website for public display. This initiative showcases how generative AI can enhance web development and digital artistry in educational contexts.

**Technology Stack:** CSS, HTML, JavaScript, Google Colaboratory

**GAICAs:** Gonzalo Pelenur

---

**Project Name:** Live Generative Transformations

This project utilized live Stable Diffusion transformations to convert images and videos captured on a green screen into generated visuals. Initially deployed in **SLAVIC121**, the technology allowed students to visually explore the mathematics of dance movements in innovative ways. The course focused on the spatial imaging and language of dance choreography, particularly how movements are communicated and preserved. This application of AI not only prototypes future educational tools but also opens new methodologies for analyzing and codifying dance movements through advanced visual processing.

**Technology Stack:** Stable Diffusion, green screen technology, video processing software

**GAICAs:** Gonzalo Pelenur
**Learning Lab Undergraduate Fellows (LLUFs):** Elisa Weyer, Joel Kizito


---