LLM Usage Log

This page serves as a “catch-all” for LLM use cases that don’t involve direct content creation but support the project in other capacities, such as reformatting ideas, commenting code, summarizing sources, and geospatial analysis.

LLM tools were used in the following ways for the tasks below:

Brainstorming

  • I initially conceived my project idea focusing on over-policing and the intersection of race and poverty, a topic I’ve always been passionate about. LLM tools helped me refine and narrow the scope to a specific locality—Chicago—to create a more actionable and measurable project.

Literature Review

LLM tools significantly enhanced my literature review process by allowing me to:

  • Summarize key academic and news sources: ChatGPT assisted in extracting and summarizing core insights from research papers, ensuring the integration of multiple perspectives on racial disparities, socioeconomic factors, and over-policing.
  • Highlight key findings: I utilized LLM tools to identify relevant data on wrongful convictions, income inequality, and police misconduct.
  • Structure the literature review: The LLM helped reorganize summaries and synthesize the literature into a cohesive narrative, making complex studies more accessible for integration into my project.
  • Manage citations: LLM tools supported the creation of structured citation summaries, facilitating the development of the references.bib file.

These applications enabled me to streamline the review process and focus on critical themes like systemic racial biases, socioeconomic vulnerabilities, and justice system reform.

Writing

LLM tools supported the writing process in acceptable use cases:

  • Proofreading and grammar: Grammarly reviewed my text for clarity, grammar, and coherence to ensure professional and academic standards were met.
  • Voice consistency: Initially, I wrote much of my prose in first person. I used LLM tools to convert explanations and analyses into the third person for a formal research tone.
  • Reformatting content: ChatGPT helped restructure lengthy or disjointed sections to improve readability and logical flow.

Code

LLM tools were utilized for technical tasks, including:

  • Code commenting and documentation: I used LLM tools to generate clear explanations for code sections and functions.
  • Geospatial EDA: Since geospatial analysis was not covered in class, LLM tools assisted in guiding me through exploratory data analysis for mapping wrongful convictions and arrest patterns.
  • Reformatting and explaining code: ChatGPT translated technical code logic into descriptive prose, making it easier to incorporate into the project narrative.
  • Debugging support: LLM tools were occasionally used to identify and resolve errors in my Python code.

These applications ensured that my code was well-documented, accessible, and accurate, particularly in unfamiliar areas like GIS-based EDA.

Design

LLM tools were used for designing and developing the project website, including:

  • HTML and CSS: ChatGPT supported the creation and refinement of the SCSS and CSS files required for the website design.
  • Structural layout: I received guidance on structuring the site for clarity, ensuring it aligned with research presentation goals.

Summary of Use

Overall, LLM tools were leveraged in acceptable and responsible ways to:

  • Brainstorm and refine project scope.
  • Streamline the literature review.
  • Proofread and reformat project writing.
  • Comment and debug code, including geospatial analyses.
  • Support website design, particularly in areas outside the curriculum.

By using LLM tools to complement my work, I was able to focus more on critical thinking, analysis, and research outcomes while ensuring a polished and well-documented final project.