DariefLet's talk

Coding

Argumentor

Using AI to help debaters research and prepare for debates

In October 2023, I developed Argumentor, a web application designed to support the UCSD Speech and Debate team, of which I am a member. Argumentor was a precursor to tools like Microsoft Copilot, providing an innovative way to integrate GPT-3.5 with real-time internet access. By leveraging advanced AI capabilities, I created a system that could format arguments consistently, generate logical reasoning, and include trustworthy sources to back up claims.

Tools Used

Next.js
OpenAI API
Prompt Engineering
Search API
Supabase
LangChain

Goal

Help debaters research and prepare for debates faster and more efficiently

Methodology

  1. Development and Implementation

    Built Argumentor using Next.js to create a sleek and user-friendly web application interface. Integrated GPT-3.5 with internet access through LangChain and the Search API to ensure that generated arguments were supported by reliable and current sources. Designed prompts to ensure consistency in the formatting and logical flow of arguments, tailored to debate-specific standards.

  2. Core Functionality

    Allowed users to input a topic and select their stance. Generated structured arguments supported by relevant sources, using a combination of logical reasoning and real-time search results. Enabled follow-up questions, allowing users to refine their arguments or gather additional information quickly.

  3. Team Adoption and Usage

    Launched in October 2023, Argumentor quickly became a valuable resource for the debate team. Several members used it extensively to speed up research and familiarize themselves with unfamiliar topics before competitions. By February 2024, the majority of the team transitioned to using Microsoft Copilot for similar purposes, having experienced the benefits of Argumentor's approach to AI-assisted research.

Results

Argumentor significantly enhanced the debate team's research process, making it faster and more reliable. Members were able to prepare well-structured arguments in less time, boosting their preparedness and performance in debates. The tool's ability to provide trustworthy sources and answer follow-up questions proved invaluable for tackling complex topics.