In January 2024, ChatGPT Plus, Microsoft Copilot Pro, Google Gemini Advanced, and Claude Pro were four of the most popular Generative AI services available to the public. Each had subscription plans for around $20.00 a month. Meanwhile, at [LIBRARY], library student workers were creating basic metadata for scanned electronic theses and dissertations (ETDs) from the 1920s-1990s. Due to time constraints, these historical ETDs only had basic metadata fields such as title, author, and date. How could the Oklahoma State University library add valuable metadata with limited staff and time? To solve this problem, [LIBRARY] student workers began testing these four, subscription-based Generative AI services to see if they could create reliable, accurate abstracts and keywords. To do this, student workers learned how to create accurate prompts to gather the best possible metadata, and then they learned how to evaluate the results with a rubric provided by their supervisor. This presentation will demonstrate how student workers used Generative AI to create abstracts and keywords for electronic dissertations and theses for an institutional depository. Then it will compare these four Generative AI options, evaluate their performance for creating metadata, and envision the future of generative AI and metadata.
