14. Weaving AI to life: Physically visualizing generative AI responses to data librarian reference questions

Ada Lovelace, often heralded as the first computer programmer, saw the potential in Charles Babbage’s Analytical Engine by comparing its possibilities beyond number-crunching with that of the mechanical loom, which used mechanical binary punchcards to weave intricately patterned fabric. Inspired by this revelation to conduct a contemporary reimagining, I propose a poster that investigates the potential and pitfalls of generative AI and then visualizes the results through a woven data physicalization.

As generative AI rapidly advances, there continues to be a need to stay abreast in its applications and implications. In particular, I plan to investigate how well different generative AI products fare in answering various reference questions that a data librarian may be asked; how deep can a certain generative AI tool be asked to go, before it starts to hallucinate or lead to results not fitting the prompt? How well can it unearth datasets that meet a researcher’s very specific needs, when data can be embedded in a variety of places, some behind paywalls or low SEO?

In homage to the long lineage of technology, the results of my investigation will be visualized by creating a textile-based data physicalization, where an accurate result will be woven with one colour and an inaccurate result in another. Each woven row will represent the results of a different reference question and AI product combination; the surrounding area of the poster will contain details of the question and product combination. Thus, technology again programs the textile’s pattern.