Have you ever had a sudden flash of inspiration right before falling asleep? I had one such moment recently for a great visualization about word graphs. It didn't actually solve my problem but lead to a rabbit hole to create a puzzle game called Wordspose (with a full demo). Continue reading →
Is AI de-skilling us, or leading to a new age of creativity? Both are discussed at the recent Creativity & Cognition conference. I provide a timeline of my text visualizations, and ways that I use LLMs to aid creative approaches. Continue reading →
Perceptual scalability is the challenge of preserving meaningful visual detail as datasets grow. Unlike summary charts, expert analysis often depends on seeing anomalies and relationships. Our latest research paper is for perceptual scalability for massive graphs with millions of connections. We combine many visual techniques including hierarchical grouping, clustering, edge bundling,…
In 2023, a group of us ran a workshop to experiment with novel visualizations for humanities. The results are now published in the book Seeing Interaction Design (Springer). Further, I revisit one of those concepts - using AI to generate connections on the fly in a visualization to aid rapid, exploratory hypothesis-building. Then, Claude and I build it: illustrating how quickly speculative AI…
Can LLMs be creative collaborators for visualization design? I use sketches sketches, images, and text prompts to iteratively generate and refine an expressive text layout for Alice in Wonderland. I believe that multimodal AI can aid democratization of design, enabling both visual and non-visual thinkers to prototype, refine, and communicate novel visualization ideas collaboratively. Continue…
How can we visualize contemporary phrases? Contemporary art suggests possibilities, but there's a need for additional context. With AI, I find phrases, do semantic grouping, create example sentences, and encode connotative colors, into a composition. This little AI-assisted design project aid reconsideration of design rules regarding layout and color. Continue reading →
LLM's help me animate the text in ways that it perhaps should not be animated. Why? It's about understanding breadth of design possibilities. Continue reading →
After failing to create an effective scrolling news ticker, the author explores kinetic typography's potential for data visualization. "Dreamy Alice," is an interactive version of Alice in Wonderland where text movement semantically encodes plot emotions, demonstrating how animation can dynamically enhance large-scale reading. Continue reading →
A Washington Post analysis of earnings calls visualizes rising CEO concerns by sloping keyword changes in context. For another “text on paths” approach, I explore placing quotations on circular rings, enabled by rapid LLM-assisted visualization design. Continue reading →
An LLM-driven exploration of visualizing 800 National Film Registry movies. After failed genre and text-embedding approaches, the author uses LLMs to generate nuanced attributes, cluster films, and design an interactive visualization that reveals meaningful similarities without standard categories, reflecting on strengths and limits of “Team LLM” design iteration. Continue reading →