1. Read Open Sourcing the Origin Stories: The ml5.js Model and Data Provenance Project by Ellen Nickles and reflect on the the following questions:
   • What questions do you still have about the model and the associated data? Are there elements you would propose including in the biography?
   • How does understanding the provenance of the model and its data inform your creative process?

Even with the research done on ml5.js models, I still have lingering questions about the hidden decisions that shaped their creation. For instance, what demographic distributions exist in the training data race, gender, age, geography and how do these affect model accuracy for different groups? Were subjects aware that their images or voices would be used in machine learning, and what does consent look like in these contexts? Provenance often stops at the first layer, but many datasets are themselves derived from others, creating cascades of ownership and responsibility that are easy to lose track of. To strengthen the “biography” framework, I would include demographic composition and known biases, clear notes about licensing and consent, and an outline of known risks and limitations. It would also help to show whether a model is static or periodically retrained, since origins are only one part of its life cycle.

Understanding the provenance of a model informs my creative process by reminding me that these tools are not neutral but shaped by human choices. When I know the bias in a dataset, I can adjust my design decisions and avoid applying a model in contexts where it may cause harm, such as using a body-tracking model on groups it wasn’t trained to recognize. Provenance also inspires me to bring transparency into my own projects acknowledging the “ghosts in the data” as part of the artistic narrative rather than ignoring them. Sometimes this means rejecting a model if its origins are too opaque; other times it means weaving the dataset’s history into the work itself. Overall, provenance shifts my perspective from asking only “What can this model do?” to also asking “Whose values does it carry, and how should I use it responsibly in my creative practice?”

Link to week 3 project :https://editor.p5js.org/Sitong_Zhou_Silvia/full/vglIU_gmn

Demo Video:

Week3 demo video.mov

This is the model i choose to play with.

My plan：

Key points Particles: Particles cluster around human body keypoints, and when detection is lost, they scatter and wander.

Hand trajectories Neon trails: Each wrist leaves a glowing trail behind it.

Pose-triggered backgrounds:

• Hands Up (both hands raised above the shoulders): Background switches to a pulsing ripple.
• T-Pose (arms extended horizontally): Background switches to stripes.
• Hands Together (palms pressed or fists touching): Background triggers a one-time energy pulse.

Here is what I’ve add for background.