Mixing Movement and Machine
I found this article really interesting because it emphasized shifting the question from what the machine can do to what I want the machine to do, which isn’t a perspective that I looked from before. I realized I often focus just on experimenting with new tools, but sometimes I forget to think about whether they actually help me with my creativity or my project itself. I found the concept of technological restraint very interesting, because it reminded me that sometimes not trying to make it perfect can actually make the work better in the sense that the outcome is unexpected. I can also relate to the idea between novelty and meaning. For example, in my own projects, I’ve tried really bold effects just to realize they distracted from the main narrative that I wanted to tell.
Humans of AI
I really liked how the article focuses on the hidden labor and histories behind AI systems, which I have never really paid attention to before. Such as the photographers, annotators, and data sources cannot be seen, yet they are the very fundamental structure of what “machine vision” is. I also found the concept of “restoring human context” interesting, like when they reclaim image titles and stories suppressed in datasets. It is because it challenges the idea of machines as purely objective which now that I think of it isn’t very true. I now think about how systems that seem neutral are actually built on choices (like past data) about what gets left out or used.
Open Sourcing the Origin Stories
Reading what Ellen wrote made me more aware of how much is hidden behind the models that I use daily, and lowkey take for granted. I still have questions about the gaps in provenance, especially for models like Face-Api, which is kinda creepy to me? Like as in who exactly collected the data, how consent was handled, and whether the images were truly meant for these uses. I think the biography could also include an explicit section on ethics and consent and not just technical sources (because i feel like we don’t talk about it enough and sometimes users aren’t aware of what they’re consenting to), so that people like me studying AI know how datasets are gathered. Understanding provenance also changes how I take on my creative projects, because it makes me think not only about what a model can do, and what I want it to do for me, but about whether using this tool will be a threat or not.
P5.js Sketch link: https://editor.p5js.org/Anna_Tang/sketches/AA41nB8qX
Reflection: I found this project both fun and a little tricky. I struggled at first to figure out how to pull out the right key points from the BodyPose model, but once I understood the naming, it became much easier. I was proud that I got different elements, like the circle following my nose and the background changing with my wrist distance, to work smoothly. I also learned how to test with my own body movement and adjust the mapping so that the interaction feels natural.