Rebuilding Memories with Generative AI
After years of working with generative AI tools, I have learned to see the similarities in how my thoughts, dreams, and memories also work. I tend to dream in a duotone spectrum, with detail that fades from the center and visualizations that are almost generated in real-time when I focus on any detail. This process is not dissimilar to how a diffusion model recreates images! Identical to AI, the image has not been stored in my mind, but the concept of it has. The memory is reconstructed not so much by a visual imprint but rather by a pattern formula of probabilities that approximates reality without truly being an identical depiction.
I believe Generative AI is opening many creative ways of approaching content, and one of my favorite expressions is attempting to “photograph” situations that do not exist anymore. AI allows me to describe the memory as I remember it and, with the subtle use of prompts and parameters, arrive at a very close approximation of how that memory looks to me… independent of how it was. If being realistic is all I wanted, I might peruse scans of old Ektachrome photos in our collective internet repository until finding the closest one, but that is not my intention… This growing collection will reflect a very personal (and hopefully at times entertaining) perspective on the memories of the culture I grew from.
Conclusion
Artificial Intelligence is not only a tool for accelerating creative output but also a model for understanding ourselves better: From figuring out what being “intelligent” or “self-aware” means, to decoding how we humans store information and later reconstruct when remembering things. For me, it has opened a door for a process that I call “dreaming while awake”.
0 Comments