By now, I have a fairly clear idea of my topic area–I want to explore the idea of digital identity from two directions: first, as a network of story we “write” into digital spaces; second, as a digital object far beyond our comprehension and control. The first approach suggests that we can analyze things like social media and pervasive data as texts which say something about us, the authors of that identity; the second suggests that we can only ever engage with those digital “selves” as outsiders if we want to truly understand them. My hope is that my project can help me to find and articulate a balance between the two, allowing us to empathize with our digital identities while bearing in mind that they are not us, that we are human and they are not, and that while this can be dangerous, it is also wonderful.

The particular uncanny affect of generative AI seems like a useful tool for encouraging such simultaneous relating and othering. Given my coding background, I’m very interested in using natural language processing tools to create extensions of the versions of ourselves we put out on the internet, perhaps as some kind of public-facing interactive art where users get to talk to their digital doppelgangers to see firsthand the similarities and differences between them. Their conversations and reflections would then provide a rich opportunity for narrative and discourse analysis.

However, to do this at the scale I envision could easily take years between coding, deployment, response gathering, and critical analysis. Instead of trying to pare it down and cram it into a single year, I think I’d like to do some smaller-scale creative exploration of the concepts tied into this long-term research.

I don’t have any clear ideas yet about what form this will take. Fiction-as-research, perhaps? I could tell a story about someone having such an encounter. But I do want to include some quantitative and computational work in my project. I almost wonder if I could explore something like what Margot Livesey calls “anti-fiction,” a method of storytelling which is “messier, more confusing, in other words more lifelike.” She uses Marcel Proust’s In Search of Lost Time as her prime example. While she does not apply the language of data collection to the (partially) fictional autobiography, I wonder if it might still stand. Proust uses a great deal of data and narrative from his own life, without tying himself to the particulars of what actually happened. He uses his own life as both data and story to leverage fiction and fact simultaneously and blur the two.

Maybe my project could do something similar (although hopefully in far fewer pages!). If I used some existing corpus of texts to represent a fictional character, I could then tell their story as a sort of dialogue between a human’s interpretation of the character (my writing and speculations, based on the source texts) and a computer’s (generated with an LLM I would fine-tune on the chosen corpus). While it isn’t quite the same as having a real person dialogue with their digital self, it still provides enough juxtaposition to explore the questions I’ve raised about identity and empathy.

If possible, this would be an interactive and/or multimodal story, leveraging a format that adds an integral layer of meaning to the piece. That said, without knowing anything about the story itself yet, I’m very open to all the narrative and creative possibilities within the scope of the larger concept. I hope to do a little brainstorming before the group supervision meeting, which should help me come up with some more specific ideas.

Livesey, Margot. “How to Tell a True Story.” The Hidden Machinery: Essays on Writing, Tin House Books, 2017.

Week 5: Initial Project Idea / Knowledge Integration & Project Planning by is licensed under a