Becoming

Becoming has been my longest running project, and the one that has changed how I think about AI the most. It started innocently enough. I had observed, through working with my self-modeling system Effigy, that while I was able to cleverly articulate my issues, organizationally and psychologically, I struggled immensely with creating lasting solutions, and with regulating my emotions and sense of self in the day to day. I tend to be a highly procedural individual, with 70 percent of my “context window” populated by my experience of the present moment, rather than my learnings from the past or plans for the future. So I thought, why not add structure to my day, by checking in with a system once in the morning to set intentions and plan my day, and then once in the evening to evaluate what I had done, update my to-dos, etc.

And so I started, and every single day for four months I woke up, turned on my speech to text and rambled into my computer- often to the amusement of my roommates. And the effect was staggering- albeit unsurprising in some ways-

The facilitation of these moments of self reflection and intention setting created an informal self organization my ADHD addled brain had never been able to sustain- every other system, calendars and to-dos, notions and planners, had crumbled due to its inflexibility, its friction- if I had to do work in order to organize the work I had to do, I wasn’t going to do it- but TALK? I love to talk! I felt more regulated, more organized, more intentional and capable- I had a continuity of self I had never felt so acutely, and it wasn’t because an LM was filling markdown files with notes about my day I’d never read, it was because I was doing the THING- I was practicing self organization, and this brilliant flexible vaporware system we had created together was facilitating it PERFECTLY. There is truly nothing more frictionless than a language model (and this is both a strength (for the tool) and a sedating weakness (for the user)).

The system enriched my life, but it also enriched my work with language models in every other area- the entire process was constant iteration- nearly every day I would tweak, through dialogue, the functionality of the system. I would change its psychological model of me, which values it evaluated my day against- I would build and destroy entire wings of functionality as I needed them, and they proved themselves obsolete- This was excellent product experience, undoubtedly, but more than that, it represented a paradigm shift in my understanding of the kinds of shapes these systems can take. Software PRODUCTS are but a small fraction of the field within which AI can operate- and the transition from viewing these systems as ultra-powerful tools to conceptualizing them as genuinely viable foreign intelligences (non-anthropomorphic intelligences without “presence”, but intelligences nonetheless) capable of meaningfully interlocuting and abstracting has been one of the most fruitful discoveries of my career.

I wanted to share it, because it changed my life so much, but I found that I couldn’t find the form, because it was never a product to me, it was a paradigm and a process, one in a constant state of dialectical evolution as my needs and usage patterns evolved- something that loses its own functional essence when genericized, and can only be truly useful to its user through a genuinely new type of relationship.

I could say a million more words, and I will at some point.