Newsletter #153
You may have noticed an absence this past Friday, but that’s only because I spent most of the week living the dullest - but busiest and most chaotic - version of The Neon Tempest you can imagine.
Let me explain.
There were sadly no mirror shades, drones, crazy sci-fi weaponry, or the more stylish accoutrement typically associated with cyberpunk; no cool holograms, flying cars, cyber-decks or the like. Instead, imagine what of the pandemic looked like: there were a lot of bored faces on screens listening to other talking heads go on and on about this, that, or the other; it just so happened those meetings revolved around artificial intelligence and its capabilities.
Most of you know that, in addition to writer and educator, I work as an AI analyst and consultant, and though I can’t talk about the client or other specifics of the work I do, I assure you that it sounds far more exciting than it is in reality. This isn’t to say that I don’t find the work fascinating. I work, in particular, on lingual models, so the granular aspects of what I do are right up my educational alley. At the same time, those things would likely be dull as dishwater to anyone who isn’t much of a word nerd. I do love the work, and hope I can discuss it more openly at some point in the future. In the meantime, I assure you that we aren’t in the Matrix, SkyNet isn’t self-aware, and - sadly - Rosie the Robot maids ain’t on our near horizon.
The discourse around AI could do with some leavening, I think, but I’m also seeing things from a privileged perspective; there have been and will continue to be ethical questions around models and how they’re built, what information they use, and how they acquire that information. The only solution to some of those current quandaries, really, is to train models using data produced specifically for the model or to train models using information in the public domain, or both. Alone, one of these is labor intensive and the other unnecessarily limiting. Either way, the process is both slower than one might like and faster than one might expect, and combining them both creates a plethora of unforeseen issues that have to be navigated and re-navigated. Often, in this type of work, the client doesn’t know what they want until they see what they don’t want, so negotiating expectations becomes a lot like revising a story or playing a tabletop game - asking ‘what if?,’ breaking things apart and jigsawing them back together, giving and taking, but ultimately working with what you’re given to make it better. Sometimes that process takes over, and bleeds from week to weekend, then back to week again before we know it, and we don’t have the time to do the things for which we ought to make time.
TL;DR - I had a long week and weekend and never let anyone tell you that playing Dungeons & Dragons doesn’t involve applicable life skills.
Have a good week, y’all! Don’t work too much, but take a little extra time for yourself if you do. You deserve it. I’ll see y’all Friday!
Wado!