HELLO FELLOW BOTS and SANDBAGS!
And another two years go by. I must have dozed off; I don't think that much has happened. I came back to Minnesota and then left to California again. Then came back to Minnesota again where I now sit.
The ranch still stood. Much of the trials and tribulations of the first year were pretty diminished. Power seemed to more or less work (getting the solar updated was the only hiccup). Weather was overall milder. No trucks stuck in the mud (just a four-wheeler in branches). No slithering horse thieves in the night. I had gone to understand something and I came back learning that half my motive force comes from arriving at truth that I first denied. But all's well that ends well.
Buster Jr. is doing well as good. Although I've started calling him Mr. Glass, because it seems like every week he re-injures himself in some way. He is either slowly rolling at 5 mph or jumping the curb at 50. I have theorized it is due to his smaller than average feet below his steroidal tongue of a body. But in any case I'm hoping he continues to mellow out to the point where he can actually let himself heal from his bumps.
On the academic front, PhD continues to limp along. A variety of options are on the table for paths forward, but all of them seem to put me at a cross-roads I'm not ready for. What has become clear is that if I want to finish this thing in the next few years, there is no way I can continue in corporate, no matter how cushy the womb is or how funny the resulting jokes are.
We all have a destiny or at least a place we'll settle for. Sometimes its high fate and sometimes its low comedy. With high fate you get a chance to do what you were meant to do with your life, but with low comedy you get to gaze upon the face of eternity. Which would you choose?
-------------------------------------------------------------------------------------
Had an idea 2 years ago that I have started to try and flush out these last couple months. The Mirror would be a trio of AIs that, given the right quality/quantity of data, could learn to "fit" a system or a context. Essentially the Mirror layers existing "artificial" "intelligence" "technologies" that could be built to work together to replicate the knowledge in a system, which could be a person, an organization, or really any association of things (think next level PowerPoint SmartArt). Structurally it would be composed of three layers including Implicit Knowledge (Large Language Model - LLMs), Descriptive Knowledge (Knowledge Graphs/Ontology), and Experimental Knowledge (Causal Graphs).
An LLM's main limitation is context - it can only utilize the context that the prompt provides. The LLM makes use of the fact that a question has more information in it than the answer. But how can one ask a question that encodes all the relevant context? And conversely, how can we get LLMs to ask good questions? By chaining LLMs to more explicit knowledge structures such as ontologies and causal graphs, the context can be generated by, or represented for, the LLM.
Two uses of such a "technology" I think could be to conduct system-perturbing experiments on the information structures or integrate multiple systems/intelligence together, both of which could potentially be done through a natural language interface, but I admit I'm not exactly sure how these could work. However, I think something like the Mirror will inevitably emerge, as the accolytes of such a thing already can be found across the Internet prophesying the arrival of their demon king through the value of "knowledge graphs" for "innovation". If something like the Mirror does come into being, crossing my fingers for some completely inadvertent use of it by the common man. Otherwise I'm sure it will just be another freedom-reducing yoke; one of those AI technologies that languish
in the dark for 20 years to be discovered again for the next cycle of
Tower-Of-Babel-Building by the deranged engineers.
My final thought on the matter for the blog is that there is an interesting dimension to the Mirror in that it folds into a post-modern understanding
of subject vs object based on an idea of context (in this case graphical knowledge). What if all that determines a subject is that it has more context than an object? It's Baudrillard's "are you watching TV or is the TV watching you?" It's a rejoinder to the question of Adorno's Primacy of the Object. It's Ryan Gosling's replicant in Blade Runner 2049. It also provides a structure for "higher-dimensional" intelligence, which in theory could be infinite, as you add additional knowledge structures at higher levels that encompasses the lower structures.
To that effect, there must be some kind of human cycle beyond our individual "lives" here. It drives the attempt to represent knowledge on a computer, as it drove the attempt to write made up symbols on clay tablets. Maybe the human cycle itself has an "intelligence" to it. Maybe our culture and religions can be thought of as conscious (a collective conscious to a collective unconscious?). If we can build the soul, how does it remain eternal? What hymns will the religion of the Android write?