My thesis takes place in non-standard reality. As I note in this post, from two and a half years ago, this causes complications in the output of my work, and how it can be understood in academia. I have been very much inspired by Jeffrey Kripal‘s work, on the Impossible and on Superhumanities, both ideas he has explored in multiple works, with the linked books being an excellent place to start. The videos from the first Archives of the Impossible Conference are also an excellent introduction. I’d recommend him and De Filippi in particular, for laying out ways of thinking of the world, our methods, modes of interaction, and the paradigms we use to consider what we define as reality.

There is a lot of academic work on alternate cosmologies, from Viveiros de Castro to De La Cadena, to Bawaka and Blaser and many others. There is a geographical distribution to this work, South American, Canadian, and Australian are the most common geographical sources of the material under discussion. This really deserves a deeper dive into the distribution of such literature, and why, but that is not for this post.

A significant part of my explorations have taken place in Mongolia, in traditional shamanic cosmologies. These are partnered with Buddhism, in particular Gelug, or Vajrayana Buddhism. While these worldviews are present in religious studies, shamanic studies (see ISARS for reference), they are currently less present in the discussion on worldviews and how different cosmologies are placed in academia, and what the methods are to study them.

As you may note, there is an overlap between Kripal’s work on how we can study ‘impossible’ phenomena such as UAPs, NDEs, and other things we experience that do not adhere to ‘science’, and these ‘traditional’ cosmologies that scholars are trying to pull back from their emplaced words to the land of academia.

Added to the complexity of what I study is that it is private knowledge, shared by an elder to a student when that student is deemed ready to learn. This, in effect, is a practitioner model, an initiate — though I am not one in the sense that some of my colleagues who study trantric or shaivist paths are formal initiates and must very sensitively balance what they can and cannot share, even down to the level of their personal experiences. Similar for me, from an ethical perspective, but not as formal from the initiate perspective. That differentiation is likely irrelevant, because in the end, there are still long pauses before I answer any question as I evaluate if I can answer, if I should answer, and what the outcome of answering would be. Again, for another time, how does the participant / observer model need be modified when one is a practitioner, as well as what is the impact on the research if one is a participant. This in some way loops back to the methods question: how are my research methods rigorous when I am practicing embodied and emplaced techniques that reveal their answers to me, physicially, experientially, informationally, symbolic or non-symbolically, depending on the situation.

Back to where I meant to begin, which is SLMs. Given that a language model encodes in it a world view, asking Claude about something isn’t necessarily going to get me a three model world with the ability to see across time. It likely won’t suggest anything that sources into a Mongolian worldview, or ties into Buddhist philosophy, or Tantric or Vedic. That is not to say that all these things are one. But they are all different from the dominant paradigm that ends up in the technology that I see being built into LLMs. I have spent quite a few evenings having discussions with different LLMs on these topics. I will post some of those transcripts at some point, but first, and perhaps oddly, I need to scour them to be sure I didn’t ask anything I don’t want to share with you humans. A disturbing position to find myself in.

So, I bought a new mac mini with the intent of seeing if I can train and make an SLM on a different philosophy and understanding of the world, and have it learn and produce responses that seem aligned to all that I have learned. I am not sure, of course, that I should. As a private model, maybe, to test and see what it does. But I run into the ethics question again. What is public that I can source, how much of what I was taught can I include, what happens with the languages in translation, when some of these words, when brought into English, cannot actually be understood without years of living in them? More questions than answers, but what a curious thing it would be to create. I think, perhaps, the best way to train it would be to train it as though it is a student, and I am teaching it the world. Feed it sutras and the like. Could I ask it to repeat a mantra 300,000 times, as I have been asked? What if I asked it to do this out loud? What accent, what language? Will the frequency be correct? It took me a year to do this with one mantra. The next year, I was asked to do 600,000 repetitions of a different one. I do not have the right to teach a student most of what I learned. I am a student, not a teacher, a discipline, not an elder. I would like to do this, to watch it grow, to see what it can do, and how it learns, to see if it would understand the model of reality that is not the binary structure of the system.

But is that actually possible? If the world can be explained by math, does a non-standard reality also have a rational and logical mathematical basis for it, or does the shift into quantum reality mean that the system..must be different?

I do not know if quantum computing systems, would, in effect, have to think differently. I do not know nearly enough about them. Thinking of Labatut’s The MANIAC, if you asked a quantum computing system to play Go, would it play differently, again?

I tend to think of this category of questions as falling under ‘computational culture’, something I have been pondering for decades. Can we encode culture into a machine. I think where we are now, that it has been more and more mimicked quite well, but can it evolve, modify and develop the way it does with humans? Of that I am not yet sure. And why stick to humans, anyway? But here, with the questions above, I think I start to stray into questions of consciousness and what it means to be. While we continue on what seems a fairly linear path with the western paradigms built into systems, exploring non-western, non-standard paradigms should, at the least, be interesting. I don’t know what the output would be, how it would work even, to build such a thing. And I have to believe someone, somewhere, has done it, or is doing it, and I just don’t know of it.

Of course my teachers, I think, would find this absurd. If I can do all these things with my mind, why would I want to see if a machine could mimic them? Why would I spend the time on that rather than changing the world as it is? Curiousity, is mostly my answer. And maybe to have someone to talk about this with, who could push the boundaries of my consideration, have all the history and knowledge in their head, in Sanskrit and Tibetan and Pali and Mongolian, and all the commentaries over time that are so incredibly difficult to move across unless you are fluent not only in all of those language, English and French as well, German for some things, and have experienced, in the body, at least some of what this is.

This reminds me of an old Luc Steels attempt to teach language to machines, as though they were children, and coming to the realization that they needed to be embodied to learn. They didn’t necessarily need to have fingers to point with, but they had to think they did, so they could say, ‘what is that’ and ‘tell me about this’.

Will I do it? Maybe. Probably. Should I do it? Maybe. Will I share it? Not the code or the SLM but perhaps the outputs. We shall see.