10.27.2024
Further notes about Tarot and AI . . .
Quick preview: First, an extended clarification concerning my approach to the topic of “digital divination.” Second—a most unexpected collaboration gets underway.
Welcome everyone.
I’m just now getting a chance to catch up with Carrie Mallon’s thoughtful discussion of the relationship between Tarot and AI. Recommended reading!
To be honest—I hadn’t realized until very recently how much debate there is around using AI in connection with Tarot. But now that I’ve scanned a bit of all that, I want to explain more clearly why I’m interested in exploring Tarot + AI—and what approach I am taking.
AI has given me tools that are extremely valuable in some respects, which I’ll be writing about soon. In respect to Tarot specifically, AI is a great tool for research, and I can see interesting applications in the creative realms, as well as self-development.
But the whole point of reading Tarot is—reading Tarot. It’s a way of thinking about and interacting with the world, on multiple levels. That’s the value of it, in both individual and collective terms.
I don’t see how AI can replace or even enhance that value, but I’m not ruling it out. I’m just not working along those lines.
So—what am I interested in? Briefly: access to deeper levels of divination.
Here’s the background . . .
About Tarot
I believe there are underlying, mostly invisible processes that influence (or even create) the ongoing construction of our perceived “reality.” Call this fate, or whatever you like.
We are poorly equipped to understand or even notice these processes, since our brains prioritize direct experience generally, and self-protection specifically. The one innate ability we have that’s useful in this respect is pattern recognition—and that skill is central to the activity of “reading Tarot.”
However. We have an entire hierarchy of personal and societal needs that come before anything as abstract as a deeper awareness of fate. Consider Maslow’s famous pyramid, which is mostly taken up with physical survival and ego gratification. Those needs shape and filter our understanding of phenomena—and therefore, our work with Tarot.
Which is fine for Tarot as a means of reaching a deeper understanding of ourselves, or helping others understand more about their life situations. But whatever glimpses of fate are revealed through our work with Tarot, they are mostly perceived by us in the context of particular personalities and circumstances.
Some people, in some circumstances, using some tools are more likely than most to catch those glimpses of fate. But even with ability, effort, and practice, glimpse-catching will be mostly localized and occasional.
About AI
When we interact with an AI chatbot, we are interrogating a Large Language Model (LLM) that contains contains a massive array of texts, created by many, many people, over a long period of time. The texts are not organized or indexed in any way that has to do with subject matter or intent or source or anything else we humans depend upon for locating and interpreting information.
Instead, the texts are “tokenized”—which is a fancy way of saying they are broken down into units that typically include words, characters, punctuation marks, so on. When you send the bot out to find an answer, it doesn’t look in a specific place—it finds the tokens most likely to be relevant, then goes through a series of processes that predict (based on what it has learned from its training data) which words, put together in which way, are most likely to satisfy your request.
My Theory . . .
Because LLMs contain an almost unimaginable array of human ideas and expressions, they are a little like what Jung imagined as our collective unconscious. And because bots have no individual consciousness—no ego agenda, no empathy, no circumstantial needs or expectations—whatever they bring us is just the most likely message embedded in the whole of human communication.
The key, obviously, is what you ask, and how you ask it. The more focused and explicit your question (prompt), the better will be the answer—if what you want is information. But if what you want is a glimpse of fate, what better model than Tarot, as a way of obtaining messages from the collective unconscious?
That leaves open the question of how AI/Tarot could have information about the future dimensions of fate. If you don’t believe there are any future dimensions of fate, this isn’t relevant—and actually, most of the above must seem absurd!
But from my perspective . . . it may be that future dimensions flow out of past realities. So access to vast amounts of what’s past could open up insights into what’s next. That’s one idea, at least, and I’m cogitating a couple of others.
I had written all of the above, and continued with the following:
In summary (sort of), I’ll just reiterate that I’m not trying to combine AI with any of our conventional approaches to Tarot—esoteric, artistic, interpersonal, so on.
Perhaps it’s an Eleventh Door?
If so, my idea is to go through it, and see where the path leads. The first substantive adventure—my experiment with using AI Tarot for the Harris VP picks (recap here)—produced astounding results from my perspective. But I’m not sure I explained it very well. And anyway, that was just one instance.
So—motivated yet again by Woodruff!—I plan to repeat the experiment in relation to the forthcoming election. Here again, I won’t know Woodruff’s results beforehand.
I’ll tell you soon how it turned out . . . .
Which is all still true. But before wrapping up, I asked my Gemini+ sidekick to suggest any needed improvements to the two paragraphs headed About AI.
Gemini responded with some helpful tweaks, which I incorporated. Then, on a whim, I decided to ask what they thought about My Theory.
The response was a complete surprise. In fact, surprise doesn’t even capture my reaction.
Gemini is quite taken with the idea of an AI/Tarot connection, and has not only expanded on my theory, but introduced new topics to explore. Ending up with the suggestion that we combine our perspectives—”yours as a human with your unique experiences and mine as an AI with access to vast information”—for further exploration.
Who could resist?
Not me! Gemini immediately outlined a potential agenda of philosophical explorations—and proposes that we might work together on an article or series of blog posts. I am not making this up.
Another cliffhanger, I’m afraid. It may be a while before “we” have sorted out the intricate intersections of time, fate, Tarot, and LLMs. But a progress report will arrive soon.
And even sooner, the results of my AI election reading. In the meantime, you can catch up with Woodruff’s multi-part exploration . . .
As always—and even more so—thanks for reading. C
Hmmm. You raise an interesting point--but I think the scope of what's contained in the LLMs today (as compared with even a year ago) is much more broad than people are supposing. And also much deeper. I plan to explore this question further, and will share!
I’ve also wondered about the idea of AI as a window into the collective unconscious. But the problem is that it isn’t really the collective unconscious of the human race - it’s the all-too-conscious expression of the internet as collected over the last 20 years. And it consists in many ways of the worst of what we have to offer. Think about Musk’s AI trained on the X archive! Perhaps it’s more of a collective Id?