Oddcast episode
November 8, 2023
Karin Valis on Magic and Artificial Intelligence
A House with Many Rooms Interview 2
We are delighted to speak with Karin Valis, machine-learning engineer and esoteric explorer, on the vast subject of how the fields of artificial intelligence and magic overlap, intertwine, and inform each other. We discuss:
- The uncanny oracular effects and synchronistic weirdnesses exhibited by large language models,
- Conversations with ChatGPT considered as invocation,
- AI as the fulfilment of the dream of the homonculus (with the attendant ethical problems which arise),
- AI as the fulfilment of esoteric alphanumeric cosmologies (and maybe, like the Sepher Yetsirah, this isn’t so esoteric after all; maybe it’s just science),
And much more.
Interview Bio:
Karin Valis is a Berlin-based machine learning engineer and writer with a deep passion for everything occult and weird. Her work focuses mainly on combining technology with the esoteric, with projects such as Tarot of the Latent Spaces (visual extraction of the Major Arcana Archetypes) and Cellulare (a tool for exploring digital non-ordinary reality for the Foundation for Shamanic Studies Europe). She co-hosted workshops, talks and panel discussions such as Arana in the Feed (Uroboros 2021), Language in the Age of AI: Deciphering Voynich Manuscript (Trans-States 2022) and Remembering Our Future: Shamanism, Oracles and AI (NYU Shanghai 2022). She writes Mercurial Minutes and hosts monthly meetings of the occult and technology enthusiasts Gnostic Technology.
Works Cited in this Episode (roughly in the order cited):
The homonculus passage in the Pseudo-Clementines: Homilies 3.26; cf. Recognitions 2.9, 10, 13–15; 3.47.
On the Book of the Cow/Liber vaccæ: see e.g. Liana Saif. The Cows and the Bees: Arabic Sources and Parallels for Pseudo- Plato’s Liber Vaccæ (Kitab al-Nawams). Journal of the Warburg and Courtauld Institutes, LXXIX:147, 2016.
‘The Measure of a Man’, Star Trek: The Next Generation Season 2, Episode 9, first aired 13 February, 1989.
Doctor Strange, dir. Scott Derrickson, 2016 Marvel Studios.
Recommended Reading:
Karin has a substack where she posts interesting things. Her recent essay Divine Embeddings is particularly relevant to the discussion of alphanumericism in the interview.
Themes
Alphanumeric Speculation, Artificial Intelligence, Divination, Golem, House with Many Rooms, Interview, Magic, Oracles, Statues, William S. Burroughs
Joshua Schrier
November 16, 2023
I’ve been collecting some notes for an “imaginary course” on addressative magic as a way to think about AI…. mostly inspired by previous episodes of SHWEP…
https://jschrier.github.io/blog/2023/08/02/Imaginary-Syllabi-Prompt-Engineering-And-Addressative-Magic.html
Earl Fontainelle
November 16, 2023
Looks cool, Joshua!
My inner (and outer) pedant requires that I point out that the plural of “syllabus” is “syllabus” (fourth declension), but I feel like an arsehole even mentioning it. But, no choice. Pedants gonna pedant.
Karin Valis
April 26, 2024
Love these resources, thanks for sharing!
Joshua Schrier
November 16, 2023
@Earl….you certainly steered me towards a rabbit hole of corrupted etymologies…
https://www.quora.com/Which-is-the-correct-plural-of-syllabus—syllabuses-or-syllabi
https://languagelog.ldc.upenn.edu/nll/?p=2684
Just to be perverse, I think I’ll start using “syllabanti” in the future 🙂
Alexander Nader
November 17, 2023
I really love sittýba! Thank you for this really magic information!
Karin Valis
April 26, 2024
Also, @Earl I just finished the 6th series of TNG, and the way Ship Computer’s capabilities align with the current AI development really surprised me.
For example – episode S06E05 where the team in this scene prompted the holodeck in exactly the same way as you would work with image generator models?
https://www.youtube.com/watch?v=7S-Meq9MMuQ&ab_channel=FarhadZ
In episode S06E09 Data browses through the external logs of a broken Exocomp and the reasoning capabilities of the computer feel *exactly* the same as if he had a large language model answering his queries. It’s excellent in information retrieval, filtering and aggregation: “Hey computer, give me the logs from the time the Exocomp broke down and look for any anomalies.” But it’s still rather dumb when it comes to higher-level reasoning: “Computer, why would this anomaly occur?” Responds with “Unknown.”
It’s quite mindblowing.