Disintegrator Marek Poliks, Roberto Alonso

Alex Reisner's writing in the Atlantic is some of the best investigative coverage of Large Language Models out there. In this episode, we talk through the mind-bogglingly vast archives of random pirated material that provide every major commercial LLMs with their linguistic faculty. 

Definitely check out his writing on https://www.theatlantic.com/author/alex-reisner/, especially the phenomenal January 11 piece on "memorization."

ALSO -- if you haven't -- submit to our call for papers on AI interfaces: link! We'd love to have you.

Refik Anadol, and by extension Refik Anadol Studio, is one of the most visible, if not the most visible, artists working with large models today. His work is everywhere, from MoMa to the Biennale Venezia, from the very first Las Vegas Exosphere art display to the front of Walt Disney Concert Hall.

We’re delighted to have had him on the pod to talk through his artistic philosophy, touching specifically on media, light, AI, and his new incredibly large-scope Nature Model project announced back in January (approximately the same time we had our conversation with him — yes, the backlog is real).

We're also accompanied in the virtual studio with Pelin Kivrak, who writes as apart of Refik Anadol Studio.

Jennifer Walshe is one of the coolest people we know. Her artistic work and thought has broken our brains for years, leaving us shipwrecked in its torrential waves of reference and irony and joy and conceptual viscera.

We talk about her recent piece for the Unsound Dispatch, 13 Ways of Looking at AI, Art & Music — a series of vignettes that in their totality assemble into one of the most coherent accountings of what it is we’re all experiencing.

Some references from the ep:Listen to Things Know Things on RTÉ Lyric FM. Hopefully you’re aware of the music duo Matmos — Jennifer references this record in the context of discussing conceptual work. Jennifer also speaks often of her close collaborator Jon Leidecker (Wobbly), who has a few absolutely killer sets with Matmos, including this one.You can interact with Walshe’s Text Score Dataset here.We continue to enjoy references to Holly Herndon and Mat Dryhurst’s Have I Been Trained (https://haveibeentrained.com/), a way to search for your (or anyone’s) work in large, public, AI training datasets.Two movies everyone should see: Catfish the Movie and HER. (We’d also recommend Catfish the TV show, of course).Jennifer mentions the computer scientist Kate Devlin’s work, especially “Turned On: Science, Sex and Robots.”If you haven’t googled a picture of Paro the Therapy Seal, do it.Jennifer’s record “A Late Anthology of Early Music Vol. 1: Ancient to Renaissance” is a top lifetime record as far as we both are concerned. Check out track 16 for that Palestrina. It’s CRAZY. To wrap it up, check out Ted Gioia’s Substack and Bruce Sterling’s writing (the concept Walshe references is "Dark Euphoria").

Go here for more information about the upcoming talk that Roberto and Marek are doing Sunday, March 10, at 10AM Pacific. It's virtual, so come join us!!!

Benjamin Bratton writes about world-spanning intelligences, grinding geopolitical tectonics, “accidental megastructures” of geotechnical cruft, the millienia-long terraforming project through which humans rendered an earth into a world, and the question of what global-scale order means in the twilight of the Westphalian nation-state.

Candidly, if either of us were to recommend a book to help you understand the present state of ‘politics’ or ‘technology’, we’d probably start with Bratton’s The Stack — written 10 years ago, but still very much descriptive of our world and illuminative of its futures.

If the first 10 minutes are too “tech industry” for you — just skip ahead. The whole conversation is seriously fire, and it spikes hit after hit of takes on privacy, bias, alignment, subjectivity, the primacy of the individual … all almost entirely unrepresented within the Discourse.

Some references:
We briefly talk about EdgeML, which essentially means the execution of ML models on small computers installed in a field location.Benjamin mentions his collaboration with renowned computer scientist and thinker Blaise Agüera y Arcas, whose work on federated learning is relevant to this stage of the conversation. Federated learning involves a distributed training approach in which a model is updated by field components who only transmit changes to a model therefore retaining the security of local training sets to their own environments only. Also - here’s a link to their collaboration on “The Model is the Message."Benjamin calls himself a bit of an “eliminative materialist” “in the Churchland mode,” meaning someone who believes that “folk psychologies” or “folk ontologies” (theories of how the mind works from metaphysics, psychoanalysis, or generalized psychology) will be replaced by frameworks from cognitive science or neuroscience.Benjamin calls out a collaboration with Chen Quifan. Check out Waste Tide — it’s excellent sci-fi.The collaboration with Anna Greenspan and Bogna Konior discussed in the pod is called “Machine Decision is Not Final” out on Urbanomic.Shoshana Zuboff is a theorist who coined the term “surveillance capitalism,” referring to capital accumulation through a process of ‘dispossession by surveillance.’ The implicit critique of “surveillance capitalism” in this episode hinges on its overemphasis on individual sovereignty.“Tay” was the infamous AI Twitter Chatbot Microsoft rolled out for 16 hours before pulling back for its controversial content.Antihumanism refers to a rejection of the ontological primacy and universalization of the human afforded to it through the philosophical stance of “humanism.” An “antihumanist" is someone who challenges the stability of the concept of the “human” or at very least its salience in cosmic affairs.Check out Benjamin’s new piece on Tank Mag (Tank.tv), it’s fire. And check out Anna Kornbluh’s AWESOME “Immediacy, or The Style of Too Late Capitalism” on Verso.

Anil Bawa-Cavia (AA Cavia) is one of our favorite writers and practitioners on the philosophy of computation. We discovered his work through Logiciel, on &&& (we 3 &&&!), both a gorgeous book in print and an elegant formal depiction of what computation might actually be (a definition that stands in striking contrast to the limitations imposed upon it by the humanities, or the comprehensive universality bestowed upon it by that particular breed of TEDx computational ‘realists’).

This conversation is a really nice parallel to Anil’s amazing chapter in Choreomata, in which he identifies the bottlenecks we are rapidly approaching through deep learning as, in part, products of incomplete thinking as to the nature of language, learning, their messy and entangled relationship to the “world,” and their reconsumptive throughput as it assembles into what we increasingly understand as something like intelligence.

We want this conversation to be accessible to as many listeners as possible, so here are some further references and definitions that might be useful:
I’ll be honest, I was surprised when I learned how radically different (and how totally gendered) the “Turing Test” was in its original formulation from what it’s become known to be. Read about it directly via: Turing - Can Machines Think (https://redirect.cs.umbc.edu/courses/471/papers/turing.pdf).It’s likely the distinction between supervised and unsupervised learning is very clear to most listeners, but if you’re unfamiliar with this distinction, see a sufficient overview here (https://www.ibm.com/blog/supervised-vs-unsupervised-learning/). This becomes important as Anil starts speaks to the implications of things like pedagogy and normativity to learning.The concept of normativity is used quite a bit here in a way that might be unfamiliar to some people. Think of normativity as the moment the word should enters into some construct — both in the prescriptive sense (“you should behave according to xyz social norms”) but also to some extent in the empirical sense (“based on what I’ve observed so far, this type of outcome should result from this interaction”). While we encode norms into language models (both through supervised learning, but also through the hidden organizing principles that are contained within complex structures like language), we do not encode “normativity” — a way of engaging with norms as norms. This is a good place to start when trying to understand the critique from inferentialism that Anil brings from Wilfred Sellars and Robert Brandom.An “embedding” is essentially the ability to place some system or configuration within another system in such a way that its general shape is retained. In the context of machine learning, language is embedded into a high-dimensional numerical space wherein meaning can be identified by the proximity of various words within that space, and translations between languages can be accomplished by looking at the position of words within one language’s embedding and correlating that to a similar set of positions in another. You don’t need to understand topology to intuit what this might look like in a way that is sufficiently useful. Anil playfully refers to “embedding” in Wilfred Sellars’ work — a philosopher who argues that everything we know is ‘embedded’ within complex webs of beliefs, norms, and meanings.Anil references Alain Badiou’s writings on finitude, and it’s our impression that this is a reference to Badiou’s completion of his enormously sprawling Being and Event trilogy (“The Immanence of Truths”). Not an essential book for this...