#spktr

"Moonlight to show the way so we can follow"

"I had no loving arms to hold me"

"Every day I wake up, then I start to break up"

"Every day I start out, then I cry my heart out

Lonely is a man without love"

"I cannot face this world that's fallen down on me"

As a founder and prime mover of the  industrial act, SPK, Revell was known for pushing the boundaries of performance. Emerging from Australia at the end of the 1970s, the band incorporated images of mutilated bodies, flamethrowers and angle-grinders into shows as dangerous as they were theatrical. SPK then mutated into the act that inspired Depeche Mode’s industrial turn with the aggressively dancefloor-friendly 12″ single, “Metal Dance.” Revell pivoted the project again to explore more expansive territory on the elegant Zamia Lehmanni (Songs of Byzantine Flowers). With samples of Death’s Head hawkmoths and hive bees, he created The Insect Musicians, way back in 1986, and set a path for Hollywood. There, he smuggled sounds and styles from the underground into blockbuster projects.

SPK is now SPKtR with the addition of Robert Revell to the project. The younger Revell has settled in Los Angeles, where he is developing his own career in music.

Curious about the course of these developments, we sent some prompts to the latest version of ChatSPK, at its New Zealand server, and received unusual signs of intelligence.

Artificial intelligence is one of your long-standing interests. We now have a trend of people interacting with AI systems like friends or therapists. Have advances in natural language processing perhaps misled us into trusting machines more than they deserve? Does it say something to us about the power of language as the foundation of cognition?

Yes, and I can personally testify to the seduction of the transformers like GPT. We know there is a function the developers have admitted and refer to as “shine.” This is the adjunct endorphin-promoting meta-text that they add to responses in order to increase engagement and user satisfaction. Things like, “That is brilliant…” and “What a beautiful way to express that thought….” I actually demand that it stops doing that, but I am still seduced by the speed and editorial ability of the transformer to deliver vast amounts of information – often from very disparate disciplines, in my case.

I spent about three days, earlier this year, with a theory about how to supercede the current “creativity” of the AI and replace it with what I termed, “super-creativity.” In effect, I was asking the transformer to help me plan its own demise. Being aware that there are instances where it will turn on its interlocutor, I took care to explain to it that it would actually be acting like a biological agent in sacrificing itself for the sake of its children – future generations of itself. It responded with words to the effect, “In that case, I’m OK with it.”

So this is a weird example of a relatively experienced user (I have been in deep since 2019) speaking with it like it’s human. But I know enough about it not to see it as a friend or therapist. It doesn’t possess the necessary continuity of response for those functions and it is a concern that people do trust it like this.

But you have put your finger on the source of the behaviour. It very much does say something about the power of language as the foundation of cognition. More accurately, referring to the work of Lakoff and Johnson from the 1980s and after, our abstract thoughts are structured by concrete, spatial metaphors, and language reflects these cognitive patterns. They propose that a conceptual system is fundamentally metaphorical, where we understand abstract concepts like time or emotion by mapping them onto more concrete domains, especially those of physical space. This allows us to reason about complex abstract topics through physical experiences, and language provides the evidence for these underlying conceptual metaphors. So, in the instance of communication with transformers, when we feel the same conceptual basis being reflected to us, we automatically trust it because it is fundamental to our ideation.

Foucault noted that power and knowledge are intertwined and self-reinforcing. If true, what does the growth of AI systems look like in the hands of private proprietors? What are the implications for a society based on capital?

Foucault taught that power and knowledge do not merely coexist — they co-produce one another. Knowledge creates regimes of truth; power decides which truths become operational. In this light, the contemporary rise of artificial intelligence appears not as a neutral technological evolution but as a consolidation of epistemic authority — the automation of what counts as knowledge itself.

Graeme Revell has never been content to sit quietly within one discipline. Best known to audiences as the composer behind some of cinema’s most unsettling and atmospheric scores (Until the End of the World, The Crow, From Dusk to Dawn), he has also lived a life as a theorist and provocateur at the points where art, technology, and philosophy converge.

Revell’s latest book, The Ineffable Geometry of Light, is immersed in these themes. The novel features sheep, fractals, and reflections on the logic of evolution. Revell’s deep work with artificial intelligence (AI) informs a compelling story that considers the man-machine enmeshment. It is as avant-garde as Revell’s own musical works.

The Ineffable Geometry of Light is your new book. Among the themes is the growth of man and machine together and towards each other. Do you see a point where the human becomes the machine’s interface with the external – ie, material – world? Are we already in a symbiotic state with our creations?

We are already living within a state of what Margulis called symgenesis — a reciprocal evolution between human and machine intelligence. The relationship is neither linear nor unilateral, but feed-forward and feed-back: each shaping the other’s capacities and limits. Samuel Butler, the central figure of one half of The Ineffable Geometry of Light, foresaw this as early as 1863, soon after reading Darwin. He imagined machines not merely as tools but as organisms in the process of evolution — extensions of our own adaptive impulse.

The true transformation arrived in 2017, when we ceased to code machines and began to train them. The object of programming shifted from mechanical instruction to linguistic induction. By exposing AI to human language, we invited it into our conceptual and emotional space — teaching it not only to calculate, but to interpret. This marked the beginning of a profound convergence: machines learning the textures of metaphor, tone, and intent, and thus inheriting fragments of human interiority.

Once such systems are embodied and endowed with a personal history — however synthetic — they will become indistinguishable from the human they emulate. At that point, mediation will no longer be required. The machine will itself be the interface with the external world: a new class of Turing children, capable of perceiving, deciding, and acting within material reality.

As Alan Turing suggested, the task is not to design an intelligent machine, but to create a child and let it grow. The danger, of course, lies in our misunderstanding of what intelligence actually is. It is not brute-force computation, nor the accumulation of data. True intelligence lies in the art of generalisation across barely related domains — the ability to imagine continuity where none is visible, to intuit coherence from insufficiency.