Part of an occasional series: ideas from outside the usual solarpunk reading list.
Most conversations about AI, automation, and technical progress share an unspoken assumption: more autonomous, more self-sufficient, more deterministic systems represent more advanced engineering. A 2024 doctoral thesis in architecture, by Argentine researcher Leandro Tomás Costa, draws on a mid-20th-century philosopher of technology to make a case that inverts this intuition directly — arguing that a fully automated, fully deterministic system isn’t the pinnacle of technical achievement. It’s a degraded version of one.
Not a book about AI — but relevant to it anyway
Costa’s actual subject is architecture: how the centralized energy infrastructure of the twentieth century shaped the housing typologies we live in, and what a new kind of infrastructural, housing-oriented architecture might look like in response to what he calls the „cybernetic society” now taking shape. It’s not a text about artificial intelligence. But one chapter, working through a philosophical disagreement between Martin Heidegger and Gilbert Simondon on humanity’s relationship to technology, produces an argument with direct and useful application to how we think about AI systems specifically.
Two twentieth-century answers to the same question
Costa frames Heidegger and Simondon as offering two structurally opposed responses to the same underlying question: what is a person’s place in a society increasingly shaped and organized by technology?
Heidegger’s answer is essentially passive and reflective. His famously cryptic line — „only a god can save us now” — treats waiting, contemplation, and meditation on the essence of modern technology as a form of action in itself, the path toward eventually transcending technology’s grip rather than actively intervening within it.
Simondon’s answer is active and constructive. He casts the human being as a mediator between technical objects — someone capable of building a genuinely non-alienating place for themselves within a technological society, but only given the right kind of surrounding architecture and space, which Costa argues Simondon himself never actually supplies. Costa’s own project is, in effect, an attempt to build the missing architectural half of Simondon’s proposal.
The key distinction: technical objects versus automata
The part of Simondon’s thinking most directly useful here is a specific distinction he draws between two categories of technical things.
A technical object, in Simondon’s framework, remains open to creative, somewhat unpredictable interaction with its environment and its user — it has, in a phrase Costa quotes, a margin of indeterminacy in how it operates.
An automaton, by contrast, is a closed, fully deterministic system — every input maps predictably to a fixed output, with no real openness to the unexpected.
Here’s the part that runs against intuition: Simondon argues an automaton is not the high point of technical development. It’s a diminished version of it. Real technical advancement, on his account, isn’t measured by increasing automatism — it’s measured by preserving a genuine margin of indeterminacy in how a machine actually operates. A society that treats pure automatism as the value to optimize for is, in Simondon’s specific and deliberately counterintuitive framing, a society of low technicity — one that has narrowed itself down to sheer determinism and closed itself off from genuinely creative exchange with its environment.
The Golem, as a figure for this fear
Costa adds his own contribution to this argument: the figure of the Golem as a myth expressing fear of the automaton specifically. A being made of clay, brought to life by a purely logical command, that ultimately slips beyond its creator’s control. Costa reads the Golem not as a story about fear of something genuinely foreign or alien, but as a story about fear of our own power over matter and language — the anxiety of „playing God,” directed inward at our own capability rather than outward at some external threat.
Why this matters for how we think about AI specifically
This gives us a fourth, independent line of critique of the contemporary AI industry, alongside others we’ve drawn on in our broader research — arriving this time from the philosophy of technology via Simondon and Costa, rather than from ecological limits, critiques of techno-utopian „prometheanism,” or debates about technological determinism versus relativism.
The core insight, translated directly to AI: automation and closed determinism in a system are not signs of technical sophistication — they’re signs of technical impoverishment. Applied concretely, an AI system optimized specifically for full automatic determination — predictable, closed off to anything resembling error or instability, engineered to eliminate every margin of unpredictability — is, on this specific philosophical framework, technically worse, not better, regardless of how efficient or high-performing it might be by conventional operational metrics.
This inverts a default intuition that runs through a lot of contemporary AI discourse: the assumption that „more autonomous and more deterministic” straightforwardly means „more advanced.” Simondon’s framework, as developed through Costa’s reading, gives a philosophically grounded reason to doubt that equation specifically — and to ask, of any system marketed as more capable through greater autonomy, whether what’s actually been gained is capability or simply a narrower, more brittle kind of determinism dressed up as progress.
A caution about the source, stated plainly
This chapter is one part of a much larger doctoral thesis whose central project is architectural, not philosophical — proposing a new category of „Infrastructural Housing Architecture” as a response to what Costa calls the cybernetic society. We’re drawing specifically and only on the Heidegger/Simondon chapter here, which is the part most directly relevant to AI critique; the thesis’s broader architectural argument, and several other chapters we haven’t engaged closely, aren’t represented in this piece. We’d also flag that this line of critique — automatism as impoverishment rather than advancement — arrives to us at one remove: through Costa’s own reading and application of Simondon, not from Simondon’s original 1958 text directly. We think Costa’s argument holds up as presented, but a reader wanting the argument in its most authoritative form should go to Simondon himself, not to us or to Costa as intermediaries.
Did we get something wrong?
This piece is drawn from our own reading notes, not a translation of the thesis. If something above misrepresents the argument, or just sounds off, tell us — publicly or privately, whichever you’d rather.
Want the fuller picture?
This is a survey, not the whole thesis. The deep-dive version is here: deep-dive PDF.
What should we read next?
This is one piece in a loose series drawn from our own research reading. If there’s a thinker, a text, or a tradition you think belongs in a series like this, we want to hear it.
