A few years ago you probably read about the ‘internet of things,’ a comfortable but baggy way of describing a network of interconnected electronic devices. The idea occupied a central place in that other forgotten project, the ‘fourth industrial revolution,’ which jammed together everything from genomics to virtual reality.
The goal of the internet of things was to make everyday machines ‘smart,’ a mission that I suppose has been accomplished. I allegedly have a smart speaker, a smart TV, and even a smart oven that I tell myself I will one day connect to the wifi network. It’s a fittingly dull task for a technology that doesn’t get the blood pumping, one that in some ways follows the path laid down by electricity or telephony. Like the smart-everything, these are remarkable things that became first mundane and then invisible.
If you can gloss over the horrible phraseology, the connected device is a useful thread to pull for making sense of assumptions about what good machines do. When an old appliance breaks, it’s replaced by a new model that insists on attaching itself to your wifi. The physical layer gets wider as more sensors, communication nodes, and platforms are folded into the network. But it also gets deeper as the connected devices become a little more lively.
Boosters describe these devices as manifestations of ‘artificial intelligence’ because they identify changes in the environment and change state accordingly. Weighing a smart washing machine against a large language model seems a bit overwrought, but it does raise an important point about the nature of digital intelligence: it doesn’t care about the shape of its container.
ChatGPT may animate your computer or phone, but the real magic is happening in an Arizona data centre. Even without access to the internet, compression techniques now let a three billion parameter language model run on a phone’s battery without cooking it. Researchers are doing the same for the hardware layer, proving that models can live next to the signal rather than a continent away.
This points us towards a curious observation about AI in the popular imagination. Large models can be anywhere with enough processing power or with sufficient connectivity, but we only tend to picture them residing in a small set of physical platforms. Humanoid robots loom especially large in the public psyche, giving the impression that digital intelligence obeys the same rules as our own.
But AI doesn’t work like that. It will populate the world around us and turn appliances, devices, and computers into talking (and in some instances, walking) machines that interact with us from whatever vantage point they can cling to.
For the purposes of this post, I’m going to make a few assumptions about the future. These are (a) the models will broadly maintain their current rate of improvement for the foreseeable future; (b) the best models of any given moment will shrink to allow for local deployments; (c) different models will be able to communicate with each other; and (d) lots of physical platforms are capable of hosting the models in one form or another.
If each of these assumptions hold, there’s no reason to think we won’t have one AI model (or a handful of models) that live across many substrates within the next couple of years. After all, we already have Claude in a vending machine, Grok inside the newest Teslas, and LLM-powered assistants in fridges, ovens, and dishwashers.
Animism through the ages
Animism is the conviction that spirit inhabits matter, one that shows up far earlier than the word itself. Palaeolithic hunters painted animals on rock walls and carefully arranged bear skulls in a way that archaeologists interpret as negotiations with animal persons. When Edward Burnett Tylor coined the term animism in his 1871 book Primitive Culture, he formalised that observation by defining early religion as ‘the doctrine of souls and other spiritual beings in general’ that could reside in the natural world.
By the Bronze Age, the impulse to see spirit in matter hardened into liturgy. In Mesopotamia a newly carved cult statue underwent the mîs-pî (‘washing of the mouth’) procession. Craftsmen led the image to a riverbank where its lips were ritually cleansed, then ‘opened’ with cedar oil. From that moment on, the wood and precious metal counted as the god’s living presence and was capable of eating offerings, signing treaties, and punishing neglect.
A similar performance took place in the Egyptian Old Kingdom. Priests put peseshkaf blades to the mouth and eyes of statues in the ‘opening of the mouth’ ritual that enabled a figure to enjoy food and speak in the afterlife. Eberhard Otto’s Das ägyptische Mundöffnungsritual from 1960 identified 75 examples of the practice, an effort that emphasises the institutional heft underpinning animistic practice.
Centuries later, objects did god’s work in the churches of the Byzantine Empire. Painted boards, splinters of bone, weapons, and other artefacts were thought to contain divine energy (energeia). John of Damascus made the theology explicit when he said: ‘I do not worship matter; I worship the God of matter, who became matter for my sake.’
But probably the most famous example of animism comes from Japan’s Shinto, a religious tradition whose roots lie in early agrarian rites that personified the forces sustaining rice cultivation. At its centre are kami, the packets of vitality saturating both nature and human-made objects that invite reverence through craftsmanship or long use.
Everyday life in a Shinto frame assumes a world that listens back. A shrine gate marks a threshold where rock and tree possess their own intentions and household rituals treat the cooking fire or well as moral participants. That habitual attribution of inner life is what scholars describe as animism.
When the electric telegraph connected Europe to America in the nineteenth century, the Victorians seized the new medium as proof that voices could travel between planes of existence. Séances were framed as ‘circuits’ and mediums styled themselves human telegraph stations. A New York weekly even titled itself The Spiritual Telegraph, reporting on dispatches from the afterlife in the form of a newspaper. As Desmond G. Fitzgerald, the editor of the Electrician, put it in May 1862:
“Telegraphy has been until lately an art occult even to many of the votaries of electrical science. Submarine telegraphy, initiated by a bold and tentative process – the laying of the Dover cable in the year 1850 – opened out a vast field of opportunity both to merit and competency, and to unscrupulous determination. For the purposes of the latter, the field was to be kept close [sic], and science, which can alone be secured by merit, more or less ignored.”
The author Jeffrey Sconce notes that popular magazines styled the séance room as a kind of domestic telegraph station. In Haunted Media, he describes the rise of spiritualism as a utopian response to the electronic powers presented by telegraphy and connects the emergence of the radio with an ‘atomized vision of the afterlife.’
The upshot is that animism has a knack for rerouting through the technologies of the day, whether that’s burial artefacts, swords, or telegraph wires. People may not worship microwave ovens in the future, but I wouldn’t rule out the adoption patterns of animistic living that stress interconnectedness, vibrancy, and agency.
Intelligence in the pipes
Animism is more habit than philosophy, a set of reflexes we use when faced with unexpected forms of mediation. Its long life reminds us that we've done this before, and that what feels new — doorbells and ovens and thermostats that listen — is the latest chapter in a much older story about how we relate to our surroundings.
The diffusion of AI into the world is a well-trodden cultural negotiation, one that should make us weary of the image of an embodied intelligence that only exists within humanoid robots or specialist AI hardware. That will surely happen, but these kinds of deployments will only represent the most visible form of physical manifestation.
Just as a smart thermostat slips into invisibility, large models may come to occupy our surroundings in ways that feel almost as unremarkable. Animism has some utility here. It encourages us to notice intelligence in the places we aren’t used to looking, to imagine the world as a little more vibrant.
An interesting discussion of assistance but I'm not certain of intelligence. All of the objects referenced seems to have a limited purpose. In case of object held to be sacred, that object has a specific role to play in the larger environment of humanity. In the case of intelligence applliances, I'm certainly glad mine stay put where they are and only do what they are programmed to do. I can only imagine the chaos of my intelligent tv got together with my intelligent dishwasher to discuss what to do about my intelligent mower. Oh my.