Ambient Computing
Ambient computing refers to the idea that technology runs invisibly in the environment and reacts in the background, rather than being consciously operated. Instead of picking up a device, one simply speaks, walks, or acts – and the technology draws its conclusions from that.
The normal way of dealing with technology works like this: you pick up a device, unlock it, look for an app, and type something in. Ambient computing describes the opposite of that. Here, technology is distributed throughout the environment – in lamps, speakers, sensors, watches, cars – and keeps working in the background without being consciously operated. The term comes from English: “ambient” means something like “surrounding” or “present all around.” So one is no longer meant to think of a single computer, but of a room that thinks along with you. The computer scientist Mark Weiser formulated this vision as early as the beginning of the 1990s, long before the matching devices even existed.
Why technology is meant to disappear
Every device costs attention. A smartphone interrupts conversations, draws the gaze, and wants to be unlocked. The basic idea behind ambient computing is to reduce these costs. Technology should only become visible when it is truly needed. Weiser called this “calm technology” – quiet technology that stays at the periphery of perception.
For companies, there is an economic interest behind this. Whoever fills a person’s environment with devices is constantly present without having to fight for screen time. That is why Amazon, Google, and Apple have for years been building ecosystems of speakers, watches, headphones, and displays that are all connected to one another. Once someone is inside such an ecosystem, they rarely switch brands, because otherwise the entire environment would have to be set up again from scratch.
There is also a serious field of application beyond mere convenience. In care settings, sensors in the floor can detect whether an elderly person has fallen, without them having to press a button. But this is precisely where the price becomes apparent: an environment that is permanently attentive is an environment that is permanently observing.
Sensors, evaluation, and response
Technically, ambient computing consists of three steps. First, sensors capture something: microphones listen, cameras see, motion detectors register presence, watches measure pulse and steps. Then software evaluates this raw data and classifies it. Finally, the system triggers a response, such as turning on a light, lowering the heating, or reading out a message.
The difficult part is the middle step. From a sound, the system has to infer whether someone gave a command or whether it was just the television playing. This requires models that have learned to recognize patterns from many examples. That is precisely why ambient computing only became practically usable once progress was made in AI methods.
An important distinction concerns where the computing work takes place. Much of it runs in data centers on the internet, i.e. in the so-called cloud. Increasingly, however, the devices themselves do the computing – this is called edge computing. That is faster and more privacy-friendly, because recordings don’t have to leave the house. Ambient computing should not be confused with the “Internet of Things”: there, the focus is initially just on devices being networked. Ambient computing additionally requires that they respond independently in a meaningful way.
From voice assistant to AI glasses
In everyday life, one has long encountered this principle, usually without knowing the term. A smart speaker in the living room that waits for a wake word belongs to this category. So does a smartwatch that reports an irregular heartbeat, or a car that unlocks as the key approaches. Automatic doors in the supermarket are also a very simple precursor.
In tech news, the term mainly appears when companies introduce new device categories. Smart glasses with a camera and voice assistant, AI pins without a screen, or networked earbuds are regularly marketed as a step toward ambient computing. Several such products have flopped, however, because the benefit did not justify the effort.
A common misconception is that ambient computing simply means “lots of smart devices.” What matters is not the number of devices, but whether one stops operating them individually. As long as you have to open a separate app for every single lamp, you have achieved exactly the opposite.