One of the overall guidelines should be "to keep the human in the loop" when designing smart environments. Without entering into the philosophical discussion of
when it is justified to call an artefact or environment ‘smart’ or what we consider ‘smart’ or ‘intelligent’ behaviour in general, the following distinction turns out to be useful (Streitz et al.,
System-Oriented, Importunate Smartness
An environment is ‘smart’ if it enables certain self-directed (re)actions of individual artefacts (or by the environment in case of an ensemble of artefacts) based on previously and continuously collected information. For example, a space or a place can be ‘smart’ by having and exploiting knowledge about which people and artefacts are currently situated within its area, who and what was there before, when and how long, and what kind of activities took place. In this version of ‘smartness’, the space would be active, (in many cases even proactive) and in control of the situation by making decisions on what to do next and actually take action and execute them automatically (without a human in the loop). For example, in a smart home, we have access controls for the house and other functions like heating, closing windows and blinds are being done automatically. Some of these actions could be importunate. Take the almost classic example of a smart refrigerator in a home analyzing consumption patterns of the inhabitants and autonomously ordering depleting food. While we might appreciate that the fridge makes suggestions on recipes that are based on the food currently available (that would be still on the supportive side), we might get very upset in case it is autonomously ordering food that we will not consume for reasons beyond its knowledge, such as a sudden vacation, sickness, or a temporal change in taste.
People-Oriented, Empowering Smartness
The above view can be contrasted by another perspective where the empowering function is in the foreground and which can be summarized as ‘smart spaces make people smarter’. This is achieved by keeping ‘the human in the loop’ thus empowering people to make informed decisions and take actions as mature and responsible people who are in control. In this case, the environment will also collect data about what is going on and aggregates the data but provides and communicates the resulting information - hopefully in an intuitive way so that ordinary people can comprehend it easily - for guidance and subsequent actions determined by the people. In this case, a smart space might also make suggestions based on the information collected but the people are still in the loop and in control of what to do next. Here, the place supports smart, intelligent behaviour of the people present (or in remote interaction scenarios people being away ‘on the road’ but connected to the space). This view can be summarized as ‘smart spaces make people smarter’.
Of course, these two points of view will often not exist in their pure distinct forms. They rather represent the end points of a dimension where we can position weighted combinations of both somewhere in between. What kind of combination will be realized is different for different cases and depends very much on the application domain.
* N. A. Streitz, C. Röcker, Th. Prante, D. van Alphen, R. Stenzel, C. Magerkurth (2005). Designing Smart Artifacts for Smart Environments. IEEE Computer,
March 2005. pp. 41-49.