Title: EXAMPLE OF INTELLIGENT SHOWERS

The definition of the term intelligence can, at this stage, best be explained by a simple example:  

A household of four people uses a thermostatic shower that can (but must not) be activated with a fingerprint. Each fingerprint is linked to an individual. When a certain member of the family takes a shower for the first time, the default temperature is 26 degrees. After several showers, the software will deduce from the behaviour of an individual that he likes a temperature of about 18 degrees. For instance after taking three showers, the default temperature for that individual will become 18 degrees instead of 26 degrees.

By automatically collecting the shower activity of an individual, the shower ‘learns’ that the individual likes relatively cold showers of 18 degrees. If, due to a cold winter, the individual adjusts the temperature to 24 degrees, the default position will be automatically changed to 24 degrees. The same applies to the other members of the family. Now, when this person takes a shower in a hotel that uses the same technology, the shower in the hotel could also adjust the shower temperature to the preferences of the individual - if the profile of this individual is communicated to the Hotel shower software.

In this deliverable, one of the pertinent issues is whether it is possible to store the dynamic profile on the personal device of a user, instead of a network or a remote database, and - if possible - to which extent this falls within the scope of Ambient Intelligence. In a further elaboration of the example, the profile could take into account more specific situations, such as a person taking a shower after a tennis game prefers much colder water than when taking a shower in the evening, etc.  

As to the use of the term intelligence, it should be stressed that this intelligence depends on the one hand on the input of those that write the software and the algorithms involved i.e. computers only act according to the codes written for them. However, on the other hand, the intelligence of the software depends on the fact that the software moves beyond the original input, building on emerging patterns and thus learning in the sense that knowledge is constructed, applied, reconstructed and so on. Whether this type of learning measures up in any significant way to human learning and human intelligence is out of this document’s scope, however, the fact is that the computing powers of the hardware involved are far beyond those of humans. This allows computers to produce correlations based on enormous sets of data not usually accessible to a human mind. The speed with which these correlations can be made and the fact that they are made on the basis of algorithms designed by a human person, brings to mind a tortoise slowly following its path: ‘I may be going slowly, but then I may be going in the wrong direction’. However fast a computer works and however complex the correlations found, a computer can do nothing other than apply a code written by a human person.  

The interesting question is perhaps not whether computers can think or whether they are intelligent, but in which way they think differently (or their intelligence works differently) and what this should mean for the application of computer technologies. 

Personalisation

In principle, AmI technologies could be used to create a responsive environment for animals or even plants, allowing a reduction in the cost of human intervention, for example, in cattle-breeding (AmI in the stable) or horticulture (AmI in the greenhouse). One could, for instance, imagine a Smart Home that provides personalised services for a cat that is left alone for some days (regulating temperature, providing the right amount of food in time, maybe some robotic device to clean the litter tray). In this case we are not talking about personalisation, as the environment will only have to deal with a set of plants or animals, not persons.  

In this deliverable we will focus on AmI as a technology to personalise the environment of individual human persons. This personalisation can consist of providing all kinds of services without an explicit action on the part of the ‘customer’, but it can also consist of taking over routine decision-making processes, thus reducing the amount of choices to be made without loosing out on interesting opportunities. The point here is that the intelligent device continuously constructs and reconstructs behavioural profiles that indicate certain preferences, enabling the construction of a responsive environment that seems to know one’s preferences before they have surfaced in consciousness.  

The human-centred approach and the concept of personalised services is important because it indicates that the AmI system needs a never-ending stream of personal data to (re) construct personal profiles that indicate what people want and to provide them with customised services. Thus in order to provide personalised goods and services, AmI needs to be provided with personal information. The more (useful) personal information the AmI system has access to, the more personalised (and intelligent or enhanced) its services can be. It should be noted that precisely this aspect of AmI raises questions about privacy and security.