Home / Blog / AI is becoming a marketing word and it’s a pity -VII

AI is becoming a marketing word and it’s a pity -VII

The tracking of power consumption through a digital smart meter can be used to feed Machine Learning software creating a digital signature of the home. This in turns will be used to detect when an appliance is operating and how it is operating. Image credit: AI and Machine Learning – Google Cloud

Home Intelligence

If there is an area where AI hype is in full swing that is the home environment. We are bombarded by intelligent vacuum cleaners, intelligent microwave ovens,… even intelligent toobrush.

True to tell, some advertisers downplay the intelligence a bit and talk about “smart”: the smart kitchen, the smart air conditioning,…

As a matter of fact a home, thanks to the growing number of IoTs inhabiting it permanently, in appliances and home infrastructures, or occasionally, being worn by people staying or visiting the home, is the source of a growing number and variety of data. Not “big data” as we usually define them but nevertheless quite a bit of them.

Are these data enough to have intelligence emerging? Do we have sufficient processing power to make this intelligence emerge? Yes and no.

If we take a snapshot of these data there is very little meaning that can be extracted and that can give rise to an intelligent “behaviour”. On the other hand if there is the possibility of following the dynamic change of those data AND comparing them with other sources of data then we might have just enough raw material to work on.

Several appliances manufacturers, particularly those producing white goods, have started to accrue the data provided by an appliance as it operates and track the changes resulting from change in operation environment and from problems, fatigue – degrade- of the various parts of the appliance. They have also been able to generate a unique pattern identifying the appliance and its different ways of operation. This is usually called the appliance digital signature. As an example, the power consumption of the appliance provides hints on what is going on and on the type of the appliance. It is called digital signature because it is indifferent to external conditions meaning that a pattern uniquely identifying that type of appliance can be recognised. More than that. Over time this unique “type” pattern can grow into a unique pattern of that specific appliance, of that instance making it possible to distinguish that specific appliance from another of the same type.

This pattern extraction and instantiation is achieved through machine learning software, through artificial intelligence. However, notice that the intelligence is not “inside” the appliance.

A washing machine having gone through the process of creating a unique pattern, a digital signature, may be more or less smart but it is not intelligent. However that washing machine by sharing its operational data can enable an external application to detect signs of possible issues. This external application can indeed be an intelligent one.

This is what is pointed out in the figure, created by Google researchers. By looking at the power consumption patterns in a home (accessing the smart digital meter data) they have been able to train an application to identify different types of appliances in a home and looking at their working the application can come up with “intelligent” suggestion.

I took here the example of the home, mostly because this is the area where we are bombarded by ads emphasising the “intelligence” of this and that, but the basic idea that through the harvesting of many data over long period of time we can train, through machine learning an application to derive meaning on what is going on has a very general application, in industry on the manufacturing plan, in a smart city, in health care …

The concept of digital signature is a very important one for initiating analyses and having intelligence emerging.

So, in my opinion we do not have “intelligent appliances” today (although we have some very very smart ones like the Roombas’ family that can learn the lay of the land in your home and calculate when it is better to vacuum clean it…) and we may not have any of them in the next decade. Further down the lane it may become possible to have local intelligence. However, we can have an intelligent service through an appliance. Further releases of Alexa and the like, through Cloud processing and data analyses, will become more and more intelligent. This will also depend in how much we are willing to share with them (and with the cloud). In the end it is not that different from you and me. If we talk to one another and share knowledge and experience each of us will improve his intelligence. The only difference is that our inference engine is embedded in our skull, for our appliances for a few more years it will have to float in some clouds.

About Roberto Saracco

Roberto Saracco fell in love with technology and its implications long time ago. His background is in math and computer science. Until April 2017 he led the EIT Digital Italian Node and then was head of the Industrial Doctoral School of EIT Digital up to September 2018. Previously, up to December 2011 he was the Director of the Telecom Italia Future Centre in Venice, looking at the interplay of technology evolution, economics and society. At the turn of the century he led a World Bank-Infodev project to stimulate entrepreneurship in Latin America. He is a senior member of IEEE where he leads the Industry Advisory Board within the Future Directions Committee and co-chairs the Digital Reality Initiative. He teaches a Master course on Technology Forecasting and Market impact at the University of Trento. He has published over 100 papers in journals and magazines and 14 books.