Spiria logo.

Machine Learning, or Elevator Talk

March 16, 2017.
Machine Learning, or automatic learning, is a buzzword that rhymes with science-fiction for an entire generation of us who still remember Hal, of 2001: A Space Odyssey, the computer blessed with a superior intelligence and consciousness that were equal parts fascinating and terrifying.

The fact is, automatic learning is a close relative of artificial intelligence, which studies the way computers execute algorithms on existing data to create new models. These models in turn predict or make decisions based on reams of new data.

The learning component comes in when you slip a factor of exploration, or randomness, into the flow of algorithms executed by the computer. Due to the large masses of data and variables involved, curveball input often yields unexpected results, providing new learning opportunities thanks to new predictive models.

The Wonderful World of Elevators

Kone, the Finnish escalator and elevator company, has partnered up with IBM to launch 24/7 Connected Services, a new initiative to explore the full potential of its famous artificial intelligence platform, Watson IoT.

Kone maintains and operates over one million elevators and escalators all over the world. It plans to connect its entire fleet of devices with multiple sensors to analyze user behaviour, detect and predict malfunctions, track frequency of use, and monitor temperature and humidity in all of its lifts, among other things. And by compiling comparable data from hundreds of thousands or even millions of lifts, it hopes to eventually minimize the number of breakdowns and more quickly solve any issues that do arise.

Better yet, the company’s customers will also be able to track the operation of their own elevators and escalators in real time. For example, London’s Heathrow airport is equipped with over 1,000 Kone elevators and escalators, used by over 190,000 people daily. Not surprisingly, the failure of even a single one of the devices has a huge impact in terms of delays and congestion. Kone hopes to enhance overall client experience through predictive failure models obtained through automatic learning. The data obtained will be used not just to prevent breakdowns but also to glean information on the fleet as a whole, for example which lifts are the most heavily used.

Elevator Talk

To demonstrate Watson’s capabilities, Kone has created a Web site where you can listen to elevators from all over the world talk to Kone servers on the cloud.

“Kone initiated ‘machine conversations’ as a way to help others understand how machines may talk to each other, all in the spirit of creating a better customer and building-owner experience,” said Danilo Elez, Kone’s senior vice-president in charge of service for the company’s American branch. “When our elevators share information about their performance with IBM Watson, we’re working to improve their availability and ride quality in real time.”

Unsurprisingly, elevator talk is rather dull, but pretty creepy at the same time. It provides a glimpse of a not-so-distant future when all machines will chat amongst themselves to learn more about one other.

Fascinating or terrifying? You decide.