Why Don’t Elevators Listen?

We’ve all done it. You walk into an elevator at the ground floor. You press 5. You wait a couple of seconds. Someone else walks in. They press 7. You quickly glance at one another. You press 5 again. A few seconds go by. You check your watch and with a sigh, you press 5 a couple more times. After a brief pause, the elevator finally complies. The doors close.

When you press an elevator button, the elevator talks back — sometimes. The button lights up. The elevator says “Ok. I’ll stop at 5 for you.” But the dialog ends there. Other people will get on and they can say their piece as well. But that’s it.

We talk to machines all the time. And they are getting more and more sophisticated in terms of the tasks they’re capable of completing. But where machines still fall short is in the breadth of dialogue they’re willing to embark upon with us. Better yet, why can’t machines pay more attention in general?

Consider the following questions:

  • Why can’t the elevator realize that after a minute or so of me pressing the same floor and nobody else coming on and pressing any other floors that it’s time to go?
  • Why can’t the button illuminate a bit brighter each time I press it, thus signaling to me that it is listening?
  • Why can’t the elevator keep track of the patterns of the typical work day (the same way trains and buses do with schedules)? “Rush hour” happens in the morning and late afternoon so it should behave differently at those times. At 3AM, it should immediately respond to a single request.

The technology to do all this is not only there but pretty bare-bones. This logic can all sit on a single chip. The hurdles are clearly not technical but rather conceptual. Thinking about machines as beings that pay attention isn’t typically baked into your typical elevator business requirements. The weight of the payload. The time of the day. The days of the week (weekends vs. weekdays). The patterns of use. And of course, the odd patterns and habits of passengers as they click away at those buttons. 5. Door Close. 5. 5. Door Close.

As people on the other end of this dialogue, we can’t help but feel like we are talking to this machine – and that it’s listening. What we should do, as technologists and product designers, is aid these machines with the ability to listen, pay attention and let people know that they are paying attention.

Ever walk into a shop and buy some gum or soda and the shopkeeper doesn’t even acknowledge you or make eye contact? Instead of handing you the change, he tosses it on the counter and says nothing. Now you may not be trying to make friends, but it’s not a very good feeling. We love to be acknowledged and recognized. The more interactive machines are, the more of an affinity and loyalty we’re willing to build towards them. Is there a business case for smarter elevators? Possibly. But there’s a value to all this that is beyond efficiency and algorithms.

Ultimately, interaction design is about people. Today, people’s lives are filled with stress and anxiety. In our work lives, we’re often subjected to these impersonal routines that can eat away at us. This isn’t about improving the IQ of elevators. There are plenty of really smart people with high IQ’s that are really rude.

This is about making elevators a little smarter…and a little more thoughtful.

4 Comments Why Don’t Elevators Listen?

  1. Josh Tynjala

    Some elevators actually are kind of “smart”. I’ve heard of some that will automatically return to the ground floor in the morning if they aren’t currently in use so that people won’t have to wait when they arrive to work. Likewise, the elevator will go back to higher floors in the evening to be more accessible.
    I can’t imagine it would be much more difficult for it to remember (or have someone program in) other traffic patterns.

  2. Amy Hoy

    As much as I don’t like the elevators in the building where my company leases office space (they’re kind of creepy and bumpy and have been out of service numerous times), they’re pretty “considerate” as far as things go.
    They have the habit of opening automatically when you walk into the area near them. I don’t believe they’re tied into any kind of motion detector system; must be some kind of programming based on times of day and usage patterns.
    On the one hand, it’s very strange to start walking towards an elevator and have it begin yawning open for you without touching anything. On the other hand, once you get over the strange, it’s really convenient.

  3. Tom

    You’ve hit upon the problem of moving between the hardware world and the software world. If your UI isn’t quite elegant enough, you can release a patch and two minutes later, a whole new UI appears. Commercial elevators for human transport have been around for over 100 years (and in NYC, you see some that appear to be almost that old). If your elevator already has the chip in it to send it to the right floors at the right times (and what the hell, it may as well even analyze traffic patterns over time and refine its logic), then updating the software with improved logic is a no brainer. Hell, you could connect it to RFID employee cards so that it knows who’s on the elevator.
    But for the people in the elevator business, they need to justify the investment of retrofitting exisitng elevators. Now how does the amount of improvement you would get from a “smart” elevator balance against the cost of the improvement (and even worse, against the increased maintenance costs associated with additional machinery and code? Not very well, I suspect.

  4. Sam Musleh

    Don’t blame the elevator unless it’s 20+ years old, I suppose. Blame the guys setting it up. I run the software R&D for a major US elevator controls company and I can tell you that almost all of the smart features you’ve been pining for in the blog above and subsequent comments, and then some, have all been developed as part of our iControl elevator controller. You wouldn’t believe the sophistication and “brains” behind how that elevator truly operates.
    Our controller can detect how many people are in the car so that if a “Bart Simpson” presses all the floor buttons, the prank will be immediately detected and the car calls will be canceled. If someone on a maternity ward tries to take a baby into the stairway or onto an elevator and triggers the alarm, all elevators at that floor will not move and elevators coming to that floor will just keep going on past. Our dispatcher will “learn” the traffic patterns throughout the day in your building and will automatically switch to Lobby-peak mode in the morning arrival rush then switch to down-peak mode when the majority of building tenants take coffee and lunch breaks, etc, and will use artificial intelligence to determine where is the best place to park all cars for fastest response for every hour throughout the day! And yes, they can be set – and usually are – to park at the lobby in the mornings.
    Long dwell times for example, the time that the elevator sits there after you’ve entered and nobody else has, are set aggressively in most cases… until a tenant complains that they didn’t reach the elevator before it closed. If this tenant is important enough, settings get changed and often aren’t changed back unless someone else complains.
    Building managers and elevator contractors are always looking for feedback on ways to make the elevator system more efficient and useful. The key is to be the squeaky wheel. Until someone tells you that your wild wish is beyond the elevator system’s capability, trust me, my team has probably thought of it and built it into our system.


Leave a Reply

Your email address will not be published. Required fields are marked *