October 3, 2019 Rodney Cowled
Smart Dumb Home

Your Smart Home is Really Dumb


It’s 10pm on a Thursday and I’ve just turned off my bedroom light.

When I turn off my bedroom light after 10pm on a weeknight, it’s highly likely I’m about to go to sleep – Imagine if my smart home could work this out itself, do a quick status check of my devices, & point out anything that’s out of the norm. 

For example, it’s after my usual bedtime and I’ve asked Alexa to turn off my bedroom light. A moment before the light goes off, Alexa asks “The front door is currently unlocked, would you like me to lock it? 

My smart home has noticed a pattern. My front door is usually locked when I turn off my bedroom light after a set time on a weeknight. This time however the door isn’t locked, so the voice component of my home brings it to my attention and asks if I’d like to take action.   

Imagine a home…

A home which is able to adjust schedules, routines & scenes to suit the changing environment. A smart home which is programmed with desires rather than with specific parameters. Would it work? 

Rather than programming “Turn on X light when I arrive home, but only after 5:30pm & before 8am.” You instead, say “I don’t want it to be dark when I get home” (or, you don’t say anything at all, your home takes care of it based on your patterns). 

But how far is too far? 

How far do we take it? What kind of decision can a smart home make on its own? Should our home simply bring something to our attention or should it just act? 

Where we draw this line will likely move over time.

Think how we’ve changed our opinions on whats acceptable in this post-shared-economy world. Previously, you’d never have considered jumping into a strangers car, but with uber, it has become common place.

Where we draw the line today, will likely be considered very conservative tomorrow. 

What could it look like? 

Alexa: “Hey, I’ve just noticed X, is this unusual?” 

Me: No. 

Alexa: analyses my response and weighs the result next time a similar scenario appears. 

Alexa: “Hey, I’ve just noticed Y, should I do something about it?” 

Me: Yes. 

Alexa: Again, analyses the response and weighs this result next time a similar scenario appears. 

With time, Alexa* could become the keeper of your home. 

*substitute your favourite smart home assistant, or one that’s not even on the market yet.

How would it work? Direct or Implicit Feedback.

Direct Explicit Feedback

Above example is an example of direct feedback, a response loop. The voice assistance asks “Did I do the right thing?” you then respond with “yes” or “no”. The voice assistant then uses that feedback to adjust its responses going forward.

This method of feedback is effective, it’s clear, it’s explicit…but it’s also annoying.

No one wants to do a (tiny) survey every time they turn on a light or listen to the radio.

Another option is implicit feedback.

According to Rohit Prasad, Vice President and Head Scientist – Alexa Artificial Intelligence. Alexa launched self-learning, which lets the system automatically make corrections based on context clues.

Prasad provides an example: Say you ask your Echo to “play XM Chill,” and the request fails because Alexa doesn’t catalogue the station that way. If you follow up by saying “play Sirius channel 53,” and continuing listening, Alexa will learn that XM Chill and Sirius channel 53 are the same, all on its own. “That’s a big deal for AI systems,” says Prasad. “This is where it’s learning from implicit feedback.”

Alexa learns from implicit feedback, and it is great.

Here’s my personal experience & how alexa adjusted her responses to react to our personal smart home setup.

Me: “Alexa, Turn off the stair pendent”

Alexa: “I can’t find a device called stair panda” 

Wife: *chuckling*

Me: “Alexa, turn off the [email protected]#$^ STAIR PENDENT”

Alexa: “ok”

Alexa no longer tells me she can’t find my “stair panda”. When she hears “stair panda” she knows I mean “stair pendent” & reacts accordingly.

Ultimate Personalisation

Overtime, the sum total of our responses and data on our device activity could inform the decisions of your smart home. Your smart home, would be engaged in an endless study of your habits, asking for your feedback – one subtle question at a time.

It could become the ultimate in personal customisation without selecting a single parameter. Well, at least without realising we’re selecting them.

Challenges and Creepiness

As your smart home assistant embarks on this endless journey of endless customisation, there could be some speed bumps. Likely, there will be some awkward conversations too. 

The success of a shift like this will come down to our management of the creepiness-factor, and our ability to ensure that the benefits far out way the intrusion of the research & feedback questions.

The success of a shift like this will come down to our management of the creepiness-factor

1. Awkward Interruptions

It’s middle of the night, Alexa asks if you’d like the kitchen light turned off. Yes Alexa, I want the light turned off, but I do NOT want to be woken up at 2am to check if I want it turned off. It’s just not THAT important.

2. Creepiness Plus.

Your assistant notices a habit you didn’t even know existed. 

  • Alexa: “ Hi Rod, it’s 11pm and I’ve not noticed motion in the kitchen. Also, you’re almost out of chocolate, should I order more?”
  • Me: *Thinks to self* I’ve got a chocolate addiction and a late-night snacking problem. Even Alexa’s noticed…let’s hope she doesn’t tell my wife. 

What habits & routines could our smart assistants notice, that we’d prefer them not to notice? Let me know your thoughts.

3. Catering for Changing Routines.

Our routines change over time, we take up new hobbies, start new jobs, kids grow up. How do we cater for these changing routines?

How does Alexa know to ignore 2 years of history in favour of this new scenario which has only occurred for the last week?

4. Catering for one-offs.

Your sister is arriving from the airport late tonight, so you decide to leave the front door unlocked and the hall light on. An important question would be how we cater for these one-off changes? 

We don’t want Alexa to automatically lock the door. We also don’t want Alexa to leave the door unlocked tomorrow night. 

What are do we do? Perhaps, Alexa could ask clarifying questions. E.g. “I’ve noticed the doors unlocked, is this a one off or should I consider this going forward?”

Dumb Home to Smart Home

Simply put, your smart home is a connected home. It doesn’t think, it doesn’t learn – is it really smart?

This isn’t some far off future filled with flying ubers (wink). This doesn’t require some sort of big technological breakthrough. We have the technology, we can build it. It’s the same technology that behind your Netflix recommendations.

Don’t get me wrong, machine learning/deep learning technology is being used in the smart home. Amazon uses it & Google too. The implied feedback mentioned before is an example of this. 

Rather, this is a vision for a smart(er) home. One more deeply focused on your smart home assistant becoming the brain of your smart home, rather than just the messenger.


Post Script: Alexa hunches, I’ve not noticed any of its functionality. Have you?

Tagged: , , ,