(Source: Shutterstock)
For all of the trouble and strife inflicted by the pandemic, it’s proven a powerful catalyst for the adoption and advancement of technology. With lockdowns confining many to their homes, consumers chose to invest in the environment that they could control, looking to bring quality of life changes into the home with the money once reserved for experiences and luxuries that, at the time, were unobtainable.
With that in mind, we may look back on the restrictions of the pandemic as the genesis of the smart home movement. A survey conducted by Voicebot Research reveals that over 38% of UK adults owned a smart speaker at the start of 2021, overtaking U.S. adoption, and Ampere reports that more than half (52%) of internet households now own one. That’s a lot of units.
These speakers are the pioneers of smart home technology, teaching their users how simple voice interaction can be, and how verbal commands can impact other devices.
However, they can only do so by listening to you. All the time.
Smart speakers are using 2-4 W of power to idle in the heart of your home, monitoring every conversation within earshot. It’s not purely a case of listening for keywords to bring the speaker to life. Smart speakers are digitizing and then commodifying the things you speak about, recorded in crystalline quality.
If you think that’s an over exaggeration, Alexa recordings have been brought as evidence for murder cases in New Hampshire and Florida – such is the sophistication and quality of its recordings.
For many, there is a balance between being recognized and being monitored that goes beyond what they would be comfortable with in almost any other setting. You wouldn’t, for example, tolerate a stranger eavesdropping on private conversations, let alone if they interrupted that conversation to try and sell you something you’d mentioned.
But such is the passivity of the smart speaker, and the primitive understanding of consumer data that many of us have, that convenience trumps diligence. The price of that is our data.
Why Fi?
Once that data is beamed to the cloud, it’s no longer yours. As the terms and conditions of the Alexa Communication Schedule state:
“Amazon processes and retains your messages in the cloud to provide you with the service, including speech to text transcription and vice versa, to provide additional functionality, and to improve the quality of our services.”
“Certain Alexa Communication services are provided by our third-party service providers, and we may provide them with information, such as telephone numbers, to provide those services.”
That transmission to cloud usually happens via Wi-Fi, which invites a host of well-documented security issues in itself – especially in an age where more people than ever are working from home.
On the one hand, we want to be recognized as individuals by the smart home systems that understand our preferences and behaviors. But on the other, we don’t want to be perpetually observed by a device that can commercialize every word with relative impunity.
Ideally, we’d be able to rely on smart home devices that aren’t required to transmit data beyond the smart home environment at all. Rather than running constantly at a low wattage, recording every detail just in case a keyword is spoken, we should be working towards more privacy-friendly, energy-efficient devices that react only to specific commands as they’re spoken.
So, how do we do that?
Use your brain
The key is to make devices more intelligent in and of themselves, rather than incorporating a dependency on cloud networks to interpret and action commands on the device’s behalf.
Many modern smart speakers rely heavily upon these networks. Instructions are identified by the speaker and then bounced to a cloud network for contextualization – and recording. The content of the command, from a particular service to vocal patterns, are recorded as a reflection of your interests and behavior as the command is actioned.
That’s the norm, but it’s not necessary. The artificial intelligence of things (AIoT) market is advancing to the point that processors are capable of interpreting and actioning commands locally, on device, rather than being beholden to the cumbersome process of cloud-driven interpretation.
If a device could turn word into deed without relying upon the cloud, it would represent a huge advancement in consumer privacy. Deprioritizing cloud interactions drastically reduces the need to transmit data outside of the smart home environment, protecting the user from invasive data capture.
Sixth Sense (or)
The standard counterargument to this privacy is an inability to personalize your experience. How smart can a speaker be if it doesn’t even know who you are?
Fortunately, you don’t need a cloud account for more private devices to recognize you. The sensor array incorporated into smart speakers (and other smart appliances) provide an alternative for the minutia within personal data that normally informs how such devices treat you.
As a simple example, visual and audio sensors are capable of determining age without reference to the cloud. Smart devices could thereby distinguish between parents and children in a family home, ignoring commands from children to turn the oven on, or to order “beer for daddy.”
Moreover, if local intelligent sensors on device can coexist with cloud services, they’re still able to maintain privacy. For example, devices can send metadata to the cloud rather than raw data, capturing key information or features that trigger the appropriate response from a smart device without sharing the exact information or origin of the data.
Looking forwards
In the longer term, the smart home will eventually evolve into an environment where the sensors from different devices can inform wider action. You can imagine visual and sound sensors being able to observe a room in which someone injures themselves in a fall, alerting the emergency services as a result.
There’s also the potential to move away from capture-intensive search engines to interpret voice commands, adding another layer of privacy. DuckDuckGo, for example, is a search engine renowned for its user protection: it doesn’t track IP addresses and so doesn’t deliver targeted ads or capture personal info.
Ultimately, these ambitions require devices that have the intelligence and the collective sensor array to paint that picture. The shift towards privacy, prioritizing sensor suitability over data collection, may well be the start of that paradigm shift.
Private matters
We do not need to sell our digital souls to be recognized as individuals. The smart home environment is currently too dependent on devices that capture, store, and use your data to understand you, at the cost of offering or selling that data to stores and services.
Challenging that zeitgeist not only demonstrates a genuine respect of consumer privacy, but it also makes devices cheaper to run and less environmentally damaging. Given that aforementioned research from Voicebot reported around 20,000,000 smart speakers in the UK alone, you can imagine that these devices produce an immense amount of CO2 just to idle over and observe your every conversation.
about XMOS