All the major tech companies have their voice assistants. Google has Google Assistant, Amazon has Alexa, Apple has Siri, Samsung has Bixby and Microsoft has Cortana. And other companies are also experimenting with these voice assistants.
When the voice assistants initially arrived, they were primarily made for smartphones. It was to make the smartphone actually smarter and voice-driven.
Fast-forward to now, voice assistants power smart speakers, which are now becoming mainstream and common electronic devices in homes.
But it didn’t stop there. Now, these voice assistants are making their way into streaming devices, headphones, TVs, electronic appliances like refrigerators, microwaves, AC, doorbells, bulbs, and even simple things such as switches.
And even though they slap AI-powered tag on all these devices, essentially these are just microphones that connect to the web. So when you speak to your Echo Dot with Alexa, there’s no SoC on these speakers to carry complex NLP algorithms. Instead, your voice is sent over to Amazon servers, which have hardware capabilities to process your voice inputs and send responses back to your speaker.
So why smart speakers are being pushed so aggressively
Voice data has a lot more nuance and more personal, as it varies with a person’s mood and situation. Thus it is a lot valuable and useful for companies to understand user demographics and use them for better and more personalized ad targeting.
That’s why Facebook which is actually a social media business, bought Oculus for $2B dollars, which makes AR/VR headsets. And we all know how Meta( Facebook) is more of an advertising platform, like Google.
And both Google and Amazon devices were caught listening all the time, even when the device wasn’t triggered with their voice commands. Even though they shrugged it off as a bug, the device is still capable of listening.
These devices also collect location data which can also be a tool for mass surveillance.
Training datasets for improving accuracy
If you’re familiar with natural language processing, you might know that training your algorithm with a wide range of voice data is necessary to make the algorithm more accurate. And this voice data is hard to come by for these companies. And it’s expensive too, to pay people to record their voices. So these tech companies took the easier and deceptive route of using regular consumer data to use as training data sets.
And if you go over the settings of these voice assistants, they opt you in to save your recordings in their servers by default. And you have to request for your voice data to be deleted or not to be used for targeted ads.
Locking into the ecosystem
Once you’re accustomed to a specific voice assistant, you would buy smart devices that actually are built with this assistant. Apple is one company that does this very well. Siri powers every device that Apple makes, so people tend to buy Apple products more for that seamless experience across devices.
That’s the reason why Amazon is flooding the market with Alexa devices like Fire tv sticks, smart speakers, smart speakers with screens, smart plugs and it just goes on.