Do companies need to have humans sample your voice translations?

Seems that companies are doubling down on their position that they need to have humans listen in on your private conversations (at least in language translations) to improve the accuracy. Microsoft has doubled down on that by just amending its terms of service to be more transparent. Google Apple, Amazon and Facebook all have listened in. Apple is amending it’s eavesdropping on voice translation hoping that folks will opt-in. At least they are asking permission. And it may be necessary particularly in language translation. Machine learning maybe not that smart yet and needs to be tweaked by humans. Do you think all companies should move to opt-in in their privacy practices? What do you think Alexa?

Interesting to find that some people are surprised to learn that “they” are listening to private conversations without explicit approval; “they” being the large tech companies listed above. Much of what we view as ‘privacy’ has already been given away when we accept the many EULA’s from our phones, downloaded apps, access to our contacts & photos, etc. Machine learning progresses faster if humans perform some “tweaking”, such as listening to conversations, and then correction. It would be nice to have an opt-in/out for this, but my experience is that folks don’t really care about their info being collected. Most say they care about their personal privacy, but how many put that into action?

Nice to meet you, Brian. Thank you for your thoughtful post. Most people who aren’t in the tech business don’t have a clue about AI and machine language training. They just want it to work inside a magic black box. They’ve heard of Artificial Intelligence through the media or movies. I’ve been a fan of speech recognition since the early days and I knew the lead developer of Dragon Dictate. I didn’t think twice about using it in those days but of course, it must have had to be trained extensively with humans listening and tweaking it all the way along. And your right, machine learning gets better with human tweaking and it gets more complicated and less accurate with translation from one language to another language.

The Cambridge Analytica scandal really turned up the heat on privacy, the collection of data points and the potential for abuse. The heat in the pot got so hot the proverbial frog jumped out. And maybe that was a good thing. It should be obvious now to a lot of people that we are part of the product. EULAs and Terms of Service have always been hard to read so GDPR makes an effort to make that simpler. And at least Apps are now telling you more about what information they collect. We haven’t even gotten into the subject of endpoint security in voice assistants!


Featured Webinars


Advanced Phishing and
Training

Monday 1:30 PM – 2:30 PM
» Learn More
Outlook Phish Alert Button
Tuesday 1:30 PM – 2:30 PM
» Learn More
Customizing Phishing Templates, Landing Pages, & Training Notifications
Wednesday 1:30 PM – 2:30 PM
» Learn More
Active Directory Integration
(ADI) Setup

Thursday 1:30 PM – 2:30 PM
» Learn More
Gold/Platinum/Diamond
Features

Friday 1:30 PM – 2:30 PM
» Learn More

Privacy Policy | Terms of Service