Wikifriend

Today I was answering a question on Quora, and I think that could be a good share :)

The question was: Does the Amazon Echo Dot really spy on your conversations?

The answer to this question is not as simple as it can seem. One thing that can be easily verified is just googling about it, there are many authoritative articles about this topic, like this one:

Amazon admits employees listen to audio from Echo devices

But I would focus more on something more objective than this. I suggest to examine the actual technology on the market and the way the companies in this field use this technology and why.

Today we hear talking about the benefits of using cloud technologies but is not really clear for who is the benefit 😊

Most people think that giving their personal data to “any” provider offering a free and good service is a normal compromise. Who cares that a free email provider that gives me 1 tera of free space, uses my personal data to profile me and create targeted advertisement? Probably this sounds quite reasonable. But the point is that our personal data is not used just for that. Actually, we don’t know how our data is used. The only thing we know for sure is that we are deeply profiled on these cloud services.

So now, to come to the question: Why smart speakers use cloud services instead of on-premises one?

Amazon, like Google and the other brands producing smart speakers, don’t provide this technology because they have a revenue by selling it, instead the real value consists in the data they can acquire. So, there is no valuable reason for them to provide this technology running locally. That is the benefit of the cloud! (for them….)

If we look it with these terms, the answer is definitively YES.

Smart devices are always listening, to improve…. This is called deep learning… again, learning what exactly? All about us!

What is good, is that people today start to care about this topic: GDPR in Europe is a clear consequence of this awareness. The new thread in smart services is privacy-by-design and privacy-by-default. On-premises is the new request. ASR, NLP, TTS are now returning to run locally, at least for those companies that put the user privacy at the first position.

Of course, this topic is a very sensitive issue, and Wikifriend is complaining about this since many years now. What we are trying to let people be aware of, is that our personal data (and all the data related to us like for example our children’s data) can be used by companies like insurance, HR departments of our workplace, government organizations and so on, to take critical decisions regarding our future or the future of our relatives. To make it clearer, let’s think about a serious disease in the family, and that at home there is a discussion about that and about the right insurance to choose, to handle the future therapies… What value could have this information for the insurance companies? I made it very simple here, but I hope this helps to understand how critical is to underestimate the consequences of a lack of privacy at home.

Another good example is given by this Wikipedia article: https://en.wikipedia.org/wiki/XKeyscore

As I said, this is a really complex topic, just answering YES is not enough.