Issue 28

It's not just Alexa listening

Delivered on 16 April 2019 by Justin Pyvis. About a 5 min read.

In news that really should have surprised no one, Bloomberg revealed that audio captured by Amazon Alexa-enabled devices is actually listened to by humans:

"The team listens to voice recordings captured in Echo owners’ homes and offices. The recordings are transcribed, annotated and then fed back into the software as part of an effort to eliminate gaps in Alexa’s understanding of human speech and help it better respond to commands... A screenshot reviewed by Bloomberg shows that the recordings sent to the Alexa reviewers don’t provide a user’s full name and address but are associated with an account number, as well as the user’s first name and the device’s serial number."

So much for artificial intelligence (AI) and machine learning. As I wrote last week, these things can't actually understand anything; they use algorithms to scan an enormous database and pick a response based on the probability that it's correct. Humans will always be needed to improve those algorithms given that language constantly evolves:

"But sometimes Alexa gets it wrong—especially when grappling with new slang, regional colloquialisms or languages other than English. In French, avec sa, “with his” or “with her,” can confuse the software into thinking someone is using the Alexa wake word. Hecho, Spanish for a fact or deed, is sometimes misinterpreted as Echo. And so on. That’s why Amazon recruited human helpers to fill in the gaps missed by the algorithms."

Terms like machine learning and artificial intelligence are deceptive in that "learning" and "intelligence" are misdefined. There's no learning or intelligence happening, at least as the average person would understand those words to mean. These algorithms are giant probability machines, with humans doing the learning for them by informing the algorithm whether a response was appropriate or not, effectively tweaking and redefining its probability model.

Artificial intelligence and machine learning are the result of brilliant marketing, and the probability models that they use are fantastic. But as we've now discovered with Amazon (Apple's Siri also uses humans in the same way), it has created a bit of an issue in terms of privacy - people don't think that other people are listening in on their conversations:

"You don’t necessarily think of another human listening to what you’re telling your smart speaker in the intimacy of your home," said Florian Schaub, a professor at the University of Michigan who has researched privacy issues related to smart speakers.

"I think we’ve been conditioned to the [assumption] that these machines are just doing magic machine learning. But the fact is there is still manual processing involved."

“Whether that’s a privacy concern or not depends on how cautious Amazon and other companies are in what type of information they have manually annotated, and how they present that information to someone,” he added.

If the constant data breaches at Facebook (and many other companies / governments) have taught us anything, large central databases filled with a diverse set of user information is a honey pot just begging to be hacked. If you absolutely must have a voice activated assistant to play your music or set an alarm, consider privacy-friendly alternatives such as Mycroft or Snips, which give you full control over your data.

Artificial intelligence. Really?

I'm not against artificial intelligence, although I detest the name. Likewise machine learning. The marketing of those two words has gone too far, and as a result so have the expectations. It just needs to be dialled back a bit and its limitations acknowledged.

Learn more:

Blockchain's back (it never left)

While artificial intelligence steals the spotlight, blockchain will continue to do its thing in the background. Crypto prices have actually been quite stable for about six months now, which is a good thing.

Learn more:

Regulating big tech is a bad idea

Everyone wants to regulate big tech because they're monopolies. But they're really not even close (check our archives for why; I've written extensively on the subject). From the top article linked below:

"It is hard to come up with any sectors that have been more consistently innovative than what we call Big Tech, and that’s without even mentioning Amazon and Apple, two major players that also would be affected by Ms. Warren’s breakup plan. Yes, there is a good case for reform at the edges – but let’s not shoot ourselves in the foot."

Learn more:

Unintended consequences of privacy regulation

The European Union is leading the rest of the world on this front so it's wise to pay attention. Especially when members of that Union are falsely flagging hundreds of innocent websites as containing "Terrorist Content".

"The European Parliament is set to vote on legislation that would require websites that host user-generated content to take down material reported as terrorist content within one hour. We have some examples of current notices sent to the Internet Archive that we think illustrate very well why this requirement would be harmful to the free sharing of information and freedom of speech that the European Union pledges to safeguard."

Learn more:

Other bits of interest

Image of the week

Do people care about their privacy, or owning their data? Not according to this survey (usual 'pinch of salt' survey caveat applies):

"The share of U.S. adults who say they use certain online platforms or apps is statistically unchanged from where it stood in early 2018 despite a long stretch of controversies over privacy, fake news and censorship on social media, according to a new Pew Research Center survey conducted Jan. 8 to Feb. 7, 2019."

Issue 28: It's not just Alexa listening was compiled by Justin Pyvis and delivered on 16 April 2019. Join the conversation on the fediverse at