Sunday, November 8, 2015

Google's Smart E-mail

Admittedly, I don't watch the news very often. But I was on the subway this past week and while I was sitting there I saw a quick glimpse of a segment about Google's new feature they are launching in Gmail that shocked me.

As we have discussed a lot in class, we really do not know how much of our internet activity is being monitored. However, with this new feature called "smart-reply" we are able to see that every email that goes in and out of our inbox is being read and interpreted by Google. Let me give you a quick synopsis of what I mean.

The second you receive an email, Google analyzes the text and will automatically offer the receiver three tailored responses based on the text within the incoming email. Of course, you can also choose to customize the message. This type of technology requires something called "deep learning." Or in other words, artificial intelligence. Google has been using this deep learning for a while to filter out spam emails identifies photo subjects, translates text etc.

The obvious issue with this is the question over the amount of privacy we are given as users of Google. How do we know what they are doing with this information? If Google is able to see all of our emails in order to create their own responses, that means they have access to information most users forget is not just theirs. Many articles I have read have been promoting this new feature saying that it will be able to really help users because typing your own responses to your own emails is so difficult and time consuming that you would need a computer to generate a response for you. I personally can't see any benefits of smart-reply other than the benefits offered to Google. They are able to keep a close watch on its users activity and who really knows where this information will go from there?

And to make matters worse, this feature is also able to take into consideration human interactions with their emails. For example, it can better understand the user by the responses the person chooses or doesn't choose. Eventually they will utilize a fully operating machine-learning system being called "deep neural networks." This is basically a program that allows a computer to absorb observational data and analyze it for other purposes. But how far will this program go to collect data on us?

No comments:

Post a Comment