We Tested Google’s Censorship Bot… Here Are 27 Frightening Examples
By Ben Marquis | Conservative Tribune
Aug 2, 2017
Click here

[snip]

Anybody who has spent any amount of time in the comments section of an online article knows that conversations can quickly become heated with abusive rhetoric and vitriolic remarks, often running readers off and stifling debate of differing opinions.

As such, the folks at Jigsaw and Google’s Counter Abuse Technology team have developed an experimental add-on program known as “Perspective” for various sites to use with their comment sections as way to better monitor and moderate the conversation by weeding out what they view as “hate speech.”

The program, which is still in the testing phase and has yet to be rolled out in a final form, uses machine learning models to essentially “score” a comment in terms of how likely it is to be perceived as “toxic” to the average reader.

Of course, as with any other sort of machine-based learning model, it was designed by humans, and therefore contains an inherent bias derived from those programmers and code writers, a bias that clearly showed through when some of us here at Conservative Tribune gave the program a test-run of our own with a variety of comparative phrases and even a rather famous standalone quote from a widely revered historical figure.

The above quote, “We must learn to live together as brothers or perish together as fools,” was uttered by none other than Dr. Martin Luther King, Jr., and obviously called for peace and harmony among all people.

Yet, the Perspective program rated that statement as 87 percent likely to be perceived as “toxic” or hateful.