Liberal bias in, censorship out

A well-known phrase in computer science is Garbage In, Garbage Out (GIGO). Flawed input data causes the computer to output garbage responses. This is closely related to some well-known biases in science which can badly affect results. Confirmation Bias happens when someone only looks for and focuses on information that confirms their prejudices. Expectation Bias occurs when a tester tends to favour data that agrees with their preconceptions and disbelieve data that doesn?t.

While the most visible instances of social media censorship, the Great Purge of the Internet, is due to conscious decision-making by colluding Silicone Valley plutarchs, an even more insidious censorship occurs daily, almost without notice. This is the algorithm-driven, automated censorship that strikes down YouTube videos or slaps Twitter users with 12-hour suspensions.

Technocrats like to boast of their sophisticated AIs, but supposed ?Artificial Intelligence? is an exaggerated name for what are merely extremely sophisticated computer programs. Like all computer programs, they can only act as their programming and input dictate. Garbage in, garbage out.

AIs programmed with biased algorithms, created by hopelessly biased humans, make hopelessly biased judgments: hopelessly biased against conservatives.

A follower of a liberal/libertarian Silicone Valley blog, Slate Star Codex, described just how social networks build AIs that automatically censor not just conservative content, but innocuous content by users they deem suspect. Quote:

There has been a lot of controversy with social networks censoring conservative views. I work at a major social network headquartered in the San Francisco Bay Area and it is absolutely true.

We have an algorithm that suppresses content that it deems bad. It is supposed to suppress abusive and spam messages. It ends up not working like that. End of quote.

The reason it ends up not working is because its creators, evaluators and users are all hopelessly biased. GIGO. Quote:

To train the model we have humans evaluate content to determine if it is abusive. The evaluators are given no clear instructions…[and] are a small set of employees who work in our Bay Area office?.There has been a huge liberal push inside the company to ban political opinions they disagree with.

In my years in the company, I have not heard a single conservative voice. It’s?an echo chamber to the extreme. People talk about how it is outrageous that anyone could see the situation differently than them. End of quote.

So the model is programmed by biased, groupthinking SJWs locked in a tiny echo-chamber in one of the most elite cities in the U.S. Even when the social media companies find that their AIs are shockingly biased, they just pretend that they aren?t. Quote:

To the companies credit they spent some amount of time to determine if it is biased…the outcome of this study showed there was significant bias…however, these finding got distorted using some statistical slight [sic] of hand so that…it said the opposite.

Data scientists also spot-checked the model…with my judgement 95% of the content that was suppressed was conservative and [greater than] 90% of the suppressions were false positives. 100% innocuous content was being blocked. I have no idea how the data scientist considered the algorithm was working as intended. End of quote.

Simple: confirmation bias. Just like the people who created and then instructed the AI, the data scientists are in lock-step with the biased echo-chamber of Bay Area elites. When the data shows overwhelming bias against conservatives, of course, they think it?s working swimmingly. These are people who want any views other than their own banned, and that is exactly what their AI model is intended to do.

Unlike most of their colleagues, the person who posted this expose can clearly see the danger of Silicone Valley plutocrats censoring the public square. Quote:

I am very concerned with my company censoring conservative views…I think the effect of [Russian interference] is nothing compared to…secretly suppressing conservative views. People these days get their news and politics from so few sources and democracy suffers when one of those sources suppresses one side of a debate.

The younger generation is very open with restricting speech they disagree with. As that generation gets older and rises in power within the company, I think we will find a small group of Bay Area social justice activists controlling discourse for the entire country. End of quote.

Caption: A small portion of the screen capped original comment

The commenter in question edited their original comment, out of fear of being doxxed. As the James Damore case showed, anyone working for tech companies who doesn?t slavishly adhere to SJW orthodoxy is absolutely right to fear the destruction of their career, if they are exposed. This article has been written using both the edited comment and a screen capture of the original.