Tech Overlords Frustrated That AI Won’t Banish Trump’s Tweets
The tech overlords from Google, Facebook and Twitter keep insisting that “the algorithm did it” every time that they ban, suppress or censor a conservative’s speech online. This is the tech overlord equivalent of “the dog ate my homework,” of course. They’re lying to us, as whistleblower after whistleblower has proven.
Humans at these companies are manually pushing a button to suppress conservative speech. But that doesn’t mean that the tech companies are not trying to create a computerized way to censor conservatives. They are doing so, but it keeps backfiring on them.
All of the tech companies are developing Artificial Intelligence (AI) programs right now to try to take the dirty work of censoring conservatives off of their human hands. It will be so much easier to suppress the tweets of people like Donald Trump, Ann Coulter or Kurt Schlichter if they can just get a computer program to do it for them!
Wired, which used to be a decent tech magazine but is now just another “woke” social justice rag, decided to do a report on how well Google’s AI program known as “Perspective” is doing. The writers at Wired are all-in when it comes to social justice.
They know that people like Donald Trump, Faith Goldy, Stephan Molyneux and others engage in toxic hate speech. Their speech should be banned. Meanwhile, all Democrat social justice groups are virtuous, and their speech should never be suppressed or banned in any way.
So, the writers at Wired ran an experiment to have the Perspective AI categorize the “toxicity” of the tweets of prominent conservatives and compared those to the toxicity of every virtuous Democrat transgender contestant who has ever appeared on the TV show Ru Paul’s Drag Race.
AI was going to be the salvation of the left. It was going to finally show us the light of whose hate speech should be banned, and whose tweets deserve to be seen by all!
The results… are hilarious.
In fact, the results are a disaster not only for the writers at Wired magazine, but also for the tech overlords who were hoping to develop a digital form of salvation from conservative speech online.
If the toxicity level of a person’s various tweets exceeds a threshold of 50%, Perspective would presumably then recommend suppressing that speech (if it had the power to do so – which it doesn’t since Google doesn’t own Twitter).
Every drag queen that ever participated on Ru Paul’s show has a toxicity level above 90% in their tweets, according to Perspective. Meanwhile, “racists” like Trump, Molyneux and Goldy never cracked the 30th percentile on their Twitter feeds when it comes to toxicity.
The “hate speech” of these hated figures was nowhere near as hateful as the speech of the virtuous leftwing drag queens that Wired compared them to.
It turns out that the AI determined that calling people “fags” and “sissies” all day, every day on Twitter, as the drag queens do, is pretty toxic. Meanwhile, Stephan Molyneux citing current scientific studies about racial IQ differences is pretty tame.
Stupid AI! Why doesn’t know which Bad People to ban and which Good People to promote? The AI, of course, is analyzing the actual text of the online speech of everyone and using neutral metrics to impart a score to it.
The computer program isn’t looking at the skin color or sexual proclivities of anyone. It’s just using the objective data of the actual written words to rate them. Wired is, of course, devastated by this.
In one of the most hilarious examples from their study of Perspective, they turned to the Twitter feed of the boogeyman of the left in every presidential election: The totally irrelevant former Klan leader David Duke. Surely the AI would condemn the tweets of such an evil person, right?
In one example, David Duke suggested that maybe it’s not a great idea for the nation of Italy to replace its own population with millions of “Somalians” (sp) so they could have new cuisine. Perspective rated that tweet’s toxicity at 7% — which is only half as toxic as Michelle Obama’s total average.
Let that sink in for a second. Michelle Obama’s online speech is twice as toxic as David Duke’s, when they are compared based on their words alone.
It turns out that when AI learns on its own, its decisions about online speech are remarkably similar to those of normal people. Democrats and leftwing interest groups engage in nasty, venomous speech against their enemies, while everyone else is pretty mild-mannered by comparison. That’s a big problem for the tech overlords, and it must be frustrating them to no end.