Is doing he explains is limiting the information the AI system is exposed to so that its breadth of information is limited to biased sources this is programming the system with its own political biases which can undermine the basic freedoms of online speech we speak with Zach Voorhees about the system how
It relates to the New Revelations on censorship at Twitter and what this means for the future of free speech and online expression Zach Voorhees great having you on Crossroads great to be back and always a pleasure well I mean you know we’re watching right now what’s happening with
Twitter Elon Musk releasing the Twitter files and people are saying we’re watching how the FBI different federal agencies were involved with censorship on the platform that’s like a big that’s become a big Scandal and a lot of people are asking well we know this about Twitter now what about Google what about
Facebook you know YouTube of course under Google you are The Whistleblower on some of this tell us what’s really happening look like right now it’s official that Twitter and Facebook both had backdoor access to the FBI they were giving requests about what they wanted to you know have removed
And I think it’s pretty obvious that this isn’t just a problem with Twitter this just isn’t a problem with Facebook this is a problem with YouTube this is a problem with Google um it’s happening you’re one of the Google Google whistleblowers yeah specifically yeah and the thing was is
That like I always had this question of who was writing these blacklists that I was watching inside of Google and then the backdoor access kind of came out with the Twitter and I started to reevaluate like maybe some of these blacklists are actually being written by you know from requests from the FBI
Requesting certain content certain keywords being banned uh on social media and I think that that’s pretty much what what the ultimate story is going to be when it’s all revealed is that at the end they gave back door access to the FBI and other three-letter agency three-letter agencies and that’s
Happening all across the world so let’s step it back a little bit you are working at Google what what year was that what was your position you know yes I worked at YouTube from 2016 to 2019 when pretty much everything went crazy with uh Donald Trump’s election and the
Censorship and the defining of fake news and then when they really started cracking down yeah and then you know I heard about the censorship regime called project dragonfly and I went to try to find Project dragonfly and it turns out that it didn’t exist but what I found
Instead was the real censorship engine which was called machine learning fairness machine learning fairness what is that machine learning fairness is Google’s AI that censors its main products uh Google search Google News YouTube and uh it pretty much classifies all data found on the platform and then
They can know which signals to amplify and which signals to us suppress well because a lot of people if I go on YouTube I assume they’re giving me stuff based on my search history and my interests you’re saying it’s not like that there’s there may be to an extent
But there’s intentional manipulation for political reasons look if you like to go you know look at videos that involve baking you’re gonna get more stuff about baking but if you go to like one of these blacklisted terms uh they’re gonna try to not give you more of that content
They’ll give you alternative content on the next up algorithm and uh that’s pretty much how it works like everything works great until you get something that is politically sensitive and then they’re going to suppress it and the next time you search for it it may be
Even harder to find that’s what a lot of journalists are finding is like oh I used to be able to find it like a month ago and now it’s gone and that happens all the time is because of this uh um you know machine learning fairness that they’re using that constantly evolves
What it is you can find on the internet and if they can shift that over to the window to the left then people like us are essentially programmed by it well and I I can see it I mean as a journalist I I do research all the time
I remember the algorithms you know the search functions on Google for example used to be very easy to find like certain articles I was looking for based on certain keywords you know give me a pretty good you know selection you know you know pretty good results based on
What I was looking for and it was really good I’d often find exactly what I was looking for I find these days that I can’t find half of it yeah even searching my own content because as soon as I remember like oh I did I did an
Article on this years ago I’ll look for it I can’t even find it using Google I have to use other I have to actually use other search engines typically to find it you’re saying that basically this is part of a intentional manipu or maybe not this specifically but generally this
Is a potentially intentional manipulation process based around political interest that they’re doing or they’re making it more difficult to find things that go against you know the narrative they don’t want based basically right and the thing is is that they’re boosting authoritative content so what the BBC says what the guardian
Says these mainstream leftist organizations are ranked within Google as having the highest authoritative value right so that’s what they say credible sources right exactly so it’s like the New York Times top of the list the New York Times writes a hit piece about a certain topic or person and
Let’s say that’s part of the local election you’re going to search for that and the first five you know links are going to be to this is what the mainstream media has to say about that so they can redefine reality you know if Wikipedia gets changed then it’s like oh
The authoritative source of Wikipedia says that this thing is you know uh not happening or a conspiracy theory and not real and now people don’t know what to think because they don’t even know that their psychological warfare and an influence operation being targeted directly at their minds well and we see
This as well because they kind of circled the wagons so to speak around like the the systems they control Wikipedia as you mentioned will only take we’ll only take references from like basically mainstream media and like journals and so on which are all typically of the generally the same like
Political leaning or like just the establishment in general if source is outside of that tiny boundary right are not regarded as credible sources then what is like you know I mean I I don’t think epic times can even go I don’t think we can even be cited on Wikipedia
Actually last I checked and so you know if people say what is your proof what is your evidence you know when you’re talking about things that maybe debunk established narratives right it becomes the cathedral in which everyone is self-referencing each other and uh if you can’t if you’re not part of that
Inner Court then you can’t get your opinion heard in your definition of reality no longer makes sense in their sort of echo chamber now I know you wrote a whole book about kind of the way these algorithms work and you had actually internal Google Documents detailing some of this
Uh we’re we’re you know we’re watching what’s happening at Twitter we’re not really seeing the inner workings like the mechanisms at play like you you actually get into the mechanisms of what was happening like how the systems work what did you see there that made you concerned what I was concerned about was
How they were curating their data to generate these AIS that were then biased with uh social justice values calm leftist Progressive whatever but they’re pretty much adhering to the current thing the current Narrative of the mainstream media so we talk about algorithms right you’re you’re you could you could technically call an algorithm
And AI as long as it has like a machine learning element you were saying this internet though the censorship is within the machine learning though yes explain that so AI is a product of the data that gets fed into it and so if you want to create an AI That’s Got social justice
Values uh what you’re going to do is you’re going to only feed it information that confirms that bias wow so by biasing the information you can bias the AI they’re they’re pretty much coupled and so that’s what they’re doing
