the infinite propaganda machine

or, AI will destroy the internet, but not before it destroys democracy

Nearly three years ago, I wrote about how the primary use for AI will be propaganda generated at such an astonishing rate that there will be no guardrails against it, that it very well could destabilize democracy.

Now, I am an infamous AI hater and I will continue to hate it until it or I die, but my prediction timing was off. I thought the 2024 election would be inundated with AI propaganda.

That didn’t happen.

But recently we have all likely seen lots of videos going viral of Venezuelans crying for joy in the streets, celebrating that the US illegally snatched the sitting president of the country. One might reasonably conclude that we are now at war with Venezuela, but that’s a whole different discussion.

All these Venezuelans weeping! Such joy! So many tears!

Someone with more curiosity might try to dig a bit deeper into this and figure out what’s going on in Venezuela or, rather, what was going on there last week or last year or over the last decade or since this new millennium began.

But, of course, you don’t actually care and aren’t even interested. Easier to just scroll on by. But when you hear, later, people talk about Venezuela, you might say something like, “It looks like everyone in Venezuela is really happy with what happened.”

Which is a bit of the point.

No one reads past the headline or whatever is clipped to spread on social media. Even people with enormous followings don’t bother to actually read or listen to the things they share.

And since no one reads anything anyway, text based argumentation over facts becomes largely a waste of time. Especially when a picture is worth a thousand words and a video is likely worth at least ten thousand words.

Which brings me to the fact these videos are all AI. All fake! Now, there are certainly Venezuelans who are happy about Maduro being taken out of power. Though that’s also distinct from them being happy that the US unilaterally—and without any kind of democratic discussion or oversight within the US—came in and snatched up the dictator Nicolas Maduro.

Even if you show people that these videos they’ve seen or shared were AI, it’s convincing enough that it will stick in their heads, possibly forever.

The truth of AI is that it’s getting increasingly convincing, especially with video. I don’t think you’ll ever be able to make anything artistic with AI, but you will be able to create mountains of this shit.

Here’s another example, though non-political in nature.

If you click, you’ll see how someone can use AI to wear a near perfect digital mask of someone else. In the video, a man is wearing the face of attractive young women on webcam. This may not seem like a big deal to you, but wait until this same technology begins to be exploited for political goals. You no longer need to create a deepfake through AI prompting but can now just wear an AOC or Ben Shapiro mask and say whatever you want.

It won’t be real but it will be convincing enough to fool a lot of people.

As another example, here’s a deepfake of Neil deGrasse Tyson explaining how he’s become convinced that the Earth is flat.

Again, this may not matter at all. It’s all for the laughs and so on, but we are very clearly at a point where most people, if not everyone, can no longer tell the difference between AI and reality. That’s not a critique of the unwashed, sweaty, uneducated masses.

It’s just something true about humans. We have now created a technology that fools the eye and ear so effectively that we cannot distinguish between AI generated photos and videos and real photos and videos of real life people.

That’s a problem.

It may not yet be one for you or for your country, but it is coming. And it will likely be very very bad.

So what do we do about this? What will we do when the president and even members of his cabinet are fooled by AI and make policy based on literally fake information.

MIT has already studied how AI can change people’s minds, and this is just chatbots that people know are chatbots! What might the effect be when they see video of a speech given by Trump or Gavin Newsom or some other major politician that they don’t realize is AI?

Nothing good.

And so what do we do about this? How do we safeguard ourselves from this kind of manipulation? How do we safeguard society from it?

Well, the simplest option is to go set a bomb in every data center in the world.

The trickier solution is to get people to actually care about the information they ingest at such a dizzying rate. To get people to look past the headline and actually read the article. More than that, to get people to look past the single article they read and investigate the sources of the information.

But this is really not scalable, especially since you can generate these videos and images at the click of a button. You could hire a handful of teenagers in Vietnam to generate thousands of photos and videos this week for an astonishingly low price tag, if you were so motivated.

We’re already past the tipping point where the internet is now majority AI content.

That should scare people. Think of all the stupid listicles Buzzfeed put out over a decade and consider how all of that can be spat out in an afternoon with chatgpt. Consider how many billions of people over the last twenty years have produced content for the internet.

In just a few short years, that is being eclipsed by AI content.

AI is making the internet increasingly unusable. My dearest hope is that this will force people back offline. That people will come to realize that scrolling through AI sewage isn’t worth their time so they may as well just put their phone away and get back out in the world.

And I do think that will happen. A few generations of adults have now been trained to look online for everything, but those habits will break when everything you come across online is fake or possibly a lie.

While I may find this a hopeful outcome, the short term will be disastrous. Because I do think this level and rate of propaganda has the potential to shatter what remains of the world’s democracies, to the degree that they are functional democracies.

When citizens cannot even agree on basic reality, though, it makes any kind of consensus impossible.