Photo Credit: pexels
The deep fakes created by AI keep getting better and better. The first generation of deep fakes were easy to spot, and did not represent a real risk to the democratic election process. But the current generation of deep fakes, powered by generative artificial intelligence (AI), are eerily real. If you’re not paying attention, and just scrolling through endless posts on social media, it’s not out of the question that you might fall for one of these.
AI avatars warning about AI
To underscore this point, Arizona Secretary of State Adrian Fontes recently appeared on NBC’s “Meet the Press” to demonstrate an AI version of himself delivering a warning about AI to voters. And, indeed, if you weren’t paying attention, you’d be convinced that it was actually him, and not an AI avatar.
Today’s deep fakes have gone way beyond what was possible several years ago, when a deep fake version of Barack Obama saying bizarre things went famously viral. Back then, AI seemed like a cute parlor trick, sort of like listening to a ventriloquist and a misbehaving sidekick. But the new AI avatars are nothing to laugh at. And that’s what has me convinced that this might be the first presidential election decided by AI.
Is that the real Joe Biden or Donald Trump?
Here’s the nightmare scenario: a deep fake version of Joe Biden or Donald Trump goes viral right around Election Day. The candidate appears to say or do something outlandish, and there’s simply not enough time between the launch of the deep fake and the election to fact-check things.
Don’t think it could happen? Well, how about the alleged deep fake of Joe Biden that went viral back in July 2022, when he commented on the events of January 6? To this day, political commentators are still trying to figure out if it was really him or not. The voice sounded “off,” and the person claiming to be Joe Biden did not blink once in the 17-second clip. Just try doing that at home – you’ll immediately realize how improbable that is.
The end of fact-checking as we know it
On top of all that, Google is now experimenting with AI for its search results, and that is resulting in all kinds of “AI hallucinations.” These occur when AI says things that are obviously false, but does so in a way that attempts to be convincing. The tech site Gizmodo, for example, recently detailed a number of disturbing Google AI hallucinations, including a recipe for “gasoline-infused spaghetti.” If given the right prompt, Google AI will also tell you plenty of misinformation about political candidates.
So we can’t even trust the fact-checkers any more. Maybe the fact check is just another AI hallucination? After all, can you really trust any fact-checker who thinks that “gasoline-infused spaghetti” is a real recipe?
How to distinguish between fact and fiction on social media?
So good luck trying to figure out what you’re hearing, seeing, or reading on social media is actually true. Using generative AI, it’s simple to generate any type of content you want, ranging from text documents to short video clips. You don’t need a room full of Ph.D. scientists to pull this off, and that means both campaign teams might be tempted to dabble in AI.
We’ve already seen plenty of examples of generative AI being used to create content about Biden and Trump. This went well beyond just photos and brief video clips. For example, Twitch.TV recently hosted a parody “AI debate” between AI avatars for Trump and Biden. (Spoiler alert: it was more entertaining than the real debate.)
So, 2024 is already shaping up to be the presidential campaign determined by AI. Who knows? Maybe the AI is already in control, and humans don’t even realize it yet.
That explanation might sound preposterous, but it also might be the only way to explain some of the bizarre glitches in the matrix – like Joe Biden frozen in place at a Juneteenth event – that we’ve already seen this year.
