The contenders for Argentina’s presidential election have been whittled down to Sergio Massa and Javier Milei with both candidates making extensive use of AI in their campaigns.
Propaganda and political posters are common features of elections but with the advances in generative AI, campaigns are leaning heavily into the technology.
The candidates themselves are an interesting side story to our focus on the AI aspect, but for context here’s a brief catchup.
In this corner, Sergio Massa, Peronist Minister in charge of Argentina’s beleaguered economy. In the opposite corner, Javier Milei, hard right, tantric sex instructor, former rock singer, and fan of Donald Trump.
In an effort to appeal to the socialist ideals of his supporters, Massa’s campaign put out a series of AI-generated posters, some of which would not look out of place in Soviet-era Russia.
View this post on Instagram
Leaning into this, Milei’s campaign tweeted an image of Massa portrayed as AI’s aggregation of Mao meets Stalin.
So far it seems a little tame, but then Massa’s team upped its AI electioneering game by exercising some cinematic creative license.
In an attempt to portray Massa as a soldier fighting on the side of right, his campaign used AI to insert him into a battle scene from the WW1 action movie 1917.
View this post on Instagram
And then, just to make sure you realize who the bad guy is, Massa’s campaign used AI to portray Milei as Alex, the ultraviolent, drug-laced milk-swilling character from A Clockwork Orange.
View this post on Instagram
None of these pieces of propaganda would be considered deepfakes created with the intent to mislead voters, but they are great examples of how AI is supercharging the creative outputs of campaign teams.
The line between truth and fiction got a little more blurred with some of the other content Massa’s team put out.
Milei is a libertarian with some interesting ideas including his stated support for the legalization of trading in human organs.
Massa’s team thought it would be a good idea to create a deepfake video purporting to show Milei explaining his views on the organ trade.
In the video, which is no longer on Instagram, ‘Milei’ says, “Imagine having kids and thinking that each is a long-term investment. Not in the traditional sense, but thinking of the economic potential of their organs.”
Massa’s campaign claimed the video was made in jest and obviously not intended to be seen as anything other than satire. But with gen AI getting as good as it has, very little is obvious anymore.
The boy who cried ‘fake!’
When politicians employ AI to the extent they have in these Argentinian elections, the distinction between truth and lie all but disappears. And often this confusion is used to dismiss uncomfortable truths as AI-generated lies.
Some of Massa’s critics accused him of being under the influence of drugs after a video did the rounds showing him looking exhausted after a campaign event.
His supporters, presumably drunk on their own AI-fueled campaign, immediately claimed the video was an AI fake. Massa’s drug habits are unclear, but the video turned out to be real.
AI is so good at faking things that aren’t true, that it’s become all too easy to call ‘Fake!’ when the truth is a little awkward.
In their 2018 paper titled “Deep Fakes: A Looming Challenge for Privacy, Democracy, and National Security”, lawyers Chesney and Citron coined this paradox “the liar’s dividend’.
The abstract of the paper said that AI’s “ability to distort reality has taken an exponential leap forward with “deep fake” technology.” When you consider how gen AI has advanced since this paper was written, 2018 feels like it was decades ago.
One of the beneficiaries of the liar’s dividend was Patricia Bullrich, a candidate who dropped out earlier in the election.
Her pick for economy minister, Carlos Melconian, was accused of sexual harassment and offering jobs for sex. When incriminating audio of Melconian making these comments and speaking about “taming” women leaked, Bullrich immediately dismissed the recordings as AI fakes.
People can “make voices using Artificial Intelligence, cut videos, insert audios that nobody knows where they’re from,” she said.
It turns out the clip was made by editing audio that was recorded somewhere between 2015 and 2017. Despite the editing, the individual pieces of audio were eventually acknowledged as authentic by Bullricht’s colleagues.
Entertaining and scary
It’s easy to be entertained by the drama and creativity that AI has brought to these elections. But beyond the entertainment value lies the prospect of AI destroying the credibility of candidates, or endowing it where there was little to start with.
When you look at how quickly AI tech is advancing we begin to get some idea of what the US presidential election campaigns will look like next year.
Meta, Google, and YouTube all require different degrees of disclosure now when political ads use AI-generated content. How X and Instagram deal with ‘creative’ AI political ads remains to be seen.
Even so, expect AI to be a key enabler of the electioneering videos, ads, and robocalls the American public will be subject to.
Argentina’s AI election is likely to be just a foregleam of things to come. Gen AI in 2024 will, to borrow Sam Altman’s words, make Argentinian AI electioneering propaganda look “quaint”.
