Political parties and campaign consultants are using artificial intelligence to create candidate biographies and parse the data at lightning speed. Some have gone a step further. This technology is already used to generate video ads.
An ad posted on the Republican National Committee’s YouTube channel last week used AI technology to depict thousands of people walking to the Golden Gate Bridge for drug-related crimes.
But are we really on the verge of a revolutionary AI development that will fundamentally change political horse racing? Answers from most experts: Maybe.
The excitement around AI and its rapid evolution is real. So are the risks. But understanding how technology is implemented in the real world is still in many ways a guessing game.
The Criterion spoke to political organizers, data scientists and other experts about how AI is already being used in politics and what might happen next.
trust but confirm
Political consultants say the most obvious advances in AI are at a basic level, such as creating bios and basic speech outlines with programs like ChatGPT. But there have also been amazing advances in creating “deep fakes.” These fake videos, photos and audio recordings can easily avoid detection by the naked eye or bare ears.
The deep fakes that went viral on social media during the 2020 election “have never been as scary or real as they are now,” said Elyse Samuels, video reporter for The Washington Post’s Visual Forensics team.
In a presentation at the University of Chicago’s Political Journalism Conference last week, Samuels contradicted by fact-checking Republican AI-generated videos of people fleeing San Francisco and cross-referencing them with other verified images of the Golden Gate. showed that it is possible to identify bridge.
Other deepfake images that have gone viral recently include manipulated images of Donald Trump being tackled in the street and fleeing police when he turns himself in on charges of falsifying business records.
“There are several different anomalies that you can look for in images that may be fake,” Samuels said, noting that some officers in Trump’s images have missing fingers or extra fingers, and their gaze is off-center and multiple images.
In fact, a thorough investigation is required to confirm that a deep fake is fake. The Post set up a team to verify deep fakes after Russia declared war on Ukraine. Samuels said his group of about eight reporters watched hundreds of videos purporting to show acts of war and missile attacks.
However, deep fakes are not limited to videos and photos. If AI can generate a viral new song by a fake Drake and The Weeknd, the presence of the technology will allow real audio, such as recordings of Trump making vulgar claims about molesting women, to question the veracity of his leaks. Doubts may arise.
“I was talking to Hany Farid, a professor at UC Berkeley who has done a lot of research on this, and he said that the group of people who are fighting misinformation and trying to detect this are actually: [number of] People are making it,” Samuels said.
Of course, ethical concerns extend beyond mere deep fakes. Campaign consultants worry that in the age of AI, the checks and balances of insensitive and racist ads may be lost. San Francisco supervisor Matt Dorsey, who worked as a spokesman and speechwriter before taking the elected office, said in an interview last year that acting as “the conscience of the campaign” was the communications director. told the Standard that his work as a
Using AI to target voters and craft messages may be expedient, but you can also take an approach that puts ethics aside and wins at all costs, says the Community Tech Alliance. CEO Kat Atwater said. With a progressively oriented organization.
“One of my biggest fears, especially when thinking about applying AI to data science and machine learning, is the bias it may have built-in. There are no checks or checks for that. the ethical ability to put frameworks and lenses and evaluations on it.”
About the data
Campaign experts agree that algorithms can be used to power advanced data analysis at the country level, but the work is only as good as the data it relies on.
When it comes to political campaigns, data tends to be more informative at the state and national level as there are more voters to analyze and more money to spend. According to Open Secrets, spending on state and federal elections in 2020 will total $14 billion.
Reince Priebus, former chairman of the Republican National Committee and former chief of staff to President Donald Trump, is particularly bullish about how much data the GOP has on voters in battleground states. At a journalism conference held at the University of Chicago Institute of Political Science last week, Priebus said that in the 2022 Wisconsin senator and gubernatorial elections, he’s targeting just his 50,000 voters, and he’s 120 million. said the dollar was spent.
“50,000 [people]and we really know the beer you drink, the car you drive, the number of kids you have, your income, whether your mortgage is upside down,” Priebus said. I’m finely chopping all of the
Nationwide voter files can also drill into spending habits through data obtained from credit bureaus, Priebus said. But optimizing this information to a truly useful level with the help of AI may be a long way off.
“For years we’ve had this myth about microtargeting. Specifically, that campaign chops up voters in 50 different ways and creates 50 different messages for each of those segments.” said Patrick Ruffini, founding partner of Echelon Insights, a poll and analytics firm. “And the reality is much more complicated than that.”
But that’s not to say AI isn’t in use at our local races here in the Bay Area.
Byron Philhour, a data scientist at San Francisco-based RTBiQ Political, said his company works primarily on moderate Democrat campaigns and collects data from campaigns as small as county supervisor campaigns. I’ve been using machine learning for sets.
“The problem is not using that data to identify target groups,” says Philhour. “It really is a question of the campaign’s ability to have the time, space, energy and money to generate very different messages to these targets.”
He adds: It shouldn’t be the central focus and it isn’t. People still need to have strong arguments and build relationships. ”
Short-term AI
Political insiders say one of the main benefits of AI is that instead of sitting at a desk and doing data entry, they will be able to have a personal connection with voters. is.
Emily Norman, CEO of Democratic Data Exchange, which compiles information in the Democratic National Voter File and shares it with allied organizations, said she has two employees, one of whom has a PhD. said there is. She now has a master’s degree in one and spends a lot of time reading and sorting through hundreds of thousands of responses to survey questions.
“Do you want the AI to speed it up so they can use their time to do better things?” asked Norman. “Yeah, I think they could use their time to do something better.”
Another example of AI making the leap from being a useful tool to being a game changer is political organization. Atwater proposed that AI could be used to “customize” the experience of potential voters, from website visitors and grassroots donors to campaign her workers.
“These are just the things the campaign is dreaming about organizing, from stakeholders to donors to activists to volunteers,” Atwater said.
But for now, Ruffini says AI’s biggest benefits are limited to “automating mundane tasks.”
“It’s not sexy and it’s not going to be covered.” [by the media]but that’s what I see as if it actually has an effect [2024]said Ruffini. “It’s at the back-office level, and unless someone has a very smart marketing team trying to make a sexy story out of it, there’s not going to be much written about it.
