Posted by 
Disinformation, algorithmic bias, deepfakes, and fake accounts are just some of the ways AI can negatively impact elections. As the world gears up for pivotal elections in 2024, finding ways to combat negative AI interference in elections will be paramount.
SwissCognitive Guest Blogger: Zachary Amos – “How AI Will Impact 2024 Elections”
Generative artificial intelligence — models that can create images, videos, audio or text — have become incredibly popular because they’re widely available, easy to use and fast. Unfortunately, their greatest features are also threats. Will this technology permanently improve elections or unfairly sway the polls in one candidate’s favor?
AI’s Impact on Elections Is Global
2024 is a pivotal election year — not just for the United States but the world. Residents of over 50 countries will visit the polls this year alone, including Mexico, South Korea, the United Kingdom, India, South Africa, Taiwan and the European Union.
While most voters have come to expect — and know how to spot — attack ads, online trolling and misinformation around election time, AI has brought the world into uncharted waters. Generative models can create convincing images and videos with only one minute of audio or a few lines of text.
An AI-generated deepfake — real content that has been digitally manipulated with AI — is another massive concern. This technology replaces one person’s likeness or voice with a synthetic alternative.
According to one recent survey, about 78% of people believe bad actors will use AI to influence the U.S. presidential election outcome, with 70% thinking they’ll generate fake information and 62% assuming they’ll convince people not to vote.
AI’s Negative Impacts on Elections
There’s no downplaying AI’s negative impacts on elections.
Disinformation
Most people learn about candidates and current events from social media and internet headlines. In the United States, 82% of adults get their daily news from a digital device. This is an issue in an age where bad actors can create AI-generated disinformation almost instantly.
The Center for Countering Digital Hate, a British nonprofit, recently tested six of the leading AI voice cloning tools. Each produced fake audio snippets of high-profile politicians, with 80% of the tests generating a convincing clip.
Algorithmic Bias
AI systems can learn to make biased decisions if their training data contains skewed or inaccurate information or variables like gender, age, race or sexuality. Algorithms could act with prejudice if governments use this technology to accelerate vote counting or check voter eligibility.
Deepfakes
Divyendra Singh Jadoun is known as the “Indian Deepfaker” for his work on Bollywood clips and TV commercials. Recently, he claimed hundreds of Indian politicians sought his services ahead of the country’s elections, with 50% making unethical requests like defamation or deception. He says he denied them but doesn’t doubt others would accept their offers.
A deepfake can place a politician’s likeness over any body, face and voice to make it seem like they said or did something they never have. Politicians can — and have — used fake videos to make their opponents less likable. They even use AI on themselves to cast doubt on any real wrongdoings that might surface, giving them plausible deniability.
Fake Accounts
AI-powered social media bots spread misinformation and subconsciously influence voters by posting comments, sharing articles, and liking posts about certain politicians or upcoming elections.
Examples of AI Impacting Elections
AI’s effect on elections isn’t just hypothetical — it’s already happening. Of the 112 national elections in the United Kingdom between 2023 and 2024, 19 show signs of AI interference so far. When considering the evidence of AI-generated disinformation, that figure increases.
In Slovakia, days before the election — which was to determine who would lead the country — an audio clip of one of the leading candidates spread online. In it, he bragged about rigging the election. His opponent ended up defeating him.
In the United States, a former political consultant robocalled New Hampshire voters with an AI-generated voice meant to mimic President Biden, directing them not to vote. It reached thousands of people just ahead of the presidential primary. The man faces criminal and felony charges, along with a steep $6 million fine issued by the Federal Communications Commission.
Although the number of voters influenced in these situations remains unclear, one thing is certain — they were affected by AI interference. Going forward, cases like these aren’t going to be outliers. Instead, they may become as routine as attack ads and fake news posts.
AI’s Positive Impacts on Elections
It turns out AI might not be all bad — it still stands to positively impact the election.
Heightened Awareness
People aware of AI’s capabilities may be more likely to approach social media posts, news articles and viral clips with greater skepticism. Their newfound tendency to fact-check content can protect them from disinformation.
Election Administration
AI-powered systems could help administer elections, accelerating the time it takes to count votes, register voters or remind the general public of upcoming election dates. Considering these processes are typically so time-consuming, streamlining and automating them could be substantially beneficial.
Voter Education
Governments can offer AI tools to help voters stay informed. A machine learning model can pull up the latest news, fact-check social media posts, summarize news articles or identify AI-generated content.
AI-Generated Content Will Influence Elections
While rampant disinformation around election time isn’t new, it was obvious to those who could spot the telltale signs of Photoshop or traditional digital manipulation tactics. Now, generative models have muddied the waters. How can people tell what’s real anymore? What happens when politicians shrug off real scandals as some AI-generated hoax?
The question isn’t whether the AI’s positive impacts outweigh its negatives — it’s how to combat bad actors using this technology. Generative and machine-learning models are here to stay, so voters, governments and politicians should work together to figure out how to handle them. Swift, collaborative action may soon be the only thing ensuring fair elections.
About the Author:
Zachary Amos is the Features Editor at ReHack, where he writes about artificial intelligence, cybersecurity and other technology-related topics.
Der Beitrag How AI Could Impact The 2024 Elections erschien zuerst auf SwissCognitive | AI Ventures, Advisory & Research.
