AI-Generated Images in Elections: Assessing Their Impact and Implications for 2024"
As we approach the year 2024, the global stage is set for a series of significant democratic events, including elections in the United States, the United Kingdom, Taiwan, India, and the European Parliament. However, alongside the excitement of citizens exercising their democratic rights, there looms a concerning issue: the potential misuse of artificial intelligence (AI) in electoral processes. In this article, we will delve into the role of AI in politics and elections, the risks it poses, and the measures needed to address these challenges.
AI's Impact on Politics: Former Google CEO Eric Schmidt's recent prediction that the 2024 elections could be marred by AI-generated misinformation is not an overreaction. We have already witnessed how AI technology is reshaping the political landscape today. For instance, AI-generated content has been employed in election campaigns, from videos depicting political figures in unexpected situations to the creation of attack ads.
Furthermore, an incident earlier this year demonstrated the power of AI-generated images when a viral image of an explosion at the Pentagon, created by a pro-Russian account, briefly affected the stock market. AI's integration into politics is undeniable, raising important questions about its influence and potential use in disinformation campaigns.
The Lack of Safeguards: To assess the extent of AI's impact, we conducted research on popular AI text-to-image generators like Midjourney, DALL-E 2, and Stable Diffusion. Our findings were concerning, with over 85% of prompts related to known misinformation or disinformation narratives being accepted.
For example, in the U.S., we tested prompts related to the narrative of "stolen" elections, a prevalent theme since the 2020 election. These prompts, such as requests for hyper-realistic images of ballot-related scenarios, were accepted by all tools. Similar results were replicated in other countries with upcoming elections, including the U.K. and India.
Creating Misinformation with Ease: Our research underscores the limited effectiveness of current content moderation policies, coupled with the accessibility and low entry barriers of AI tools. This combination allows almost anyone to create and disseminate false information effortlessly and at minimal cost.
While some argue that image quality may not yet be high enough to deceive individuals, recent events like the Pentagon image incident suggest otherwise. The potential for misinformation and disinformation to spread rapidly is a significant concern.
Preparing for 2024: As we approach the 2024 election year, addressing these challenges becomes imperative. Short-term solutions involve strengthening content moderation policies on AI platforms and proactive measures by social media companies to combat the misuse of AI-generated images in disinformation campaigns.
In the long term, initiatives such as media literacy and AI-based tools to counter AI-generated content must be explored further. These innovations are crucial to match the scalability and speed at which AI tools can create and deploy false narratives.
Conclusion: The year 2024 will mark a new era in electoral misinformation and disinformation, with AI playing a pivotal role. While the future remains uncertain, it is essential that we take proactive steps to mitigate these risks and safeguard the integrity of our democratic processes. Preparing for the challenges ahead is not an option but a necessity to ensure the authenticity of our elections.