[Insight & Opinion] Artificial Intelligence and Elections: Questions for 2024

It is already autumn. Recently, faces of various figures expected to appear in the early next year’s presidential primary races have frequently appeared in American news. In the Republican Party, former President Donald Trump enjoys overwhelming support, but as the first former president in history to be indicted, the future is difficult to predict. On the Democratic side, President Joe Biden, who has started his re-election race, is elderly and has very low approval ratings, leading to predictions that he might drop out midway. While many do not want to think about a Biden-Trump showdown, the possibility is not small.

As much as the candidates themselves, the use of artificial intelligence (AI) in elections is also a major concern. When radio and TV first emerged, the ability to directly hear candidates’ voices and see their faces greatly accelerated the speed and spread of news. In the 1990s, the internet, blogs, and social networking services (SNS) further accelerated this speed and spread, intensifying public interest and conflicts. The power of SNS was immense, and there were even suspicions that Russia and China used ‘fake news’ to sway public opinion in their favor.

Then came artificial intelligence. It can be seen as the latest wave of the digital revolution, but its nature is completely different from the waves caused by the internet and SNS. Although SNS contained many ‘fake’ or unreliable contents, at least they were written by ‘people,’ so there was a limit to production speed. Photos could be distorted or manipulated, but they often appeared crude or sloppy, making it relatively easy to identify them as ‘fake.’

Artificial intelligence requires no human hand at all. It is a function of a completely different dimension with limitless scope. There are two major problems. One is, again, ‘fake.’ AI can create not only false facts but also nonexistent people. The author of a text or even the person appearing in a photo may not actually exist. Voices and bodies are no exception. AI can fabricate the voice and video of an existing person or create a nonexistent person as if they were alive. Since AI can create nonexistent worlds, we will no longer be able to distinguish what is real and what is fake.

The other problem is ‘bias.’ AI system experts can utilize verified real facts and information according to their purposes. Today, many media outlets release news and write articles based on their own positions. Readers can trace these paths and hold journalists and media accountable if there are errors. AI is different. It can inject desired viewpoints toward the public and block unwanted information. As time passes, AI will increasingly dominate information production, potentially eliminating the debates necessary for forming social consensus based on diverse information. Instead, people will accept only the very trustworthy-looking information neatly organized by AI, ultimately falling under the control of those in power who manipulate AI systems.

In 2024, elections will be held in the United States, India, Mexico, Taiwan, and other countries. South Korea also has a general election scheduled. Although AI is still in its early stages, it will undoubtedly be used in elections. What impact will it have on election results? To what extent should we accept its influence? Furthermore, how can we detect the ‘fake’ and ‘bias’ created by AI, and how can we prevent its misuse? Isn’t it time for a global social discussion on these issues? These are the questions I want to pose as we approach 2024.

Robert Fauzer, Former Professor at Seoul National University

This content was produced with the assistance of AI translation services.