A voting paper shortage has caused voters to turn away in what is being described as a crisis in a developed democracy. Despite artificial intelligence (AI) predicting election outcomes every minute during the vote count and forecasting a turnout of nearly 60% for the June 3 local elections, the situation unfolded due to a failure in administrative planning by the election commission. Experts criticized the commission for not utilizing basic demand forecasting through AI.
On June 10, Aju Economy reported that the election commission conducted demand forecasting for the election while blocking information since June 2. Based on predictions made on May 27, considering the time needed to prepare voting papers, the 'basic scenario' indicated a final turnout of 58% to 61%. The 'pessimistic scenario' projected 54% to 56%, while the 'optimistic scenario' estimated 62% to 64%. Taking into account early voting and potential errors, it was recommended that at least 60% of voting papers be printed, with high-risk areas advised to prepare over 75%.
Fable5, the AI model used, analyzed factors such as the political mobilization environment and the simultaneous holding of 14 by-elections, concluding that despite a turnout of only 50.9% in the 2022 local elections, this year's turnout would reach around 60%. The actual turnout for the 9th nationwide local elections held on June 3 was 61%, closely aligning with AI predictions.
Experts agree that had the election commission conducted even basic AI simulations, the crisis could have been avoided. They noted that the data accumulated from multiple elections would have allowed for relatively accurate predictions through AI simulations. Lee Jong-ho, a professor at Seoul National University and former Minister of Science and ICT, stated, "This is a disastrous event that should not happen in a developed country. The accumulated data and trends were rich, yet the failure to conduct even basic simulations was the cause of the incident."
Professor Lee added, "It’s as if they predicted demand based on feeling, like saying it might rain because my back hurts. This was entirely predictable, and the technology was available, yet they made an amateur mistake."
Choi Byeong-ho, a professor at Korea University’s Human-Inspired AI Research Institute, emphasized that AI excels in simulation and prediction, making it essential to assess voting paper demand in advance. He noted, "Collective intelligence has its limits, which is why we should utilize AI."
Professor Choi also pointed out that there were issues not only with demand forecasting but also with real-time information sharing during the election. He stated, "If real-time data on voter turnout, spare ballots, and hourly voting status had been linked, we could have responded adequately on the day of the election."
The need for integrating vision technology with AI was also raised. This technology could have allowed for real-time tracking of voter numbers, enabling the printing of additional ballots if a shortage was anticipated.
Professor Choi remarked, "By utilizing vision technology, we could have analyzed waiting lines by time slot to determine if there was a shortage of ballots. It’s puzzling that such considerations were not made."
He further noted, "Korea is a country that exports voting technology, and by incorporating AI systems, we could reach a global standard. While this incident is unprecedented, I hope it leads to significant advancements in Korea's electoral system."
Calls for the introduction of electronic voting as a long-term solution have also emerged. Park Chun-sik, a professor in the Cybersecurity Department at Ajou University, stated, "Recurring issues such as early voting, paper shortages, and overseas voting stem from manual processes. While full electronic voting may be premature due to public trust issues, we should aim for a direction toward digitization and establish a reliable electoral system incorporating AI."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.