Choi Min-hee Targets Oh Se-hoon's Acquittal with AI Judge Bill
A bill that would put artificial intelligence (AI) judges at the forefront of political trials is expected to appear in the National Assembly.
Choi Min-hee, a Supreme Council member of the Democratic Party of Korea, criticized the judiciary after Seoul Mayor Oh Se-hoon's first-instance conviction — an effective loss of his seat — for illegally brokering poll survey funds was overturned to an acquittal on appeal. Pointing to the reality that lower and higher courts reach opposite conclusions in the same political case, she formally announced plans to prepare legislation entrusting judgment to AI rather than humans.
Background of the Bill
The immediate trigger for this proposal was Mayor Oh's trial for violating the Political Funds Act. The first-instance court sentenced him to a term resulting in disqualification from office, but the appeals court reversed the outcome with an acquittal. The same case and evidence produced divergent verdicts. Supreme Council member Choi framed this as a matter of judicial credibility, arguing that the very structure in which outcomes hinge on an individual judge's discretion must change.
Some in political circles have long argued that the divergent rulings of lower and higher courts, common in political trials, confuse voters and supporters. There are also concerns that if the perception spreads that verdicts vary with political leanings, trust in the judiciary as a whole will collapse. The AI judge proposal emerged as one alternative against this backdrop — an idea of curbing human judges' subjective bias with data and algorithms.
Key Provisions and Scope
The specific text of the bill has not yet been made public. However, according to the council member's stated intent, its scope is likely to be limited to politically charged cases rather than all litigation. One can imagine AI judging — or at least assisting in judging — cases such as election law violations, political funds violations, and impeachment-related trials, where an individual judge's value judgments have room to come into play.
The operating method could also take several forms: a fully automated approach in which AI decides guilt on its own, and an auxiliary approach in which AI organizes similar precedents and evidence for human judges. Legal experts say only the latter, assistive approach can reduce conflicts with the current legal system. The former would clash head-on with the Criminal Procedure Act governing criminal trials and the constitutional principle of the right to a fair trial, which guarantees the right to be tried by a person.
The Debate
Supporters emphasize predictability. Their logic is that if AI trained on past case data intervenes to ensure similar outcomes in similar cases, it can reduce the problem of sentences and verdicts varying sharply depending on which judge is assigned. This is closely tied to the argument that in political trials, the public's inability to anticipate verdicts undermines trust in the judiciary.
Opposition is hardly trivial. Trials are not merely the application of legal principles but a comprehensive weighing of evidence credibility, the defendant's demeanor, and social context — tasks AI can hardly replace. A more fundamental problem lies in the training data. Since AI is built by learning from past rulings, it may inherit the biases embedded in those rulings. There are also concerns that applying AI to political cases could spark a new political battle over which precedents it is trained on. Many also point to unconstitutionality, given that the constitutional right to a fair trial presupposes trial by a human being.
Legislative Process and Outlook
Significant turmoil is expected before the bill is actually submitted to the National Assembly. Reform of the judicial system, which touches on the right to a fair trial, must go through the Legislation and Judiciary Committee, and given the magnitude of the constitutional issues, review by bodies such as the Constitutional Court's litigation committee or consultation with the judiciary may be required. The People Power Party is highly likely to resist, viewing it as interference in the judiciary. The Supreme Court and bar associations are also expected to take cautious positions, as the bill touches on the essence of trials.
For now, the more persuasive reading is that this discussion is a political signal raising the issue of judicial credibility rather than concrete legislation. Still, as long as political trials continue to produce conflicting verdicts, calls for mechanisms guaranteeing consistency in judgment will not disappear. Nor can a compromise be ruled out — using AI as a decision-support tool rather than placing it in the judge's seat — emerging as an alternative during the legislative process.
