Artificial intelligence(AI) can significantly enhance the efficiency of a tribunal’s review of evidence, but its autonomy and “black box” nature may lead to risks of loss of control. In particular, the traditional view places the process control model as the core solution to regulating AI, and its feasibility is based on algorithm transparency, which is essentially a technical issue. Under current technological conditions, the worsening of the algorithmic black box, the alienation of algorithmic power, and the entrenchment of algorithmic bias are difficult to overcome, making it hard for the parties, the tribunal, and the judicial review courts to achieve process control over AI. A re-examination of the process control model reveals that its goal is to achieve result transparency through process transparency, ensuring that the parties, the tribunal, and the judicial review courts can autonomously decide whether the results of AI’s review should be valid, based on being fully informed. Therefore, as long as the parties, the tribunal, and the judicial review courts are kept informed of the AI’s review results and have the right to interpret, modify, accept, or reject them, they can control the channels through which the AI’s review results take effect, thus achieving control over the final outcomes of AI.