Abstract: The application of artificial intelligence arbitration in international commercial arbitration is on the rise, properly recognizing and appropriately addressing the the predicament of AI-assisted arbitration from the perspective of international commercial arbitration is a core issue that affects the speed of the arbitration intelligentization process. AI-assisted arbitration in international commercial arbitration faces functional dilemmas, trust dilemmas and implementation dilemmas. First, as the realization and optimization of AI functions are highly dependent on a large volume of data, the barriers to cross-border flow of data lead to restricted access to data for arbitration awards, triggering functional dilemmas; second, as the trust of arbitration parties in AI arbitration is built on the basis of transparency of algorithmic operation, the current algorithmic regulatory mechanism is unable to effectively satisfy the high demand of parties for transparency, weakening the the credibility of AI; third, as the Arbitration procedure is essentially a private procedure, when the enforcement of the arbitral award needs to rely on state coercion, the court will conduct a judicial review and make a final decision, and the lag of the existing judicial review rules to a certain extent detracts from the enforceability of the arbitral award. In view of the urgent need to promote the intelligent development of arbitration, the specific paths to overcome the predicaments should be to adopt data governance strategies, improve the algorithm regulatory mechanism and update the arbitration supervision rules.