This article postulates that there will be two stages for the implementation ofArtificial Intelligence (AI). In the short term, the first stage will lead to a complementary relationship between predictive machines and humans. After the cost of prediction decreases, new players come to the arbitration arena and the flow of capital to finance AI’s use in international arbitration is widely available, we will see the second stage’s outset where predictive machines will assist in more sophisticated tasks. AI may assist counsel in crafting arguments, and arbitrators in comparing evidence submitted, and finding conflicting fact patterns in the evidence. AI may even decide some aspects of a case. This requires a new division of labour. Lawyers will have to adapt and learn to delegate to such machines while being aware of their limitations. In response, new arbitration specialties will inevitably emerge. However, flesh-and-blood arbitrators will not be eliminated. While predictive machines may be able to decide certain aspects of arbitrations quickly and at a lower cost, the amount of data, the lack of repetitive patterns, inconsistencies, and parties’ agreement that the award shall remain confidential and state the reasons upon which it is based may hinder their capabilities. The current legal framework seems to require drastic changes to make way for AI. artificial intelligence, machine learning, international arbitration, expert systems, rule systems, the future of arbitration, division of work, intelligence