On October 20, during the 2026 Bund Summit themed "Technology, Geopolitics and the Global Economy" held in Huangpu District, Shanghai from October 19 to 22, ICBC President Liu Jun attended the conference and participated in a roundtable discussion.
Asked how to examine the current development of artificial intelligence from a broader perspective, Liu Jun said that if one abandons an impressionistic approach, AI's development looks more like Cubism, carrying distinctly postmodern characteristics, with a great deal of uncertainty and many unknown factors along its trajectory.
Liu Jun pointed out that the situation for commercial banking practitioners is extremely difficult. "It is practically like Victor Hugo's Les Misérables," he said. Commercial banks have just passed through a development stage centered on building computer systems, have only just entered the stage of digital transformation, and before completing that transformation they have already set foot on the new and unknown journey of AI.
Liu Jun stated that commercial banks can only adopt relatively mature, proven technologies and cannot rely entirely on frontier technologies that are not yet mature. The first issue to resolve is the dilemma of AI's identity positioning: whether AI is a subject or an object, whether it should become a legal subject on par with humans, or merely serve as a technical tool.
This directly raises the question of where responsibility lies. The development and application of AI involves algorithm developers, programmers, deployers, users and end users, and when combined with the algorithm black box, the determination and verification of responsibility are extremely difficult at the current stage.
Liu Jun further noted that this gives rise to several mismatches. First, a disclosure mismatch: AI algorithms and parameters sit inside a black box, making it hard to follow the existing disclosure methods used by institutional entities. Second, a data storage mismatch: data and algorithms have no national borders and flow across them, yet requirements exist for data to be stored locally. Third, a sovereignty mismatch: no country is willing to hand control over artificial intelligence to another country, and sovereign AI has drawn widespread attention. Fourth, a credibility mismatch: the banking industry has a mature credit scoring system, but it is very difficult to conduct credible assessments of various AI models, whether closed-source, open-weight or open-source.
Liu Jun emphasized in conclusion that all technological development should ultimately serve the progress of human society. He proposed that AI could be regarded as a quasi-legal subject similar to a human, that AI financial governance could be combined with the existing governance system for market participants, and that, relying on data labeling and tracking capabilities, a unified regulatory mechanism could be established to achieve full-process connectivity and solve the problems of verifying and identifying systemic risks.
Comments