[2026 Sustainability Learning Week – Evening Lecture] “Can We Trust What AI Says? From Machine Understanding and the Limits of Knowledge to Professional Judgment and Human Responsibility”

China Medical University’s (CMU) Office of Social Responsibility and Sustainability held an evening lecture as part of its Sustainability Learning Week on Thursday, November 5, 2026, from 6:00 p.m. to 8:00 p.m. at the Stanford Conference Room on the second floor of the Excellence Building at the Shui-Nan Campus, with simultaneous online participation via Google Meet. The event invited Professor Szu-Ting Chen, Director of the Institute of Philosophy at National Tsing Hua University, to deliver a keynote lecture titled “Can We Trust What AI Says? From Machine Understanding and the Limits of Knowledge to Professional Judgment and Human Responsibility.” The session was hosted by Chief Secretary Yueh-Sheng Chen, who delivered the opening remarks. The lecture attracted a large number of student participants, who gathered to explore issues concerning knowledge, ethics, and professional judgment in the development of artificial intelligence (AI).

Approaching the topic from a philosophical perspective, Professor Chen examined whether AI’s ability to generate fluent and seemingly reasonable responses necessarily means that it truly understands language and the meanings behind information. Drawing on the Turing Test and the “Chinese Room” thought experiment, he explored the distinction between producing convincing responses and genuine understanding. He also introduced the concept of embodied cognition, explaining that human professional judgment relies not only on knowledge and data but also on lived experience and an understanding of specific contexts. In the context of healthcare, Professor Chen presented a scenario in which an AI system assessed a patient's medical images as indicating a low risk, while the physician remained concerned based on the patient's symptoms and medical history. This example illustrated why AI-generated predictions should not be treated as equivalent to medical decisions. The lecture also addressed potential risks such as data bias, proxy indicators, and AI hallucinations, emphasizing that even models with a certain degree of predictive accuracy may produce unfair outcomes when datasets lack adequate representation or fail to account for the actual circumstances of different population groups.

In the concluding session, Professor Chen further emphasized the importance of AI governance and human responsibility. He explained that effective human oversight should meet three essential conditions: “understandability, intervention capability, and traceability.” These principles ensure that users not only understand the reasoning and limitations behind AI-generated decisions, but also possess the authority and ability to intervene and make corrections when necessary, while maintaining relevant records for future review and accountability. Through this lecture, students were encouraged to extend their thinking from philosophical inquiry to healthcare practice and AI governance, reconsidering the relationship between technological applications and professional responsibility. CMU hopes that students will continue to develop independent thinking, information verification, and professional judgment skills while making effective use of AI to enhance learning and workplace efficiency, becoming interdisciplinary professionals equipped with technological literacy and a strong sense of responsibility.