哲學學者帶領思考ChatGPT語言模型技術的知性德性

The Practical Philosophy Programme of the Division of Social Sciences organised a guest talk on the topic “ChatGPT and epistemic virtues” in October. Dr. Pak Hang Wong, Assistant Professor from the Department of Religious & Philosophy, HKBU was invited to share his insights on the potential risks of adopting ChatGPT and how users can address them.

Dr. Wong is a renowned philosopher and ethicist of technology, focusing on the social, ethical, and political aspects of AI and other emerging technologies. He first explained how Large Language Model like ChatGPT generates relevant responses promptly. Very roughly, it seeks to produce a continuation of word sequence through analysing the huge amount of training data harnessed from the Internet and users’ feedback etc. It then weights the probabilities of which word shall be relevant in the sequence, which is not necessarily based on truth and may lead to false statement. As a result, ChatGPT can “hallucinate”, bringing different risks such as information hazards and misinformation harm because it lacks genuine understanding. It gives rise to knowledge-related and ethical issues, when it is taken as a reliable source of knowledge or information. For example, ChatGPT sometimes provide references that do not exist. Dr. Wong instanced that in 2023, a New York lawyer cited fake cases provided by ChatGPT in court but was discerned by the human judge.

To use ChatGPT responsibly and ethically, Dr. Wong argued for a few solutions: to develop epistemic virtues, overcome cognitive biases, and focus more on skills and guidelines when utilising these tools in daily life. Next, he urged the need for students to learn the concepts and application of ChatGPT and other AI tools in order to embrace intellectual challenges but still being able to minimise the impact of the potential risks and problems on them. Some contemporary philosophers have been developing theories of virtue epistemology that aim at answering the questions, “What is a good thinker? How to develop intellectual virtues?” Reflecting on these philosophical questions has become more pressing in the age of generative AI.