Date: March 8, 2024 (Friday)
Speaker: Prof Song Fei, Lingnan University
Chair: Dr Frank Hong, The University of Hong Kong
Abstract:
In the literature on trustworthy AI, there are two main positions. The first view is in order to illuminating trustworthy AI, we need a general account of trust and trustworthiness, which is less anthropocentric. Because an account of trustworthiness with strongly anthropocentric psychological features (such as goodwill) not be generalized to our current AI (as our current AI technology is narrow AI, which lacks goodwill, virtues and etc.). A general account of trust and trustworthiness, which can be applied to AI and make sense of the idea of trustworthy AI will be a less anthropocentric one. The second prevailing position contends that AI can only be considered reliable, as it lacks autonomy, agency, and the capacity for goodwill. The primary distinction between the two main positions in the existing literature lies in their stance on whether the anthropocentric aspect, such as goodwill, should be compromised in trust relationships when applied to AI, or if we should maintain interpersonal trust while acknowledging that current AI lacks trustworthiness. Each position carries its own limitations. In this talk, I develop and advocate for an alternative approach called the network account of trustworthy AI (hereafter NTAI), which aims to overcome the drawbacks inherent in both positions. According to NTAI, a trustworthy AI requires both reliable AI technologies that consistently fulfil commitments, trustworthy human agents, including AI designers, and AI companies, who consistently fulfil commitments out of goodwill; and well-established social systems regarding the AI usage.
