I really like this paper (21 page PDF) for two reasons. First, it firmly places the ethics of using AI in contrast with the existing human systems, and in particular, it notes that "human peer review systems are also fraught with related problems, such as biases, abuses, and a lack of transparency, which already diminish credibility." Second, it approaches the question of AI ethics from a perspective of 'legitimacy' of use in scientific research (in contrast to, say, to a social justice perspective). "Science encompasses unique structures, dynamics, and factors that drive perceptions about legitimacy. Different fields and disciplines have their own value systems, priorities, and cultures, including notions of what is legitimate or not." While this article discusses peer review, I would say that the considerations apply to assessment in general across a wide variety of disciplines.
Today: 0 Total: 19 [Direct link] [Share]
