This article is more an overview of the entire issue rather than a summary that resulted in Google summarily firing their the co-lead of their ethics team, but it raises substantial questions about why Google objected to the paper so strenuously. The paper, in some important respects, makes points that have been made previously, and especially that the cost and scale of big-data AI benefits large companies disproportionately, while at the same time being nearly impossible to audit for built-in biases. Jeff Dean, the head of Google AI, said the paper didn't meet Google standards, and specifically, "it didn't mention more recent work on how to make large language models more energy-efficient and mitigate problems of bias." But Google's internal review typically focuses only on "disclosure of sensitive material, never for the quality of the literature review." So this seems more and more like a special case, for some reason.
Today: 5 Total: 134 [Direct link] [Share]
