We get only the most tantilizing glipse of this research from Liming Liu; I couldn't find a more detailed presentation of these ideas anywhere online. Essentially, the argument, based on a study of 500 billion words from 5.2 million books since 1500, while the five basic moral foundations are "care, fairness, liberty, purity, and authority," a deeper analysis shows "different patterns emerged from this, and varied strongly across four target languages." As Axel Bruns comments, "existing research on this is overreliant on observations from English-language studies, which may not translate well to other languages and cultures with their own cultural norms." No kidding. I find it ironic how writers who complain of bias in AI fail to examine the cultural bias in their own understanding of ethics.
Today: Total: [Direct link] [Share]
