> The “inclusiveness checker” highlights phrases or words the firm believes could offend someone based on their gender, age, ethnicity or sexual orientation. I

> The “inclusiveness checker” highlights phrases or words the firm believes could offend someone based on their gender, age, ethnicity or sexual orientation. I don’t think words can intrinsically have the property of “being offensive”, it can only be that particular words offend a particular person. Assume Microsoft had self-reports from every human about which words offend them. How would they turn this data into The Bad Word List? Do they just count the number of people who find each word offensive then put the top N most reported words on the list? Would I use the algorithm if it told me which words offend the particular recipient of my email or Slack message? Usually when someone reproaches me for saying “something offensive”, they are defending some hypothetical other who they say would be offended rather than reporting that they themselves are offended. So maybe I want the algorithm to tell me which words my email recipient thinks offend other people. Or I’ll just say whatever the fuck I want and deal with the consequences.