In news coverage of homicides in the Boston Globe, from 1976 to 1984, stories where the victim was black were less likely to use language that would encourage seeing the victim as a human being than stories about white victims.

This must reflect racism in some way or other. I'd be very interested in whether it is the result of racism of the writers, or some sort of systemic racism.

I am puzzled that the article says that this used "artificial intelligence", but it doesn't say what that is meant to refer to. I am dubious that something as complex as reasoning software or machine learning was needed to categorize the stories.