@gvwilson
If this is a real pattern the cause would be interesting to examine! It's unlikely copilot was trained on a lot of samples like this. Which would mean it's transferring sexism from other domains to the programming domain.
Perhaps it's not so much sexism it's transferring but having ingested a lot of text talking about a wage gap in percentages. So news highlighting sexism causing sexist code?
@faassen @gvwilson I mean it's taking literal examples from github.com: https://github.com/CKalitin/ckalitin.github.io/blob/9d5de0a13860032f2c016cdf07fa297be7e29bf9/.post-ideas.md?plain=1#L63-L65
(That's the only one I found with that literal function name, but I'm sure there are others that are close in semantic space)
((this is a place where AI could be actually useful, if I could ask it what examples were semantically close instead of literally close))
I don't expect a lot of actual public codebases with functions like this as more usually it's a pattern in the data rather than hard coded in some function.
These weird examples and others could definitely have an impact though. Though I wouldn't discount influence of news articles.
Greg Wilson
Martijn Faassen has moved
Bill Mill