[swatthatfly]

Clearly not context aware: typing "Wonder Woman and Superman" gives:

"Woman" may be insensitive, use Person, Friend, Pal, Folk, Individual instead "Superman" may be insensitive, use Titan instead

There is a lot of work done in the area of linguistic analysis and word disambiguation, and none of it it's trivial. So it will be a big leap moving from a toy project to something of actual use. Integrating with an existing ML api for disambiguation could vastly decrease the amount of false positives. Idilia, BabelNet/BabelFly and LingPipe come to mind.

[TazeTSchnitzel]

Someone pointed out another (possible) example of this further up, with the use of the word 'cripple': https://news.ycombinator.com/item?id=10199900

NLP is hard, I can't fault them for not trying to do it here. It'd be a lot of extra work and it's unnecessary for the most part, given it merely makes suggestions, so at worst an incorrect match is just annoying.

[renaudg]

""Woman" may be insensitive, use Person, Friend, Pal, Folk"

This has got to be a joke. What's wrong with these people ?

[TazeTSchnitzel]

What's wrong with it? Obviously it's not correct in all contexts, but if you can avoid excluding non-women, you should do so.

[aw3c2]

"Clearly" may be insensitive, use "maybe" instead.

[fche]

Yes, it discriminates against dirt-fluid trans-cleanists.