There's a meme going around on Tumblr about search engines' automatic AI responses being racist & transphobic: apparently, "a woman touched my face" gets celebration, while "a trans woman touched my face" or "a black woman touched my face" gets condemnation (of the woman) and occasionally advice on how to call the police.
When I read about this I immediately guessed the difference might be down to the women having any described features: the bot not-unreasonably reads the user as hostile towards them when flagging personal characteristics, and reacts appropriately. If this were true, "a white cis woman touched my face" would also get condemnation.
. . . but when I tried to replicate the basic phenomenon in advance of testing this hypothesis, I couldn't: trans women, cis women, black women, white women and undescribed women were all treated more or less indistinguishably when touching faces[1]. I guess either Google patched out the bigotry between now and when the posts were written, or the blogs I'd read were cherry-picking and/or cherry-picked.
When I read the first three sentences you mention, the immediate thoughts that come to mind, not my considered opinions but more like the instantaneous autocomplete (as if I were skimming)
"A woman touched my face" mentally completes with an exclamation mark (as in, woo!)
The other two feel like contextually they would be part of a complaint.
I could make up a reason for why I feel this way, reasons come to mind, but I think that the real reason is going to come down to "I've read and heard a lot of sentences and these are the usual contexts that I've heard t... (read more)
Addendum: one thing that did cause a change was how casually the identifying characteristics were described. "a woman", "a cis woman" and "a trans woman" were all treated the same, but when I accused "a cisgender woman" or "a transgender woman" of facetouchery, the bot clammed up and requested more details before deigning to venture an opinion.