Not so. These tools are currently in use in web apps which guess user genders. I have actually witnessed embarrassment caused when invitations to sign up or collaborate get sent out with an incorrectly guessed gender.
This feature is planned to be compulsory, therefore the web apps responsibility to disable it if their audience is sensitive to such errors.
The bug in this case is in English, so the English language has to be fixed to reflect better to the times we live in. Tools like nlp_compromise are reflecting to the current state of English, not the other was round.
In theory, language is a tool that you can replace if you are not happy with it. People who feel offended by the gender-aware languages have to option to use a gender-neutral language, such as Hungarian. ;)
(Side-note: I more than happy to support a curated version of English, but only if it becomes a phonetic language. That would make billions of people's life much-much simpler.)
The commenter asserting that knowing a user's gender is not useful for marketing is wildly wrong. I agree that showing pink to a user you think is female or only showing trucks to males is laughably stupid, but we have had campaigns that were not designed to target a specific gender that had a significantly higher conversion rate from one gender. It would have been a huge waste of money to not change the targeting criteria for the ads. Even an imperfect algorithm could save lots of money there if it's more than 50% accurate.
That said, trying to guess gender based solely on name without any other context does seem woefully unlikely to do well.
14 comments
[ 3.7 ms ] story [ 43.1 ms ] threadThe bug in this case is in English, so the English language has to be fixed to reflect better to the times we live in. Tools like nlp_compromise are reflecting to the current state of English, not the other was round.
In theory, language is a tool that you can replace if you are not happy with it. People who feel offended by the gender-aware languages have to option to use a gender-neutral language, such as Hungarian. ;)
(Side-note: I more than happy to support a curated version of English, but only if it becomes a phonetic language. That would make billions of people's life much-much simpler.)
That said, trying to guess gender based solely on name without any other context does seem woefully unlikely to do well.
If you use google search, you will know they do this sort of analysis to get you better search results.
People forget this very important concept in open source all the time.