I thought the writer means to provide directions to professional help. This is less complex than providing professional help.
Sent from Xiaomi 2014817 using GAGT
The thing is, we are taking our language for granted.
In a keyword search, you can pick up words like "suicide" or "depressed", but words can have more than one meaning, and that depends on context, history, personality, and semantics.
"Lol. The ant committed suicide" will trigger a positive response in a naive keyword search. So will "There is a depression in the flooring, how can I fix it?", or even "My actions are clearly social suicide".
In some cases (e.g. the second case), a depression hotline is unhelpful. In the first and third case, it would at best prevents true positives, and at worst make for distasteful comedy.
The problem is a hard one. You might suppose that Siri and Cortana should direct professional help, even in cases of a false positive too, but that could compromise the usefulness of these helpers, which could decrease userbase - which, in turn, decreases its suicide prevention effectiveness too.
It is not about providing professional help or directions - the problem lies in the fact that, the computer doesn't even understand or realise that help is being asked for.
