That was the entire value-add of Cambridge Analytica , whose Facebook-API data-gathering loophole has now been replaced by just engaging suckers via the platform itself and a tiny bit of NLP/sentiment analysis.
The most obvious one: with the security question type stuff, you can take over other people's accounts.
But collecting data about people is also useful if you're trying to spread an agenda. You can determine what types of messages resonate with an audience. You can group those people and target them separately -- not through ads, but through special interest accounts/groups. You can recruit people to amplify your message. You can even get people to act in real life, i.e. https://en.wikipedia.org/wiki/Internet_Research_Agency#Ralli...
If you can identity, categorize, and influence the loudest voices, you can influence public discussion and opinion.
AFAIK you absolutely can target (or could target) groups of people based on very specific criteria.
Targeted advertising lets you run a campaign that never could be run before... one where you appear to be something different to different people. If you were trying to seize control of three different warring groups, you could advertise to A, "We'll kill B and C!" , to B, "We'll kill A and C!" and to C, "We'll kill A and B!" which you couldn't do in a stump speech without people figuring out the ruse.
By building a detailed psychological profile of individuals, you can build a model that allows you to tie their responses to these questions with the political messages they're susceptible to. Cambridge Analytica paid a few hundred thousand people to do a quiz where they shared their Facebook likes and answered questions about their personality. CA then used that to build a model that showed "People who live in Slidell, Louisiana and like Dodge Ram trucks will be most receptive to messages about illegal immigration and are generally supportive of state violence". Then they can run that ad to everyone in Slidell who likes Dodge Ram pickups.
It seems the crux of the issue here is that people are being fooled into supplying data about themselves in a non-consentual manner.
The former has been happening in politics forever and imo the latter has been every tech companies MO essentially for the last decade.
Tools are not immoral.
Immoral uses of tools are immoral.
FB's targeting tools go to some lengths to only allow you to target by dems on fb and associated properties. These questions, plus profile views, allow you to extract the information for external use. Eg selling a list of <fb id, first car, birth year, favorite color, pet name, etc> tuples.