Profiling of a large population, you put them in cohorts, and slowly shift what you show to these cohorts (based on their preferences, worldviews, etc.) to slowly nudge them into a worldview you'd like. It won't be 100% effective but it can definitely shift perceptions, if each cohort is siloed into their own reality bubbles through what you show them you can stochastically nudge them into a view you want them to hold based on their preferences.
If marketing works even to the people aware of how it works, a concerted effort to use someone's profiling data telling what do they like, dislike, will definitely work on a majority of users.
It's not like it will be blunt, it only has potential if you use this data to slowly shift views by using what's most effective to each cohort, with a large amount of data you can be quite precise in defining these cohorts and using different strategies/tactics for each one depending on what's most effective.
Have you ever worked on anything that did profiling based on accumulated data? I've worked on a few projects back in the early 2010s and even at the time it was scary how much you could infer about your users based on some 100-200 data points collected over a period of 2-5 years. Weaponising that is not the complicated part, the data collection is.