At the scale of Facebook you can do interesting things that are hard to catch, like:

- Make “range” moderation rules and use the algorithmic feed for activity to be suppressed or amplified depending on desired global outcome.  Collect data and feed it to your AI processor to let it figure out which posts, groups, users, posts should get more exposure and which should be “limited” in order to achieve higher page views counts or a more positive perception of certain brand/idea/believe/concept. 

- Generate fake “reactions” to advertised content by displaying different original posts to viewers, getting likes/comments/shares and attributing them to advertised content when viewed by the advertiser. Usersndont care because of transient nature of Facebook feeds - their content reactions is buried after a few days for them (but is sold to advertisers as a main value proposition). I even made a video of this very same scam they did and forced FB refund all our advertising payments a couple of years ago. Now they banned us from advertising. 🙄 

- Condition perspectives to provoke ad-clicks/purchases. For example, promote organic posts that AI identified psychological anchors driving purchase decisions and match them with ads. Say, make people feel overconfident about they income and then push ads for personal loans. 

Now, that is the real reason why MarkZ was so adamant about getting a free pass on AI regulations when Elon Musk called for extra oversight. It’s not because they want the Messenger to give better handy tips, it’s because they want AI to save their asses and drain everything they can from the users