That's a good point, but isn't the distribution of "local boy makes good" stories a bit of an edge case? I mean do we really need AI to do that for us at the risk of something bad happening?
I'm not trying to imply that you personally agree with LinkedIn's decision, by the way; just trying to get my head around why they think it makes sense.
I think they see it as an issue where most of the time it's "close enough" that if it's wrong it'll just be wrong in a way that people will laugh off and that will still generate engagement.
As long as the cases where it goes wrong enough to be offensive and hurtful are rare enough, there's very little downside to them.
That makes sense as the closest answer (but still guessing).
They could further limit it by location/region or companies mentioned to improve the quality but that may be overkill and reduce the number of notifications too much for their tastes.
I'm not trying to imply that you personally agree with LinkedIn's decision, by the way; just trying to get my head around why they think it makes sense.