So I've actually found 1 case similar to what's mentioned above and I believe 2 cases where it may genuinely be helpful.

The bad case is due to how differently people name and date the same ancestors in trees, I suspect this is the root of the problem for many others, I sympathize with Ancestry though since compiling all the variations down without being error prone (combining people who are no the same people) is easier said than done programatically. It seems to point me back a few generations as a possible ancestor when 2 generations or so down both matches share the same group (but again name matching foiled it). When I realized that I noticed the surname and area and suspect the connection is something far different than the ancestor revealed.

In the two other cases I see some of the same name matching issues but only by a generation or so and I suspect those might genuinely be ancestors since the names they intermix with are lines that I've been blocked in early to mid 1800s or so.

Will be interesting to see them evolve this... I highly recommend everyone send feedback and to be constructive. If you don't like the feature just ignore it, simple as that.