bolt12 was invented in 2019 and hasn't succeeded to gain any adoption -- I would say it's because it is too complicated and designed as if by academics with not much concern for the practicality of the real world. I think it's more likely that Lightning will be abandoned before bolt12 becomes as pervasive as Lightning Address, not discounting the other possibility.

But no, it doesn't fix anything because normal people shouldn't be running dedicated nodes, it's too painful, and standalone mobile devices can't work with bolt12 since they are not always online and accessible, the workarounds to make them work involve centralized companion services ("LSPs") and Google push notifications.

Also it is absurdly hard for custodial providers to adopt bolt12. It requires unpacking the Lightning node daemon open, injecting the custodian's internal database of users in there, then reassembling the node back.