It's about establishing a quorum for the cluster. Generally there is an odd number of votes needed to establish quorum to ensure things can operate correctly.
A 2 node FCI would have a quorum disk or a fileshare witness or a azure storage account as the thing that establishes quorum. Should 1 item fail, you still have quorum 2 of 3 is majority, if 2 items fail, you're as dead as a dodo as 1 of 3 is not majority.
As AG's are somewhat reliant on the FCI concepts (unless you do cluster less AG's).
Now for a 3 node AG setup which is within the same site "Building1" your fine with the setup as you can loose 1 of the nodes, everything is fine. Loose 2 nodes your dead. Same as you would be in a traditional cluster. That's all well and good as your only providing local HA in "Building1"
If you want to do DR where you have 2 nodes in "Building1" and 1 node in "Building2" (B2 being your DR site), should the proverbial hit Building1 and you have to invoke DR, as you've lost 2 nodes, 1 from 3 is not majority so the cluster and resources are down. You need to force failover to bring it online (as I say that always seems dirty to me to force it).
So you need to find a way to have a majority vote should the worst happen. So add 2 nodes in DR if money can. Up the vote of the node in DR from 1 to 2. Then add in an external witness typically Azure storage if you can saves having a 3rd Building to bring a file share witness.
That way you have 2 votes in "Building1", 2 votes in "Building2", 1 vote in Azure. So if "Building1" was to die, 3 of 5 is majority, service is operational. Lose "Building2" again 3 of 5 is majority. Lost Azure, 4 of 5 is majority.
But if your not using the AG to provide disaster recovery, and only doing it for local HA then your setup you have now is setup correct.