You are mixing the two things. It is not just miners, but the entire network of full nodes that ensure only valid transactions are included in a block. When a block is mined by a miner, it is propagated on the network. All the full nodes go through a detailed checklist (see here) to ensure that the block and the transactions within the block confine with the consensus protocol. This is unlike ledger based systems where an 'authority' is appointed to vote on transactions to maintain a state of the ledger.
Since Bitcoin is peer-to-peer, there is no central authority to control it. As a result, anyone can send any kind of transaction to the network, whether or not it is valid. You could simply send a transaction that sends someone else's coins to yourself. Now, if a miner tries to mine a block with those invalid transaction, the rest of the network will reject the block. So the entire effort (read electricity and time) that the miner put in will go waste as that block will not be added to the blockchain by the rest of the participants.
The 'random hash problem' also has other benefit. It ensures that Bitcoin blocks are mined every 10 minutes. Every 2016 blocks (or ~2 weeks) the difficulty is adjusted to ensure that the average block time remains at 10 minutes.
And when do the other miner nodes validate that the proposed block is correct after one miner finishes?
As soon as the miner mines a block (finding a valid block whose header hash confines with the current network difficulty), it will propagate the block through the network. Now, the miner not just sends it to other miners, but also all the full nodes that are connected to it. If the block is valid, the full nodes/mining nodes will relay the block to the nodes connected to them. If invalid, the block will be rejected outright and not relayed. When other mining nodes see this block, they will mine the next block on top of this block (provided the block they received was valid).