There's still a very real use for smart contracts in that you change what bad actors can do to act badly. Before they may have been able to breach a contract by ignoring invoices now they breach a contract by exploiting a smart contract loophole. Basically your shuffling the trust and risk around which can be a useful tool. i.e. it can be quite costly to enforce (and do due diligencece on) a contract with someone in another country so the cost and technical risk of setting up a smart contract may be much more preferable than the posisbility of having to legally enforce redress for breach in a regular contract.
Smart contract is really the misnomer. In reality, they are automations of contractual obligations and cannot automate complete contract clauses.
I think some confusion arises because that "smart contracts" only make sense if code really is law, in the sense that any transaction executed by the contract -- even unexpected, surprising transactions -- is considered to be fully consented to by all parties interacting with the contract.
I agree that that's a terrible idea - bugs can always exist, and having no recourse when millions of dollars are lost due to a coding error is a huge and unreasonable risk.
But otherwise -- if, ultimately, courts can force "smart contract" transactions to be unwound if they are found to be exploitative, unintended or otherwise invalid -- then what's the point of having smart contracts in the first place? What's the value proposition? Why not just use regular contracts?