[tonmoy]

I imagine in real life applications you’d need to add a bit of checks and balances around the ML output (e.g. cross check against a dictionary/copyright database)