Typeless languages offer huge advantages in terms of the speed of development and maintenance.
Oh, my god, where to start?
If by speed of development, you mean working on a small project (e.g. prototype) that you write by yourself, or a small team of no more than three developers (who have similar level of mastery of the language and whatever frameworks you use). I can crank out 30K lines of Ruby/Lua codes quickly by myself and feel like speed of development is great too, because I know every line of code and what to expect of every parameter in every function and what to expect in return, etc. Try to do that in a multi-year large project. I'm working in a company where over 1000 developers are committing daily into a giant monolithic repository of Ruby/Rails code base. I'm not even going to go into how people abuse duck typing, pass hash of attributes from anywhere to anywhere without validation, include modules into classes with nothing in common, etc. Try to find where the function being called is implemented? Haha, have fun, especially when you have function names like "run", "perform", "execute" etc, with thousands of places called the same name, and especially when you have 3 dozens includes in that class. We spend so much time dealing with runtime exception that some "smart" people think we need to bolt the hacky Sorbet into the code base, while preaching the oh-so-great dynaminess of the language. Go figure. And let me tell you, adding Sorbet into Ruby is like dragging a huge metal ball behind while you run, and it was already slow to start with.
And maintenance. Man, I'm not even going to say anything.
In summary, use a typed language in large project with collaboration of a large team. It's stupid to use typeless languages in this case. I can't believe people jump through hoops to try to hack types into these things.