Can you prove that limiting the human (errors) leads to better code?

Do we need to prove this? Ok, anyway, neither TypeScript nor any other helper tool is meant to end all bugs in coding. TS is a tool designed to help, and people are supporting the idea that it helps, as shown by its adoption numbers.

Making an analogy, I would say we have Java, Python, C++, Go and many other high-level languages even when we already had much more performant low-level machine languages, because they make developer’s lives easier when it comes to understanding and traversing a codebase, just like TS in relation to JS: although the leap is smaller in this case it is still pretty significant.

This help provided by TS may reflect on the bug situation too, because many bugs are indirect consequences of a narrower view of the codebase, but as bugs may have many reasons to exist other than this, it is not realistic (and fair) to think that TS is meant to be the end to all bugs.