The programmer creates clean code. In all languages most programmers do atrocities with code because most refuse to learn real programming.
Kotlin helps to have more concise code compared to Java, nothing more. In fact the code can even be considered verbose compared to some other languages. And that’s not necessarily bad.
I have not seen anything that makes a good comparison between the codes before and after the conversion, but my experience with this type of converter is that if the semantics is not exactly the same between the languages the conversion never gets very good.
Convert C# to VB.NET and vice versa is good "80 or 90%", IE, then need to take a fancy in hand. But these languages were created almost by the same people, and it was supposed to be practically a different syntax with the same language. It is not 100% because VB.NET descends from VB that had some bad things and there is total compatibility with C#.
Kotlin has important semantic differences in relation to Java, as well as several similarities. My estimate is that it will be a "60, 70%" good.
Note that you can convert 100% (not everything), only the quality is that it will not stay 100%. I think you will have gains, but not so much when doing in hand. It can be an initial step to go through the whole code. Several things cannot be converted as well as a human would. It can help to learn, but it can help to unlearn because it will not be a good code.
In VB.NET and C# the library is the same. Kotlin brings a good part of the own library that works better and little or none of this will be converted. So depending on the case may have an even lower quality than it might be.
I think if you are going to program in Kotlin you have to embark on everything, you should avoid doing any part of the Java way, which is what will happen with the conversion. But then it’s everyone’s opinion.
Are creating a Swift to Kotlin converter, which should yield worse results.