That claim is utterly bogus. There's no indication that the languages have anything in particular to do with Africa, and indeed, no evidence that the language has 'evolved' more or less than any other. There is nothing unique to Andamanese languages to support the rapidly multiplying claims that they're 'ancient' or 'neolithic'. They're just isolated.

Yes, you're right, they're not "ancient" in any meaningful way. However, studying long isolated languages is still interesting, since whatever similaraties can still be discerned between (say) Indo-European and Andamanese languages were presumably present in the common ancestor of these languages, which probably existed more than 60k years ago. Thus by studying an isolated language, we might learn something about an ancient language.

There have been some attempts to study language drift over time and geography. The proponent that springs to mind is Joseph Greenberg. His work is an attempt to work backwards to the common core of language, the starting place.

If you knew approximately how fast language changed, and you saw two related languages, you would know when they drifted apart. There are also linguistic isolates like Basque where you can't really play this trick.

I don't believe that's something you could do with any accuracy with a claimed timespan of 70ky.