Literacy is recent. In the 1970s roughly 35% of the population was still illiterate. Go back further and you're looking at 10% or less. Most signs were pictorial. A blacksmith was denoted by an anvil, a tailor by a spool and needle, etc.
If you were dyslexic, you'd never know.
Autism is likely in a similar boat. Those individuals with less disruptive traits could probably find themselves doing well. Possibly joined the priesthood where they can apply their focus creating manuscripts. Or become hermits, where they could live somewhere less stimulating. Whereas those individuals with highly disruptive traits were probably treated quite poorly. Crammed into sanitariums or confined to monasteries to exorcise whatever evil spirit may have possessed them. Or simply died from being rejected by their families.
We can't really know. We can probably safely assume that while our knowledge of autism is new, it itself is not new.
We've known for a while that autism is highly heritable. This paper builds on a large body of work that links specific genes to autism's heritability. I haven't dug into the specific breakdown, but my understanding is that a majority of genetic variants associated with autism are inherited, rather than the random de-novo (new) mutations.