"Autism" as a diagnosis sort of prescribes people be treated in standardised discriminatory ways from early childhood for their nominal best interests, rather than simply giving you a set of advice you can pick and mix from. This is likely to result in homogenous and harmful treatment in practice.
I'm also DEEPLY sceptical of autism I.E. clinically significant differences in social behaviour all being caused by some unified biological cause, given the decades long failure to find this smoking gun, my conclusion is that DSM-V categorisations are a social construction which don't map 1:1 with what is going on biologically. The more I learn about the sheer variance autistics really have in symptoms, and how we see clusters of symptoms like insomnia or digestive issues which don't even appear on the DSM-V at all, the more convinced I get that autism is really more than 1 thing. We even know about things like fragile X syndrome, which most autistics don't have, but which reliably causes a high incidence of autism. The project which seems to have the most momentum in overhauling DSM-V taxonomy is the RDOC framework. I honestly think that autism is multiple overlapping conditions is downright obvious, and I think the main reason it isn't taken more seriously is that Autism has become a useful brand.
Maybe the confusion comes from „no two autists are alike in their symptom signature“ (which seems to be true), vs „most autism cases can be treated by X“ (which seems to be true at least sometimes). While it’s a highly individual syndrome, the metabolic and other causes could be more general. If that’s not the case - then we might actually need sub-syndromes, and I think this is actually WIP. Eg there‘s now pathological demand avoidance (PDA) which could, if it occurs outside of autism, become its own disorder.