In university I did modules about dinosaurs and modules about autism but I did better in the former so I will say on balance, I know more about dinosaurs than autism. My professor in particular was of the mind that DSM-IV and especially DSM-V make relatively common human behaviours into clinical disorders. When I was updated on DSM-V, I had to admit that I all but threw up my hands and shouted "this shit is bananas!". Very nebulous definitions, ambiguous classification and new "catch-as-catch-can" disorders made it kind of sketchy within the medical community.
I was taught that autism is a form of developmental retardation affecting emotional intelligence which often manifests as an inability to notice and understand social cues. Both neurological architecture and neurotransmitter deficiency has been blamed for it; the current science is leaning towards a bit of column A and a bit of column B. There is a certain subset of the internet that self-diagnoses, largely as an excuse for poor behaviour. Then again this is the internet, what did you expect? Clinical diagnosis is often difficult due to conflation between autistic behaviours and certain personality types including natural introverts, contemplatives and people who are simply socially awkward. As far as I know, no definitive physiological test exists for autism which puts its status as a psychiatric condition under a small degree of dispute, there is a small minority of psychiatrists that believe it to be a psychological condition rather than a physiological one. The consensus seems to be that we simply lack the tools to diagnose from a purely physiological standpoint.