today's article: self-healing programs

Today I read about IBM’s self-healing programs.
I don’t expect it to be as smooth as it is made out to be, but this is adventurous. I was first of all wondering why terms like autonomic computing, artificial intelligence, regenerative systems etc gives a kind of thrill like a fantasy or science fiction. Anyways, it need not be a macro phenomenon, but the principles could be applied in minute way in routine applications as well. Systems doing preventive maintenance sort of checks (like the on-board diagnostics of car and “check engine” warning light) – it should be possible to put in pre-coded periodic self-checks. Systems trying to do some basis error correction before throwing it out to the user (the very first data format error need not be thrown and program aborted – it could try to correct the format and continue). Systems could learn from how it is used and present better ways – frequent navigations, frequent combinations of data made available. Systems doing self auditing on a periodic basis and sending reports. Not only finding problems, but suggesting solutions for it.