/ artificial intelligence

Do we even have a Philosophy of Mistakes?

Most of the formal literature that I've found on making mistakes focuses mainly on business and applied ethical scenarios -- also included were mistakes in education, nursing and other very down to earth topics. Information on mistake making in computer science led me down paths generally tied to concepts like pulling together projects after someone has goofed and oddly enough piles of articles on women in computer science came up[1]. Searching within philosophy produced, of course, the most celebrated theory in philosophy "everyone who came before me was wrong".

So, at least insofar as I can see, Philosophy of Mistake Making is not one of the core tenets of inquiry[2], but it is pretty interesting when you get down to it. The focus is inevitably (and perhaps practically) going to be on the things that we know rather than the complement of that set, perhaps this is at least partly because it is easier to talk about what you know than what you don't, and talking about that time you were wrong can make you feel a bit like an idiot.

Error is made consciously, and unconsciously. Humans make mistakes, but we also describe things like cells 'making mistakes' and incorrect copies of DNA being made[3]. I'm not sure if we'd classify these two types of mistake as the same thing. But they both seems to derive from the same general idea of a mistake, which we'll roughly outline as 'the incorrect result of a particular action or hypothesis'.

Another distinction that can be made in types of mistakes is whether or not there is a correct answer and if it can be known. I don't know what the cubed root of 1277 is and if asked to give the answer off the top of my head unless I was frankly the best guesser in human history, I would probably make a mistake, but I could find out[4]. The confounding problem is thinking of a mistaken belief that you couldn't find the answer to -- perhaps this is because if you don't believe you are wrong you're not likely to know you need to find a better answer.

Just some thoughts, I'd like some material to review going into writing a longer paper about error in artificial intelligence but I just can't find where to begin.


  1. Should I be offended? Please advise. ↩︎

  2. Barring things like justified, true, belief being a mainstay of Epistemology ↩︎

  3. Mutations, I suppose is the proper way of putting this ↩︎

  4. ...using Wolfram Alpha ↩︎