One comment on my recent post on deontology was to express incredulity that a term so many philosophers apply to themselves could lack a coherent definition. I, on the other hand, think that’s the most plausible thing in the world, and I’ve stumbled across a good example of it while doing my research for the Singularity Institute.
The issue is the meanings of the word “computation” and “information processing.” For the gory details, see here, but here’s my layman’s summary: the words “computation” and “information processing” get thrown around a lot in cognitive science (including AI, which is why I care about them), but there are some issues with what we mean by them. With “computation,” we know what Turing-computability is, but then there’s this thing called hypercomputation, which goes beyond Turing-computation. Should hypercomputation even be considered computation? Some have assumed the answer is “no,” others say “yes.” And “information processing” seems to be even less well-defined.
This seems to be a worrisomely common situation in the academic literature. Yes, it takes a lot of work to show conclusively that it’s happened in any particular case, but once signs of it start to appear, I think they should be taken seriously.
As for the word “deontology” specifically, my main concern is to discourage the meme that people who aren’t consequentialists must therefore be deontologists (unless they’re virtue ethicists, but virtue ethics is a kind of awkward third wheel of moral theory).
By the way, it’s this kind of thing that’s eating up my time right now and keeping me from blogging more.