Definition Machine

Procrastinating again, I have been reading “Mind as Software of the Brain” by Ned Block (available here). Here’s a fun passage:

Defining a word is something we can do in our armchair, by consulting our linguistic intuitions about hypothetical cases, or, bypassing this process, by simply stipulating a meaning for a word. Defining (or explicating) the thing is an activity that involves empirical investigation into the nature of something in the world.

One can’t really argue with that. Block uses the example to point out that the Turing test examines intelligence in the first sense (whether an observer would call a machine intelligent) and not in the second sense (whether an investigator would find a machine to be intelligent); and he sets up the rest of his argument nicely with this simple distinction.

I bring up the example – and not the argument – because I want to take it on a tangent. Let’s say we build a machine that actually is intelligent. Would it define an item “j” intuitively or through an investigation?

The definition machine which defines “j” intuitively might match “j” against a list of known items to find the corresponding definition; where there is no match, the machine adds “j” to the list and provides a new definition (perhaps in terms of the other definitions). The definition machine which defines “j” following an investigation of the item may describe its concrete properties, its uses, and its material composition; the machine decides which features rank most prominent and pegs the definition of “j” to those features.

Interestingly, it’s hard to think of these as two separate machines – in part because our own minds utilize both processes – but examples of each exist. A search engine such as Google might represent an intuitive definition machine (Google simply finds things and puts them in a list, ready for near-instant reordering), and the Mars rover might represent an investigative definition machine (the rovers plot their own course across the Martian landscape while analyzing soil, sending results back to Earth). Now, imagine you put Google search algorithms on the Mars rovers: you might end up with an artificially intelligent definition machine. It won’t be able to do much besides make known certain facts about its environment, but it will do this extremely well.

This is a fun, slightly sci-fi topic that I think I’d like to return to.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: