Sunday, March 13, 2005



I think that it's a mistake to teach AI to any language other than
natural language.

Lojban is not a natural language for sure (because it wasn't really
tested for variety of real life communication purposes).

The reasons why strong AI has to be taught to a natural language, not to Lojban:
1) If AI understands natural language (NL) then it's a good sign that
the core AI design is correct and quite close to optimal.
If AI cannot learn NL then it's a sign that core AI design is wrong.
If AI can learn Lojban --- it proves nothing from strong AI standpoint.
There are a lot of VB, Pascal, C#, C++ compilers already. So what?

2) NL understanding has immediate practical sense.
Understanding of Jojban has no practical sense.

3) NL text base is huge.
Lojban language text base is tiny.

4) Society is "the must" component of intelligence.
Huge amount of people speaks/write/read NL.
Almost nobody speaks Lojban.

Bottom line:
If you spend time/money on design/teaching AI to understand Lojban ---
it would be just a waste of your resources. It has neither strategical nor tactical use.

1 comment:

Ben said...

The idea is not to teach an AI Lojban *instead of* English, but rather that teaching it Lojban might be a valuable *intermediary step*.

I'm confident that Novamente *can* learn NL, either via the intermediary of learning Lojban first, or not.

The question is merely of the fastest route of teaching, which may be different for an AI than for a human.

You say

If AI can learn Lojban --- it proves nothing from strong AI standpoint. There are a lot of VB, Pascal, C#, C++ compilers already. So what?

But those languages are programming languages, not languages for communicating commonsense knowledge about the world.

This comment suggests that you don't really understand what Lojban is.

Teaching an AI Lojban has the practical sense that Lojban can be used to interact with the system and teach it all sorts of useful commonsense information about how to interact, think and so forth, as well as about the human and natural world.

This knowledge can then be very helpful to the system in learning English, among other purposes.

I am quite *sure* that teaching a Lojban-fluent AI English would be a lot easier than teaching a nonlinguistic AI English. This is obvious.

Now, whether getting to English via Lojban is the optimal path, is a whole other issue. I don't claim to be sure about that. All I can say at this moment is that it seems to be a path well worth exploring further. Your arguments definitely have not convinced me otherwise; particularly your analogy between Lojban and programming languages seems quite ill-conceived. In fact Lojban is far more similar to natural languages in both intent, semantics and syntax than to any of the programming languages you mention.