----- Original Message ----- From: "Benjamin Goertzel" <[EMAIL PROTECTED]> To: <[email protected]> Sent: Wednesday, April 18, 2007 3:51 PM Subject: Re: [agi] rule-based NL system
> If you're going to restrict your AI to a special subset of English, > then it can't read free text anyway... all you can do is chat with it. > So why not just chat to it in Lojban which has full expressive power, > barely ambiguous semantics, and totally unambiguous syntax? One main reason. How many people know Lojban? How much time is required by a person to be proficient? People that have only a single spoken language now (computer languages excluded) have a *very* difficult time learning a new language. (Unlike some others) The internet is full of the English language, *not* Lojban. etc Unambiguous syntax is trivial and all computer languages have this ability. Why not use Java instead of Lojban if only non ambiguity is important? (I don't advocate this at all!) The problem of any NL is that it requires context and understanding to correctly interpret a persons' meaning. A grammatical parser and dictionary are far from adequate to interpret NL. This problem needs to be solved whether Lojban or English are used. If the understanding and context are solved, the ambiguous nature of English won't be a problem for the AGI any more than it is for all those that speak and work in English now. My conclusion is that Lojban over English is a solution to a problem that won't help the AGI effort. BTW why can't an AGI communicate in a subset of English to start and grow the number and complexity of it's vocabulary over time? If a subset of English is initially used, why does this imply *restrict* as you mentioned above? Children don't start off with adult level language either! -- David Clark ----- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415&user_secret=fabd7936
