Sure, but the nature of AGI is that wizzy demos are likely to come fairly late in the development process. All of us actively working in the field understand this....
-- Ben G On 6/6/07, Mike Tintner <[EMAIL PROTECTED]> wrote:
Ben, I'd be looking for a totally different proof-of-concept for AGI. (I brought in Hawkins - not to rehash our arguments - because I consider him an example of good proof-of-concept practice). I'd be looking for proof of higher adaptivity (to use Peter's term). If Peter Voss' Maze Explorer could, for argument's sake, find a totally new kind of path through the maze, or develop a radically new way of moving - in one way or other, fundamentally change the "rules of the game" - that would do for me to start with. And that, I think, would do for the world - and get comparable attention to what Hawkins got. [Please don't hold me too closely to my definition of higher adaptivity - I would think you know essentially what I mean]. I'm only echoing what you yourself said: "I believe I have a description of software that will do this, when all the details are worked out ... but without a proof (mathematical or far better by demonstration!!), I realize this isn't going to get anyone really excited..." It takes a proof-of-concept or demonstration to get people excited - and, to repeat, it can be v. v. simple. People ARE excited even if still sceptical about Hawkins who has only demonstrated a kind of object recognition not AGI. Ben: Mike, putting together a demo of a machine learning system recognizing objects from simple line drawings would take me less than one month, using textbook technologies. Not worthwhile. Putting together a simple reinforcement learning system doing the same stuff as NM does in that fetch video would not take much more time than that either. ------------------------------ 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/?&;
----- 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=e9e40a7e
