Hi,
I know you must be frustrated with fund raising, but investor
relunctance is understandable from the perspective that for decades
now there has always been someone who said we're N years from full
blown AI, and then N years passed with nothing but narrow AI progress.
Of course, someone will end up being right at some point.
Sure ... and most of the time, the narrow AI progress achieved via AI-directed funding has not even been significant, or useful....
However, it seems to me that the degree of skepticism about AGI goes beyond what is rational. I attribute this to an unconscious reluctance on the part of most humans to conceive that **we**, the mighty and glorious human rulers of the Earth, could really be superseded by mere software programs created by mere mortal humans. Even humans who are willing to accept this theoretically, don't want to accept this pragmatically, as something that may occur in the near term.
After all, there seems to be a lot more cash around for nanotech than for AGI, and that is quite unproven technology also -- and technology that is a hell of a lot riskier and more expensive to develop than AGI software. It is not the case that investors are across the board equally skeptical of all unproven technologies -- AI seems to be viewed with an extra, and undeserved, degree of skepticism.
For the record, at the same event, Peter Voss of Adaptive AI
( http://www.adaptiveai.com/) stated his company would have AGI in 2
years. I *think* he qualified it as being at the level of a 10 year
old child. Help me out on that, if you remember.
I could help you out, but I won't, because I believe Peter asked those of us at that meeting **not** to publicly discuss the details of his presentation there (although, frankly, the details were pretty scanty). If he wants to chip in some more info himself, he is welcome to...
Peter has been more successful than Novamente has at fundraising, during the last couple years. I take my hat off to him for his marketing prowess. I also note that he is a lot more experienced than me on the business marketing side ... Novamente LLC is chock full of brilliant techie futurists, but we are not sufficiently staffed in terms of marketing and sales wizardry.
I have my disagreements with Peter's approach to AGI, inasmuch as I understand it (I know the general gist of his architecture but not the nitty-gritty details). However, I don't want to get into that in detail on this list, for fear of disclosing aspects of Peter's work that he may not want disclosed. My basic issue is that I do not, based on what I know of it, see why his architecture will be capable of representing and learning complex knowledge. I am afraid his knowledge representation and learning mechanisms may be overfitted, to an extent, to early-stage "infantile" type learning tasks. Novamente is more complex than his system, and thus getting it to master infantile learning may be a little trickier than with his system (this is one thing we're working on now ... and of course I can't make any confident comparisons because I have never worked with Peter's system and also what I do know about it is quite out-of-date), but Novamente is designed from the start to be able to deal with complex reasoning such as mathematics and science, and so once the infantile stage is surpassed, I expect progress to be EXTREMELY rapid.
Having summarized very briefly some of my technical concerns about Peter's approach, I must add that I respect his general thinking about AI very much, and admire his enthusiasm and focus at pursuing the AGI goal. I hope his approach **does** succeed, as I think he would be a responsible and competent "AGI daddy" -- however, based on what I know, I do think that Novamente has far higher odds of success...
-- Ben
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/[EMAIL PROTECTED]
