On 3/29/07, YKY (Yan King Yin) <[EMAIL PROTECTED]> wrote:
Yes, I think I have "seed" capital, that is enough to get a conventional startup started. Also I believe getting subsequent VC funding is not that difficult.
That's more than most people have! I think the reason a lot of AGI people think in terms of working for free is that if it's in the "when I'm independently wealthy" category, it might as well also go in the "and I'll make it open source" category. But it seems that AGI is more complex than a conventional startup.
Longer and trickier route to getting a product, that's for sure. It is not programming that is the problem, at least not yet. Currently we
need the ability of a group of people to come up with *algorithms* that work within a certain architecture. Some of these algorithms are not well-known to standard AI students (eg inference, belief revision, probabilistic logic, etc).
Right, and before that the architecture that will enable these algorithms to work together. I estimate that a *minimal* AGI core may consist of ~100K lines of code. So
programming is not the problem. We need people who have a good understanding of the right algorithms.
Agreed. Russell -- I recognize that our ideas are quite similar and that your views
are usually very sensible, so there is a good chance that we can work together. But IMO we should wait for more partners.
Thanks for the compliment! Sure, no hurry. Another option is to work as a spin-off of Novamente, licensing some of
Ben's technology (eg probabilistic logic). Also we may borrow an established knowledge representation scheme, eg Cyc, or Novamente's. A dilemma is that starting from scratch seems too much work, but working with older tech (eg Cyc) requires modifications as well.
I think there's value in being able to connect to older tech (I'd look at things like relational databases before e.g. Cyc). I think it is necessary to start from scratch; it's not _that_ much work (I agree with your estimate of a version 1 framework being on the order of 100 kloc) and I think the risk of having to do it is much less than the risk of ending up with the wrong design due to trying to build on existing work. ----- 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/?list_id=303
