Anastasios,
I sympatize with your skepticism, but please be mindful that resource limitations and theoretical limitations are not the same thing. I have no doubts that solving the problems you suggest will come to reality, but to do that it will be necessary to have a big machine and train it, and I am not sarcastic here, just like a kid goes to school. A pre-K does not know what chess is or what forex are, yet she can learn, and, in 20 years, she will be a trader of a world-class chess player. The difference, is that with AGI you do it only once, with kids you do it with each one. How large the pipeline? Well, the size of the human brain. Computers are just about there, but anyway, let's do an ant first. And, BTW, any computer program is a causal set. I have published the conversion from programs to causal sets. This means that, if you can write a program about it (and people have written programs about nearly everything) you can click a button and get a causal set (not kidding here either, this needs a lot of work, but it is routine work, not great science). Thinking that causal sets need to be done from scratch is a common mistake. You start with a coarse granularity, then progressively refine it. You can start with chess or forex if you want, and then refine progressively. It learns, just like a kid. Please read my post to Jim about an hour ago. Sergio From: Anastasios Tsiolakidis [mailto:[email protected]] Sent: Sunday, September 02, 2012 3:15 PM To: AGI Subject: Re: [agi] Granularity On Sun, Sep 2, 2012 at 9:56 PM, Steve Richfield <[email protected]> wrote: not only arriving at conclusions, but PROVING them right or wrong. Wouldn't it be wonderful to see a computer program that could do THAT? Well gentlemen, I would like to believe Sergio and his causal vision, but I can't bring myself to bridge the gap between information and world. Certainly everything a deterministic computer does is a PROOF, but matching the "program" to the "world" is practically and theoretically nearly intractable, especially if it has to be done from scratch, let's say by observing single photons (I have previously asked how wide a datapipe do you think your reality cruncher will need to get to grips with... well, reality). I'd like to see how well causal sets do with non-AGI problems, let's say chess and forex trading, and, dare I say, codebreaking. Of course there could be many objections to such non-smooth problems, such as my grandmother's general intelligence in the absence of chess and mathematical skills. Still, we need to draw some conclusions from domains where we know the "functional", at least to a significant degree. AT AGI | <https://www.listbox.com/member/archive/303/=now> Archives <https://www.listbox.com/member/archive/rss/303/18883996-f0d58d57> | <https://www.listbox.com/member/?&;> Modify Your Subscription <http://www.listbox.com> ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-c97d2393 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968 Powered by Listbox: http://www.listbox.com
