Aaron, I will try not to make public speculations about your psychological states. Sorry about that. Jim Bromer
On Wed, Jan 23, 2013 at 7:49 PM, Aaron Hosford <[email protected]> wrote: > Or perhaps it is just too psychologically threatening for you to take that >> next step... > > > Just a quick FYI, public speculation about my psychological motivations or > fears is psychologically annoying. Please stop. I don't need a shrink, and > if I did, I would pay someone. > > > > Speaking of needing a shrink: > >> Insanity: doing the same thing over and over again and expecting >> different results. >> *Albert Einstein <http://www.quotationspage.com/quotes/Albert_Einstein/>*, >> *(attributed)* > > * > * >> >> When a reporter asked, “How did it feel to fail 1,000 times?” Edison >> replied, “I didn’t fail 1,000 times. The light bulb was an invention with >> 1,000 steps." > > > > We learn things from our failures. Sometimes all that's needed is a minor > change, and expecting different results is no longer insanity. And besides, > why can't a project just be so big or difficult that it takes more than a > year to see progress? So when people tell me that it can't be done because > it hasn't been done, I roll my eyes and stop listening. Inductive reasoning > is not applicable to the creative process. > > > > > The fact that people can make predictions that do not come true then just >> rationalize the failure away is powerful evidence that the power of >> 'prediction' as an AGI tool is nonsense. It is the ability to create new >> ways of interpreting the evidence that drives human creatiivity, not >> verification through prediction or an exotic method of combining possible >> outcomes given the evidence of the moment into a decision process that >> determines the next step. > > > I'm no big advocate of prediction as an AGI tool, but I think you > misunderstand how to use it. If I were going to build prediction into an > AGI, it wouldn't take the form of wild speculations about years to come. It > would be things like, "If I turn this handle the right way, water will > start flowing out of the tap," or, "If I honk the horn, maybe this person > will get out of my way," or, "If I try some new perspectives, I can > probably find one that simplifies the problem down to something I > understand." > > Notice the if/then form these take. These are immediate predictions > directly useful for reaching specific goals, and without the ability to > make these sorts of simple real-world predictions, we wouldn't be able to > accomplish anything in our day-to-day lives. I comb my hair in the morning > because I predict I'll be ridiculed if I don't. I turn the key in the lock > because I predict it will keep out burglars. I cook my dinner because I > predict it will satisfy my hunger. I pay my bill because I predict my > electricity will stay on. We all make these sorts of minor, conditional, > goal-oriented predictions all day, every day. It's not about creativity. > It's about getting through the day & doing the (practical!) things that > have to get done. > > And no, none of the predictions I mentioned are 100% reliable. But they > all make great heuristics, good enough to guide my behavior and improve or > maintain my quality of life. And if one is wrong, that's when reasons and > explanations come in. I use them to revise the conditions under which I > make a particular prediction based on when it fails and when it succeeds, > so it becomes more accurate, and consequently more useful, over time. > > > > > > > On Tue, Jan 22, 2013 at 8:49 PM, Jim Bromer <[email protected]> wrote: > >> Aaron, >> I did not want to say anything that might be deemed overly critical after >> you seemed to be willing to go along with my challenge, but I get the >> feeling that you did not actually understand what I was doing. >> I do not actually think that it is likely that I will get my program to >> work within a year. I was just challenging people who talk about prediction >> in AGI to try using it in real life, but when they do they have to be >> willing to accept the results of the the actual outcomes of their >> experiences as compared to their predictions. So while it feels like I >> should be able to write an AGi program within a year, I do not actually >> expect that I will be able to do so. So I said that if a year passes (that >> is if another year passes) and I still am unable to show that I have >> something interesting, I will have to accept the results of my experiment >> and recognize that the expectations that I had were not realistic...And the >> idea that it will just take another year aren't too likely to be realistic >> either - if I do not have something interesting to show for my efforts >> after a year (or a year and a half). >> Perhaps you were not able to totally get what I was actually saying >> because you are just skimming what I wrote. Or perhaps it is just too >> psychologically threatening for you to take that next step and recognize >> that if the programming is just too complicated to get it going in a couple >> of years then maybe it is just too complicated a problem for us. So if I >> don't succeed in a year then it will also stand as evidence that it is >> unlikely that you will succeed. Ok, maybe it is only weak evidence, but if >> you don't succeed in two years that will stand as evidence that you will be >> unlikely to succeed - unless you try another tack. >> I fully realized that people used the concept of prediction in different >> ways. I just don't see how that changes anything. If it is such a powerful >> tool then try using it in a way that is consistent with its proclaimed >> value. The fact that people can make predictions that do not come true then >> just rationalize the failure away is powerful evidence that the power of >> 'prediction' as an AGI tool is nonsense. It is the ability to create new >> ways of interpreting the evidence that drives human creatiivity, not >> verification through prediction or an exotic method of combining possible >> outcomes given the evidence of the moment into a decision process that >> determines the next step. Those methods are useful but all that I am saying >> is that the driver of intelligence is the rationally creative process. But >> if creativity is not used rationally then it can turn into delusion. >> I just found that the old salvaged program that I discarded years ago was >> an earlier version of the program. The more recent version was a greatly >> improved but it was unfinished and is too difficult to be useful to me >> because the bugs are multiple at many bug points. But I think I might be >> able to use my reviving memories of what I was trying to do back then to >> build a new simpler program. And with that simpler program I can begin >> testing some simple AI / AGi ideas that I have as I go along. And if things >> work out then I could make the program more sophisticated as I go along. >> So although I do not actually believe that I will be successful, I think >> that I have a better development plan than I did last time. However, the >> discovery that the more advanced version is just too complicated and the >> less advanced program not developed enough is a major set back. It is very >> negative evidence. This insight comes directly from the schedule and the >> mature recognition that a month out of a year with nothing to show is a >> substantial negative indicator. That is an example of how rationalism can >> be combined with creative insight to produce an insight of value - even >> though it is not an encouraging insight. How can I use this setback? I have >> a lot of functions that do work, and I have a lot of plans that can be >> implemented rather quickly. And I have the most serious mistakes that I >> made in the past to work with now. But I have to get the basic program >> going pretty quickly so I can begin some early testing of my AI / AGi ideas. >> Jim Bromer >> > ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
