I watched the video and your ideas are interesting although I am not quite sure what you are getting at.
I think that AI needs learning rules. But some people might say that all modern AI paradigms have learning rules. So some disambiguation is needed right away. You seem to be saying rather than looking at a great many representations of objects (in one venue or within a narrow range of variations) an AI program needs to be able to creatively decide that some map, for instance, can refer to some particular place or kind of thing using learning rules and a collection of seemingly simple pieces of knowledge about the place or object. Your example is that a car has wheels and a compartment so a picture that shows a box with 2 circles on either side along the undercarriage can be interpreted by a human being as a representation of a car . But then again it can also be interpreted as a table with 2 chairs. I agree with that and I think that conceptual projection and creative problem solving is something that is absolutely necessary and it is something that should be easy to implement. But, part of the problem in my opinion, is that conceptual integration has to be more intricate than (for example) just projecting concepts onto other conceptualizations. So to make this quick I am working on an AI program that will allow me to make simulations of what I want the program to be able to do so I can better understand the types of events that will happen or need to happen to implement conceptual integration. I also think that learning through communication is important and my AI program is going to be designed to allow the program to learn through communication and through trial and experience. I am not certain that I via communication and you mentioned that this kind of learning could be evolutionarily fast. My idea is that ideas or concepts can play different kinds of roles when used with other ideas or concepts. These relationships can be used to narrow and/or shape the trial and error processes of learning (including creative problem solving.) In human experience there is not a clear distinction between experiencing an event and learning about it and this points to a problem with most contemporary AI methodologies. I am pretty sure you agree with this. But this also points to a methodology that should be implemented and it suggests that learning rules are not only employed while in training mode but in the every day situations where intelligent response is required. Jim Bromer On Sun, Jan 10, 2016 at 5:59 AM, Danko Nikolic < [email protected]> wrote: > Dear Jim, > > I agree with your point that deep learning machines cannot think outside > the box. > > However, there may be already some conceptual progress in this respect. > I have recently made a proposal how to make machines that think more in a > biological-like manner: The suggestion is that the machines do not store > their knowledge in synapses or by similar means but instead, in a set of > specialized learning rules. That way, when machines think, they literally > must think outside the box because they have to think by applying (very > fast) learning. That is, thinking does not occur "through internal > computations" but through interaction with the environment. The argument is > that this will enable machines to achieve understanding that J. Searle was > asking for. > > This proposal is described in this recent writing at IEET: > > http://ieet.org/index.php/IEET/more/nikolic20160108 > > And a more condensed version is in this TEDx talk: > > https://www.youtube.com/watch?v=zZMlzMTR6l8 > > > Would you think this effort is helpful for the problem that you are > pointing out? > > Thank you. > > Danko > > On 09/01/16 21:31, [email protected] wrote: > > This is a digest of messages to AGI. > Digest Contents > > 1. If Deep Learning is It then Why Are Search Engines Incapable > ofThinking (Outside the Box or Otherwise)? > <#-1559021165_20160109152915:A46324DA-B70F-11E5-AEF6-CFF8EF10038B> > > If Deep Learning is It then Why Are Search Engines Incapable ofThinking > (Outside the Box or Otherwise)? > <https://www.listbox.com/member/archive/303/2016/01/20160109152915:A46324DA-B70F-11E5-AEF6-CFF8EF10038B> > > *Sent by Jim Bromer <[email protected]> <[email protected]>* at Sat, > 9 Jan 2016 15:29:08 -0500 > If industry has AI pretty well figured out then why are search engines so > incapable of thinking outside the box? The conclusion looks inescapable to > me. Yes there will be a day when someone makes a significant achievement > while the rest of us might miss it completely but the idea that > contemporary deep search (or some other AI method) has achieved the hype or > the implied conquest that winning at chess and jeopardy seems to imply just > does not jive with the computing power Google, Bing or IBM have. There is a > substantial disconnect between low level -almost- human reasoning and deep > learning. Jim Bromer > > *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > <https://www.listbox.com/member/archive/rss/303/27154149-3c484689> | > Modify <https://www.listbox.com/member/?&;> Your Subscription > <http://www.listbox.com> > > > *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > <https://www.listbox.com/member/archive/rss/303/24379807-653794b5> | > Modify > <https://www.listbox.com/member/?&;> > 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-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
