Yes there are many program and or is it or sub-program generators in computer science. Any kind of AGI program - or an AI program - can be seen as an algorithm generator or a sub-program generator. And if an AGI program was able to create good solutions at a level comparable to humans they might be called solution generators. Since a currently feasible AGI program does not reach those levels the name "solution generator" seems pretty pretentious.
However, I believe that we can create classes of algorithm generators that would work to generate algorithms that could be used by an AGI program. At first I thought of a super-algorithm generator but I realized that even if it was feasible it would require too many parameters. So I started wondering about narrower classes of algorithms. Would this be useful in an AGi program? It can be very difficult to distinguish between a generated algorithm (or sub-program) and the output of a program that writes a sequence of steps. But my question is whether or not it would make sense to design a group of algorithm generators that are explicitly capable of potentially outputting an immense number of algorithms would be useful in an AGI program. I am studying AI Planners in one of the online courses that I am taking so I am interested in your comments, but there is no contest in the ideas that we are talking about. I am only wondering if an explicit algorithm generator, which for example, generated analytical algorithms, might be useful in an AGI project. I am not talking about solutions or programs except that the algorithms would have to be functional. That is they could not be allowed to crash and it would be important to generate interfaces to go along with them. An algorithm or sub-program is not necessarily a solution or a program but the terms are relative. It could be said that an algorithm is a solution to a kind of problem but from the vantage of a discussion about AGI I am wondering about a class of algorithms which would act as analytical algorithms. The other thing I was wondering about was whether an explicitly defined algorithm generator could have characteristics that could be ground to the parameters of the generator. Since I have shown that a class of generators would have emergent (and submergent) characteristics this would be a problem. However, that does not mean that it would never work. Jim Bromer On Sat, Feb 9, 2013 at 12:03 PM, Piaget Modeler <[email protected]>wrote: > An action could be considered a sub-program (or algorithm). > > We have such things as general purpose planers which can combine "actions" > (i.e. sub-programs). > > So we already have a "sub-program" generator in the form of general > purpose planners. > > In my vernacular sub-program = solution = plan = composite action. > > ~PM. > > > ------------------------------ > Date: Sat, 9 Feb 2013 08:22:44 -0500 > > Subject: Re: [agi] Could Algorithm Generators be a Feasible and Effective > AGI Method? > From: [email protected] > To: [email protected] > > Any effective AGI program could be considered to be a solution generator. > However, we do not know how to write a general solution generator for AGI > problems. If someone did he would have a working AGI program and we would > all be very impressed. Two exceptions to what I am saying. One is that we > can write a computer program that can solve some problems and the other is > that we really do not know much about writing algorithm generators either. > However, it is easier to write an algorithm generator which would generate > algorithms then it would be to write a solution generator which would write > solutions. > > A plan is not necessarily a solution. Plans can work as solutions as long > as they work. The wikipedia definition of an algorithm does include the > phrase, "effective procedure," so it would be a solution of some sort. > However, I do not accept this narrowed definition for a number of reasons. > One reason is that we could declare that the procedure does what it does so > it is an effective procedure to do whatever it does. This does create > problems for the definition but I think it is important to recognize that > the potential that a 'sub-program' might be an effective procedure for some > problem implies that a sub-program is a potential algorithm. So maybe I > should call it a sub-program generator in order to avoid these kinds of > disputes. Or maybe a potential-algorithm generator might be better. > > Even so, a plan is not necessarily a solution and effective procedure > algorithms are not solutions to any problem. So I really don't understand > what you are saying. > Jim > > On Sat, Feb 9, 2013 at 4:39 AM, Piaget Modeler > <[email protected]>wrote: > > For me a plan = a solution = an algorithm. > > Tell me why its' not this way with you. > > ~PM. > > ------------------------------ > Date: Fri, 8 Feb 2013 20:55:50 -0500 > > Subject: Re: [agi] Could Algorithm Generators be a Feasible and Effective > AGI Method? > From: [email protected] > To: [email protected] > > On Fri, Feb 8, 2013 at 11:56 AM, Piaget Modeler <[email protected] > > wrote: > > What kinds of data objects are analyzed and what does analysis imply? > If the algorithms are not solutions to gaps or impediments then what > purpose do they serve? > Kindly explain. > ~PM > > My guess is that I did not answer your question. In order to go from input > to 'understanding' an AGI program would have to do some analysis of the > input. The concept of analysis could be extended to an analysis of output > as well and that would be an important part of an AGI program. The > algorithms that I am talking about right now would not be 'solutions' per > se because they would be analytics. In other words I am emphasizing a part > of a greater process. The idea of 'algorithm generators' is what is > important in this thread. So while I talked about methods of reducing > complexity I was only talking about methods of reducing the complexity of > using 'algorithm generators' in an AGI program. I am thinking out loud so > to speak, trying to see if this is a good idea. And while this may seem > similar to automatic programming I am not explicitly talking about > automatic programming. The algorithms that would be generated and > preserved would be held because they would seem to be taking part in an > effective solution. However, I just do not see the idea of a AGI solution > generator as meaningful. Yes, an effective AGI program would be an > effective ;solution generator' but that sounds more like advertising than a > reference to a feasible artifact of computer science. An algorithm > generator is feasible. > Jim Bromer > > > On Fri, Feb 8, 2013 at 11:56 AM, Piaget Modeler <[email protected] > > wrote: > > What kinds of data objects are analyzed and what does analysis imply? > > If the algorithms are not solutions to gaps or impediments then what > purpose do they serve? > > Kindly explain. > > ~PM > > ------------------------------ > Date: Fri, 8 Feb 2013 11:33:10 -0500 > Subject: Re: [agi] Could Algorithm Generators be a Feasible and Effective > AGI Method? > From: [email protected] > To: [email protected] > > > On Fri, Feb 8, 2013 at 10:38 AM, Ben Goertzel <[email protected]> wrote: > > "Automated program learning" is a branch of AI that seems close to what > you have in mind... > This is what the MOSES component of OpenCog attempts to do, though it's > currently only really effective at learning simple sorts of programs... > -- Ben G > > > I described a very narrow type of programming object. An Algorithm > Generator. Because I was able to use such colloquial terms it is something > that almost everyone involved with programming should be able > to understand. Any AGI program would have to be capable of doing some > automated program learning. The question I was trying to explore was > whether or not an explicit system of algorithm generators would be useful > and how they might be used. One might argue that any AGI program that was > able to learn would effectively be creating (or generating) algorithms. > What I am talking about is the question of designing mechanisms that > explicitly generate kinds of algorithms. The algorithms that I have in mind > are not solution algorithms (per se) but analytical algorithms (including > algorithms that analyze data objects by making modifications). > Jim Bromer > > > *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > <https://www.listbox.com/member/archive/rss/303/10561250-470149cf> | > 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/303/19999924-5cfde295> | > 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/303/10561250-470149cf> | > 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/303/19999924-5cfde295> | > 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/303/10561250-470149cf> | > 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
