Ben, Jim doesn’t get it at all – and I would imagine that you can see and acknowledge that – he treats every creative problem as a logical exercise by working backwards from the solution, when, by definition, you begin a creative problem not knowing how to solve it.
However, you don’t “really” get it. You’re trying to move towards creativity, (he isn’t at all), but still moving towards it also from a somewhat unnatural position. Actually creativity is simple. To be creative, you simply have to put ANY new objects together - – like an infant and just two or three will do. Think of yourself/your AGI as an infant in a playpen – you are likely to be presented with a potentially infinite diversity of objects – bricks, cushions, boxes, cars, rattles, glasses. sausages A creative infant can form ANY of those into some kind of tower. That’s creativity. You don’t need to think in terms of something amazing like car-man or even conceptual blending – creativiity is simply taking one or more new actions on one or more new objects – handling one or more new objects in one or more new ways – actions that have to be improvised for the first time, not reproduced for the second to nth time from a pre-existing repertoire. By definition and in reality, no previous tower/combination-of-“buiilding-blocks” will prepare you for the next combination. If you’ve already put a toy block on another toy block, it won’t fully prepare you for now putting a car on a toy block, or a card on top of two cards – each new building task is somewhat different and not-wholly-derivable from the last. A creative task need not necessarily be *difficult* at all, but it is always different. You always have to move your hand and arm along somewhat new lines to handle this somewhat new form of object. Ditto creativity is being able to walk on ANY terrain.- grass, rocky, stepped, circular tunnel etc – being able to come up with whatever kind of steps are necessary to locomote across different terrains – or, if you’re a big-wheeled robot rover, configure your wheels in whatever way necessary to locomote across any terrain – again, move your effectors along somewhat new lines. “That’s all” that’s required for creativity – improvise a new combination of actions on a new combination of objects that are not formulaically (or logically) the same as past actions.(That’s what humans do continuously throughout their lives). The simplest form of creativity arguably, from a formal point of view, is building rock walls – because every rock and rock wall is different (and yet at the same time very similar). So that;s all you have to do to be an AGI. You must show how your machine can handle and combine ANY new objects presented to it (like an infant) - new objects that it will never be able to predict in advance.. You’re still not approaching it properly – even though you’re getting warmer. You have to show me that your AGI can build ANY structure with ANY materials presented to it – as human infants can. Now there is no way that Opencog or any algorithmic approach can do that. That’s like saying, there’s a recipe for ANY and EVERY dish – for combining whatever new ingredients a chef may be presented with. There are only algos/recipes for specific, known extremely limited sets of dishes. and ingredients. Creativity and AGI are about producing new structures of actions and objects for the very first time. Algorithms are routines for producing old structures for the second to nth time Creativity and AGI by extension are about building forever – building ever new kinds of structures with ever new kinds of objects. Narrow AI and algos are about building only one kind of structure. with one set of objects (and even then having difficulties at times with the “complexities” of all the possible ramifications). Junk Opencog, and you’ll find that AGI is *relatively* easy (!). – because, unbeknownst to Jim and perhaps to you, robots *don’t* have to be *algorithmically* programmed – they can be programmed to be creative and do new things, not just the same old things – move along new lines, not just the same old lines.. . From: Ben Goertzel Sent: Sunday, May 12, 2013 6:09 AM To: AGI Subject: Re: [agi] A General O.D. (Operational Definition) for all AGI projects Mike etc., In http://wiki.opencog.org/w/CogPrime_Overview#A_CogPrime_Thought_Experiment:_Build_Me_Something_I_Haven.E2.80.99t_Seen_Before I give a narrative of how a completed OpenCog system with a roughly humanoid robotic embodiment might respond to a query to "Build me something I haven't seen before" by building a car-man, displaying a combination of car properties and man properties This is a somewhat watered-down extract from a late chapter in the forthcoming (currently the text is done and I'm just fixing Latex errors) book Building Better Minds... Of course, I could explain how OpenCog would do ball-box in a similar manner to car-man, but I'm on vacation this week and will be busy when I get home, so I guess I will leave that as an exercise for the reader ;p So I have two quick reations to your "ballbox" challenge... 1) It is untrue that we AGI researchers have not thought about this sort of issue... my link above is evidence otherwise.... Concept blending is a mainstream topic in cog sci and there was a paper on it at the last AGI conference, actually... 2) The fact that you personally find our plans for achieving grounded/embodied concept blending, to be unlikely to work, doesn't particularly prove anything except that you have a different intuition about intelligence and its implementation than everyone in the AGI research community..... So what? ;) ben g On Sat, May 11, 2013 at 4:25 PM, Mike Tintner <[email protected]> wrote: What we’ve just seen with Jim is yet another example here of effective creative illiteracy –s.o. talking about their “creative” AGI project without any attempt at defining either its O.D. (the effect to be achieved) or an effective mechanism - without, to put that extremely crudely in common parlance, having any “idea”. To repeat, this is appalling – it’s simply non-creative and a waste of space. So to take further steps to eradicate this disease, let me put forward a general O.D. for A.G.I projects (and to some extent all culturally creative projects). Your project must have an E.M. for how BALL + BOX = BALLBOX i.e. you have to show how with only standard knowledge of two objects, balls & boxes, you can a) generate and/or b) understand a new, third object, “ball-box” that is derived from them by non-standard means. In this case, a BALLBOX is a box shaped like a ball rather than a cube. (That is a more concrete way of saying: “you must be able to show how your project can think outside the box”). Another example would be, you must have an E.M. for how CAT + DOG = CATOTAUR a creature, similar to a minotaur, half cat, half dog. Or, you must have an E.M. for how 2 + 2 = 5 again, your machine must from knowledge of standard maths be able to produce non-standard maths. (And for the benefit of “same old, same old” Matt, that does not mean googling an existing 2+2=5 “proof” – you have to generate/understand an altogether new one). Or, you must be able to show how 2 rocks + 2 rocks = 4 rocks. You must be able to show how having knowledge of how two rocks are laid, you can lay another two rocks on top of them. Laying rock walls is not a math operation like laying brick walls – each rock is individually formed and needs to be laid individually. (Or your system must just be able to recognize/conceptualise ROCK, since all rock forms are individual and non-standard). In all of these cases, you combine two objects to produce a third object that has never, to your knowledge, been derived from them before. You do “magic” – you put a rabbit in an empty hat and pull out a bird. You put a penis in a vagina and pull out a baby (“where did that come from?”) It’s a waste of time to even ask Jim to produce an O.D., but anyone serious about AGI will want to produce an O.D. with an E.M. - an explanation of a BALL BOX. AGI | Archives | Modify Your Subscription -- Ben Goertzel, PhD http://goertzel.org "My humanity is a constant self-overcoming" -- Friedrich Nietzsche AGI | Archives | Modify Your Subscription ------------------------------------------- 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
