I'm posting the article for completeness: Friday, December 9, 2011 Embodiment is just coordinate spaces, interactivity and modalities - not a mystery<http://artificial-mind.blogspot.com/2011/12/embodiment-is-just-coordinate-spaces.html>
"The debate" - should an AGI be embodied or not. I participated in a discussion on sensorimotor hierarchies a few months ago in the AGI list, but I'm lazy to digest it here right now (yet). I'll reword one POV of the essence which I thought of recently, though. "Body" and "embodiment" is actually about: - *Known fixed basic continuous coordinate spaces *- that simplifies search and basic operations and sets one of the basic components of the inputs and their initial "labels" - coordinates. The simplest use of this is that mind can recognize where the activity is and relate coordinates of inputs to each other. -* Interactivity* - input and output, includes changes of coordinates as smooth as possible in time-space - allows for as general as possible inputs and outputs (actions). - *Modalities* - it's about dealing with different types of lowest level (for the system's sensors) input from the environment, with the "physics" - for constructive reasons and more general input. Each system eventually should deal with some kind of lowest level input and output, the most general it is, the more general intelligence would be. *- Inter-modalities - *to relate and find models between them. *- Value-free initial inputs *- initially sensory inputs and motor outputs don't have meanings, except when related to: *- Basic reward-system - *that's in order to keep system integrity in the "physical" world where it exists and to avoid self-destruction. -- *--- Todor "Tosh" Arnaudov ---* * -- Twenkid Research:* http://research.twenkid.com -- *Self-Improving General Intelligence Conference*: http://artificial-mind.blogspot.com/2012/07/news-sigi-2012-1-first-sigi-agi.html *-- Todor Arnaudov's Researches Blog**: *http://artificial-mind.blogspot.com On Tue, Aug 21, 2012 at 12:53 PM, Todor Arnaudov <[email protected]> wrote: > > Hi all, > > -- That's not an AGI list, it's a "*court of justice"*, you are all > "lawyers", both the "prosecution" and the "defense", and everybody is > against everybody (mostly). What you talk about is most of the > time rhetorical and cheap superficial logic with entangled expressions, or > obvious stuff presented as ground-breaking. > > -- One good reason for the court-style is that email list and "blind" > non-real time text messaging is a SICK and wrong medium for friendly > communication, especially for creating bonds and respect. It starts from > high levels of abstraction in brains, emotional paths are apparently > strongly cut-off, since forums and email lists are a common place for > "trolling" and "lawyer-like" discussions. Real time - sensory experiences > kept in the short-term memory, and maybe related to the > oxytocine-generation processes during interaction with "animate beings" - > and low level inputs (sound/voice/images) is much more constructive. > > Would you personally insult somebody live? Yes, if you're sure he won't > kick your ugly face with a mavashigeri after you did, in the worst case > she''ll just turn her ASS to your face and you'll be ignored, so you would > be careful about this. > > -- Jim, it's nice that you realized the "unary" system and positional > systems as compression, I've done too some time ago, I call it "incremental > arithmetic" . However in practice the unary system is not really unary at > the low level, I'll discuss this elsewhere, though. > > -- While there are apparently schools of thought and at high level (as > discussed on this list) the claims of many people are THE SAME or are very > easily adjustable, most of you keep acting like you're the only one in your > "island", you're even ignorant and arrogant for the rest; any of you seem > to want really to collaborate; yep, the email list is not the right way - > it's conferences, meetings, business relationships and: see below. > > >One effect of conceptual relativism is that when you use concepts to > consider other concepts the concepts you use will affect the concept under > consideration. This is simple to understand and yet I don't >remember > anyone actually talking about it to me. It is one of those things that > people either ignore or don't understand or don't care about. (...) > >One thing that Boris and I seem to agree with is that you have to be able > to refer to the source of a concept (or information) in order to resolve > some issues related to data derived from it. (Since we need >to use > generalizations then you would have to refer to the simplest generalization > of the source, or an elemental source event that characterized the class of > the generalization of the concept in order to resolve >some issues > > -- Jim, relativism is so obvious ("ambiguity", "grounding", "basic > assumptions", "correct/wrong predictions"), it's as basic as that you need > to have a correct model in order to produce "correct" results given that > model. > It's interesting to whom you've explained it, if they didn't get it. > Indeed - I read your message when you say that "Nobody really knows how the > brain works". I'd ask you: > > *What "How" means? How how? And what does it mean "to really know"?* > > One obvious point is that a brain couldn't simulate itself at the highest > resolution by its own resources, it can only in a fake "mathematical" case > - it's running itself/~"simulating" itself. > > Therefore a brain - "anybody" - could never "really understand how it > works" in all details, without aids and without focusing on specific > aspects etc., but as I've tried to explain starting a decade ago, that's > true even for the simplest computer. People "don't really understand" how > any computer work at its low levels, they can't keep in mind all or even > just a few of the processes and the dynamical data alterations running, > only small fragments at a time - still we do build computers, and people > claim that they do understand computers - "logical operations", > "flip-flops", ... - generalizations, very low level and a tiny portion of > the real system. > > ... > > Regarding the aspect you and Boris "agree on", I myself has shared too > much about the hierarchical generalization and why it should start from the > lowest level on this list, but of course nobody can understand this obvious > stuff (besides some from the SoT). > > Starting from the lowest level is as simple as that it allows highest > possible resolution of perception and control in the environment (lowest > input-output), and that this is the actual data you get and output - > sensory matrices with "meaningless data" and coordinate adjusting motions > with initially "meaningless" trajectories, to which meaning is imposed and > "implanted" by generalizing the environment and driving the trajectories > from up-to-bottom. Another elementary reason is that the lowest levels work > without fully developed higher (a baby), high level without the lower is a > non-sense, if you connect the high level to the sensory input, it will turn > into low level. > > It's also obvious that humans DO use lower level data to communicate and > disambiguate, there's no other way round - again, if natural language is > used to express too low level details, it transforms in an ugly code for > expressing specific data, for example images can be expressed a pixel by > pixel in ASCII code with numbers written by NL "two hundred and eight for > red, sixty seven" etc. > > I tried to explain this to Mike regarding his confused wrong examples -*when > natural language is too clumsy and ambiguous, more specific (of higher > resolution) means of communication are used - from syntactic trees > (unveiling some of the syntactic ambiguities), to many examples from the > same class, to diagrams, maps, sound records, images, stereo images, > 3D-models in a modelling environment (so you can apply output to them, can > change their coordinates and rotation towards you), animation, video, > pointers on them and all at-once, inter-modal, making for the mind even > easier the job of finding the right/wanted correlations.* > > Keep in mind the above, and I'll continue regarding the brain: > > *In fact people DO KNOW how brain works*, every record of brain activity > (incl. any aspect and any moment of the observable human behavior) is a > by-effect of how brain works, and it displays how it does. There are > different levels of abstraction, detail, spatio-temporal area, domain, POV > in which we PERFECTLY understand how brain works - I'll join Adam and tell > you to study neuroscience, you'll see how much is known. > > Of course, another "relativistic" question here is what "to understand" > means, and a low level meaning is to be able to model it/to predict how it > will behave given particular circumstances. We DO understand how brain > works in this sense, but it depends on the spatio-temporal resolution of > causality and control for specific aspects and parts of the brain. > > ... > > -- Boris keeps repeating stuff which is like the a-b-c from the *school of > thought* he is in, and it's obvious and elementary for ones who do > understand it, still he talks like if he was saying something > ground-breaking or "substantial", he himself has published it a decade ago > or so, other researchers - too. > > The substantial things are not to be expressed in a few lines of text - > too abstract, too much "conceptual relativism", - still many of the > participants here can't manage emails of a length like of this one, and > exchange "SMS" or "tweets" here, which are hanging in no-context worlds. > > IMHO, substantial discussion requires: > > *-- Models, code, demos, samples, mind-developmental diagrams, tests and > results etc. and technical discussions on the *numbers* and those effects. > * > * > * > *That's the way.* > > Most of the stuff here is natural language blah-blah-ing. > > -- And again, at the level of blah-blahing on this list, what Boris or > Sergio, or Adam or whatever says is more than obvious and basic for ones > who're familiar with the school of thought/neuroscience sub-fields etc, > I've written about it a decade ago. However it's still substantial for the > ones who don't understand it yet. Adam says substantial things about the > embodiment, which many non-sensori-motor-neuroscience guys are probably not > aware, another wording of this by myself: Embodiment is just coordinate > spaces, interactivity and modalities - not a > mystery<http://artificial-mind.blogspot.com/2011/12/embodiment-is-just-coordinate-spaces.html> > > http://artificial-mind.blogspot.com/2011/12/embodiment-is-just-coordinate-spaces.html > > ------------------------------------------- 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
