Hi Colin You seem to be following a similar process to AI as to what was used to develop the first, nuclear bomb - various approaches were used coupled with great experimentation.
Semantically, your inclusion of the term "emergent" in your last message undersores this approach for me. I'd like to dwell on its relevance for a few seconds. Emergence is regarded as the basis for complex-systems engineering (Checkland). Further, Checkland asserted how the debate between complex and simple systems would probably give rise to what is regarded as systems thinking. This is ancient stuff I'm repeating only to stress the importance of its credibility. Thus, on the theoretically basis alone, your experimental approach could be deemed to be sound. Narrow AI, broad AI, AGI? All peas in the same pod of complex-systems thinking. The fundamentals still have no significant incentive to change. Personally, I would value such an experimental approach on the basis of rethinking the whole idea of developing AI. How else was the sound-barrier broken? In addition, if one followed the emerging trend in recent, adaptively-autonomous technologies, one would be hard pressed to write off your approach. Just one theoretically-moot point if I may, albeit a semantic one? Any institutionalised process effectively is a program of code. As an extension, any reduced process - as a procedural implementation - on a computer would become a computerized program. Hence, I suppose, your search for a generic algorithmic platform. In the sense of systemically, as soon as you'd link the "stochastic" environment to a computer chip in any way it should emerge as a form of computer program. Whilst one understands the need for research to be highly focussed on its objectives, one must still have a design framework that would not unduly restrict any design in a short-sighted Heisenbergian-Einstein debate. I would assume then that you do have a quantum-based design framework you're working from. If not though, this particular, organic approach would sooner or later come up against the eco-systemic realities of highly-abstracted implementation. This then, mainly due to the lack of navigational competency in the R&D framework to consistently and reliably perform adaptive integration. If it cannot be measured somehow, it cannot be reliably tested and I'm by no means suggesting this to be the case with your experiment. Mine are just thoughts on the interesting topic at hand. One day, when bootstrapping does occur, you'd be wanting to debug though. If only purely mathematical, then purely computational? Maybe that was how computer science emerged. Good luck with the experiment. Rob To: [email protected] From: [email protected] Subject: RE: [agi] Re: Starting to Define Algorithms that are More Powerfulthan Narrow AI Date: Wed, 6 May 2015 10:01:33 +1000 Hi, Rather busy... Having trouble devoting time here. Jim.... You ask if I am making some kind of electric circuit. Basically yes. Except it's physical instantiation is important. Materials in space. I know you won't get why that is important. That's ok for now. Just accept that it's like that for the same reason the brain is like that. What it isn't is an 'Equivalent circuit' in the traditional sense of voltage/current replication. It is designed to produce functionally equivalent action potential-style signaling AND the brain-style field system that actually expresses the voltages. The hardware will (in the field version) express an EEG and MEG like brains. Having said that I am currently designing a version that doesn't express the fields but allows their addition later...knowing what performance degradation results (it will be narrow-AI not AGI). Call it a causality mirror with a faked image in it. It is deeply self modifying. The circuits literally rewire themselves. Circuit loops duplicate/diverge and switch out/off. It accounts for the process of brain development as a kind of learning. I.e. I don't even have to design the 'brain'. It will self configure based on being in the world. Because it's not using neurons it won't automatically mimic brains in structure. I have no idea what a brain will look like. Physically its a crystalline rock. No actual material growth. Functionally it will stabilize in ways I can't know except by experiment. It means that it must be permanently juvenile.. Overexpressed neurons and overexpressed synapses culled back. Lots of wastage. But so what? Not one line of software anywhere. Any 'algorithm' it has is in the adaptation mechanisms. But they are in hardware. The state of the chip's self configuration is the only actual data involved. Yet, when you look at it there will be deep regularities in its behaviour. You could write them down. However they are all emergent. You know what the hardest part of this is? ... Giving it goals. A reason to bother. A reason for it to sustain the quasi-stable resonances that signify its functioning. I have to think of something akin to homeostasis to keep it going! ROBEOSTASIS. You know what might happen? It possibly self-sustain without human intervention or some kind of hardwiring until the fields are added. Unsure. Answering that is an experimental goal. Steve seems to be deeply inside homeostatic concerns. So that's good. I'm not here to justify anything. Experimental proof will speak for me. And if I can't get the version with and without the fields to be different in predicted ways then I will grovel at the feet of the great god computationalism. Not before.😊 I think the approach is a reversion to 'natural cybernetics' that had a brief life in the 1950s and then was lost in a tsunami called computer science. I bring it back for an upgrade. Notice that AGI failure started the moment cybernetics stopped. The actual science of artificial intelligence stopped then, too...IMO. Enough poking the bear. Gotta get back to it. I really appreciate the interest in this 'adaptive control' approach. Cheers Colin From: Jim Bromer Sent: ‎4/‎05/‎2015 12:42 AM To: AGI Subject: Re: [agi] Re: Starting to Define Algorithms that are More Powerfulthan Narrow AI I thought the ideas are interesting and Colin's description was more readable than usual but the arguments supporting the method weren't very powerful. I am curious about how Colin is implementing the method. Could you give me a little more about that? Are you designing some kind of electrical circuit? What I was trying to say in this thread is that you have to supply a little more insight about why you think that the methods that you are designing and will be implementing would rise above being 'narrow ai'. For instance, Colin's honest report on how far he has actually gotten so far sounds like it is on par with simple narrow AI. As I reread your messages I keep finding a little more in it. But back to my point. Since I can rough out the algorithms that I would use as if they were abstractions, or as if they could exist within an abstract world, it would seem that I should be able to conduct simple tests to show that they could diversify in some way that is: 1. at least better than narrow ai, and 2. useful in some way. So perhaps I should add that. I would say, for example, that artificial neural networks would pass this kind of test. However, the criticism then is, ironically given our use of the narrow ai term, that they lack efficient means to focus and they cannot be efficiently used as componential objects. So, can you guys define some abstract or simple tests that could show that your ideas would become able to adapt to the more complicated demands of actual tests? The value of the simple test is that once you can get your algorithms to pass the first test you might come up with ways to design a slightly more aggressive test. So if I could test my ideas to,say, try to learn to recognize some simple classifications then I might try to see if I can get it to try to get it to learn to utilize systems of classifications effectively and efficiently (without redesigning the program only for that specific kind of test.) So then I would have to design some other kind of test to make sure that it is somewhat general. Jim Bromer On Sun, May 3, 2015 at 3:25 AM, Colin Hales <[email protected]> wrote: > > >> On Sat, May 2, 2015 at 2:50 AM, Steve Richfield <[email protected]> >> wrote: >>> >>> Jim, >>> >>> Again, I think I see the POV to solve this. All animals, from single cells >>> to us, are fundamentally adaptive process control systems. We use our >>> intelligence to live better and more reliably, procreate, etc., much as >>> single-celled animals, only with MUCH richer functionality. Everything fits >>> this hierarchy of function leading to intelligence. >>> >>> Then, people like those on this forum start by ignoring this and trying to >>> create intelligence from whole cloth. This may be possible, but there is NO >>> existence proof for this, no data to guide the effort, etc. In short, there >>> is NO reason to expect a whole-cloth approach to work anytime during the >>> next century (or two). >>> >>> However, some of the mathematics of adaptive process control is known, and >>> I suspect the rest wouldn't be all that tough - if only SOMEONE were >>> working on it. > > > Erm.... guys. This would be me. > > I am working on it. For well over a decade now. Cognition and intelligence is > implemented as an adaptive control system replicating, inorganically, the > natural original called the human (mammal) nervous system. I simply replicate > it inorganically. Tough job but I am getting there. There's no programming. > No software. Just radically adaptively nested looping processes. In control > strategy terms it is a non-stationary system (architecture itself is > adaptive). Control loops come into existence and bifurcate and vanish > adaptively. The architecture commences at the level of single ion channels > and nest at multiple levels that then appear in tissue as neurons doing what > they do, but need not appear like this in the inorganic version. You don't > actually need cells at all. These then nest at increasing spatiotemporal > scales forming coalitions, layers, columns and finally whole tissue. All > inorganically. All the same at all scales from an adaptive control > perspective. Power-law scalable. Physically and logically. > > In my case, for the conscious version the hardware includes the > field-superposing, active additional feedback in the wave mechanics of the EM > field system produced by brain cells at specific points. The fields form an > addition/secondary loop modulation that operates orthogonally, > outside/through the space occupied by the chip substrate. > > What I am starting with is the 'zombie' or symbolically ungrounded version. > It doesn't produce the active field system (missing a whole control system > feedback mechanism) and uses supervised learning (externalised by a conscious > human trainer) to compensate for the loss of the natural role consciousness > has as an endogenous supervisor. It will, in the zombie form, underperform in > precisely the way all computer AGI underperforms. This is what is missing > when you use computers to do it all. You end up with a recipe (software) for > pulling Pinocchio's strings. Whereas my system bypasses the puppetry > altogether. It makes the little boy, not the puppet. > > However you view it, there's nothing else there in a brain except nested > loops that have power-law responses in two orthogonal axes: sensory and > cognitive. Adding the field system to the sensory axis (e.g. visual > experience) or part of the cognitive axis (e.g. emotional experience) provide > the active role for consciousness implemented through the causal impact of > the Lorentz force within the hardware. I suppose it'd be an 'adaptive control > loop' philosophy for cognition and 'EM field theory of consciousness' > combined. No computing needed whatever. Just like the brain. Most of the last > ten years has been spent figuring out the EM field bits! That I am now > omitting, knowing what I lose when I do that (i.e. consciousness). > > Teeny weeny Zombie version 0.0 this year I hope. No EM field generation. I > call it the 'circular causality controller'. I aim to add the EM fields > later. That part requires $millions. It's chip-foundry stuff. > > So chalk me in under this 'adaptive control loop' category for AGI > implementation please. I know this forum is a 'using computers to do AGI' > forum so I'll just continue to zip it. I haven't mentioned it much over the > years because it seems that most of you aren't interested in my approach. For > reference and for the record.... I am the 'AGI as adaptive control' guy. > > cheers > colin > >>> >>> >>> I suspect that when the answers are known, it will be a bit like spread >>> spectrum communications, where there is a payoff for complexity, but where >>> ultimately there is a substitute for designed-in complexity, e.g. like the >>> pseudo-random operation of spread spectrum systems. Genetics seems to >>> prefer designed-in complexity (like our brains) but there is NO need for >>> computers to have such limitations. >>> >>> Whatever path you take, you must "see a path" to have ANY chance of >>> succeeding. You must have a POV that helps you to "cut the crap" in pursuit >>> of your goal. Others here are working on whole-cloth approaches, yet >>> bristle when challenged for lacking a guiding POV. I see some hope in >>> adaptive control math. Perhaps you see something else, but it MUST have an >>> associated guiding POV for you to have any hope of succeeding - more than a >>> simple list of what it does NOT have. >>> >>> Steve ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/11721311-f886df0a Modify Your Subscription: https://www.listbox.com/member/?&; Powered by Listbox: http://www.listbox.com 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
