-----Original Message----- From: firstname.lastname@example.org [mailto:email@example.com] On Behalf Of meekerdb Sent: Friday, October 25, 2013 10:46 AM To: firstname.lastname@example.org Subject: Re: Douglas Hofstadter Article
On 10/25/2013 3:24 AM, Telmo Menezes wrote: > My high-level objection is very simple: chess was an excuse to pursue > AI. In an era of much lower computational power, people figured that > for a computer to beat a GM at chess, some meaningful AI would have to > be developed along the way. I don' thing that Deep Blue is what they > had in mind. IBM cheated in a way. I do think that Deep Blue is an > accomplishment, but not_the_ accomplishment we hoped for. >> Tree search and alpha-beta pruning have very general application so I have no doubt they are among the many techniques that human brains use. Also having a very extensive 'book' memory is something humans use. But the memorized games and position evaluation are both very specific to chess and are hard to duplicate in general problem solving. So I think chess programs did contribute a little to AI. The Mars Rover probably uses decision tree searches sometimes. Agreed. Some manner (e.g. algorithm) of pruning the uninteresting branches -- as they are discovered -- from dynamic sets of interest is fundamental in order to achieve scalability. Without being able to throw stuff out as stuff comes in -- via the senses (and meta interactions with the internal state of mind -- such as memories) -- an being will rather quickly gum up in information overload and memory exhaustion. Without pruning; growth grows geometrically out of control. There is pretty good evidence -- from what I have read about current neural science -- that the brain is indeed, throwing away a large portion of raw sensory data during the process of reifying these streams into the smooth internal construct or model of reality that we in fact experience. In other words our model -- what we "see", what we "hear", "taste", "smell", "feel", "orient" [a distinct inner ear organ] (and perhaps other senses -- such as the sense of the directional flow of time perhaps as well)... in any case this construct, which is what we perceive as real contains (and is constructed from) only a fraction of the original stream of raw sensorial data. In fact in some cases the brain can be tricked into "editing" actual real sense supplied visual reality for example literally out of the picture -- as has experimentally been demonstrated. We do not experience the real world; we experience the model of it, our brains have supplied us with, and that model, while in most cases is pretty well reflective of actual sensorial streams, it crucially depends on the mind's internal state and its pre-conscious operations... on all the pruning and editing that is going on in the buffer zone between when the brain begins working on our in-coming reality perception stream and when we -- the observer -- self-perceive our current stream of being. It also seems clear that the brain is pruning as well by drilling down and focusing in on very specific and micro-structure oriented tasks such as visual edge detection (which is a critical part of interpreting visual data) for example. If some dynamic neural micro-structure decides it has recognizes a visual edge, in this example, it probably fires some synchronized signal as expeditiously as it can, up the chain of dynamically forming and inter-acting neural-decision-nets, grabbing the next bucket in an endless stream needing immediate attention. I would argue that nervous systems that were not adept at throwing stuff out as soon as its information value decayed, long ago became a part of the food supply of long ago ancestor life forms with nervous systems that were better at throwing stuff out, as soon as it was no longer needed. I would argue there is a clear evolutionary pressure for optimizing environmental response through efficient (yet also high fidelity) pruning algorithms in order to be able to maximize neural efficiency and speed up sense perception (the reification that we perceive unfolding before us) This is also a factor in speed of operation, and in survival a fast brain is almost always better than a slow brain; slow brains lead to short lives. But not just pruning, selective & very rapid signal amplification is the flip side of pruning -- and this is also very much going on as well. For example the sudden shadow flickering on the edge of the visual field that for some reason, leaps front and center into the fore of conscious focus, as adrenalin pumps... sudden, snapping to the fore. And all this, from just a small peripheral flicker that the brain decided on some local sentinel algorithm level was in some manner out of place.... maybe because there was also a sound, directionally oriented in the same orientation. Clearly the brain is able to suddenly amplify a signal -- and also critically at any step along the way to the final synthesis of the disparate sense signals into a cohesive picture -- and jam it right up to the executive level, promoting it up the brain's attention chain much more rapidly and prominently than is normally the case. The survival benefits of this kind of alarming circuitry as well as pruning is clear, and our brains circuitry is the outcome of billions of years of selective pressure and I am fairly certain that both signal suppression (pruning) as well as signal amplification are operating at all scales. > > I believe there will be an AI renaissance and I hope to be alive to > witness it. You may be disappointed, or even dismayed. I don't think there's much reason to expect or even want to create human-like AI. That's like the old idea of achieving flight by attaching wings to people and make them like birds. Airplanes don't fly like birds. It may turn out that "real" AI, intelligence that far exceeds human capabilities, will be more like Deep Blue than Kasparov. Brent Brent -- I tend to agree with you here as well, much as it would be flattering to us if super-AI was like us, but well, just better... there is no guarantee that the actual outcome of a self-emergent process that generates a self-perpetuating AI will have much resemblance to us on an emotional/empathetic level. The prime driver for the evolution of AI is currently and has for a long time been for military applications. This is where the big money is. If it becomes a Darwinian process and the evolutionary pressure is to develop effective & increasingly autonomous killing machines then the kind of AI that I am guessing eventually emerges out from these selective pressures could potentially behave in an exceedingly unpleasant & deadly manner towards humans and in fact may not like us at all. -Chris > But for this renaissance to take place, I think two cultural shifts > have to happen: > > - A disinterest with the "science as the new religion" stance, leading > to a truly scientific detachment from findings. Currently, everything > that touches the creation of intelligence is ideologically loaded from > all sides of the discussion. This taints honest scientific inquiry; > > - New economic structures that allow humanity to pursue complex goals > outside the narrow short-term focus on profit of corporatism or the > pointless status wars of academia. > > Best, > Telmo. -- You received this message because you are subscribed to the Google Groups "Everything List" group. To unsubscribe from this group and stop receiving emails from it, send an email to everything-list+unsubscr...@googlegroups.com. To post to this group, send email to email@example.com. Visit this group at http://groups.google.com/group/everything-list. For more options, visit https://groups.google.com/groups/opt_out. -- You received this message because you are subscribed to the Google Groups "Everything List" group. To unsubscribe from this group and stop receiving emails from it, send an email to everything-list+unsubscr...@googlegroups.com. To post to this group, send email to firstname.lastname@example.org. Visit this group at http://groups.google.com/group/everything-list. For more options, visit https://groups.google.com/groups/opt_out.