RE: [agi] Complexity of Evolving an AGI
But the different trials need not be independent --- we can save the trajectory of each AI's development continuously, and then restart a new branch of AI x at time y for any recorded AI x at any recorded time point y. Also, we can intentionally form composite AI's by taking portions of AI x's mind and portions of AI y's mind and fusing them together into a new AI z... So we don't need to follow a strict process of evolutionary trial and error, which may accelerate things considerably particularly if, as experimentation progresses, we are able to learn abstract theories about what makes some AI's smarter or stabler or friendlier than others. -- Ben G My guess is that the bottleneck would be the time required to test an AI. First, the AI to be tested needs to evolve a common sense based on its interaction with the environment. Then it has to learn natural language. Only after that can we test its cognitive abilities. The time it takes for each trial will severely limit the number of instances we can test. YKY Get advanced SPAM filtering on Webmail or POP Mail ... Get Lycos Mail! http://login.mail.lycos.com/r/referral?aid=27005 --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED] --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
RE: [agi] Dr. Turing, I presume?
Brad wrote: I see your point, but I'm not so sure you're correct. If you're devoting resources specifically to getting some attention, you may indeed speed up the process. I wish you luck. Well, I'm not devoting resources to getting widespread attention for AGI right now -- because the time is not right. The time to devote resources to getting widespread attention for AGI will be after the baby mind is done being engineered and we've started teaching it I'm a bit annoyed that it's taking so long to get to that stage, but such is life (and, more to the point, such is the progress of highly ambitious and complex science/engineering projects, particularly those carried out on a part-time basis... (though a number of us are working on Novamente-based projects full-time, the AGI aspects are being done part-time while a lot of focus is on short-term narrow-AI apps of the codebase that are able to generate revenue right now)) Getting widespread attention for AGI right now would probably be possible via a well-coordinated publicity effort -- but it would be foolish. The attention would not stick well enough because of excessive skepticism on the part of the conventional academic community, and because of the lack of a continuous stream of ongoing exciting results. The attention would likely fade before we get to the teaching-baby-mind phase... and then it would be more difficult to get the attention back. On the other hand, a publicity storm when the baby mind starts being taught will create attention that will stick far better -- because the criticism by conventional academics will be more muted (assuming the baby mind has been described in publications in the right journals, which is easy enough), and because the baby mind's continual intelligence improvements will an provide ongoing stream of novel fodder for the media. Attracting and sustaining media attention is not easy, but unlike creating AGI, it's a known science ;-) However even if you do get such attention, it will still take quite a while for the repercussions to percolate through society. Yes, of course... Mike seemed to be implying a technological rapture with very rapid changes at all levels of society. I think that people at all levels will be slow to react while a small percentage of early adopters who grab hold and start creating a market. This belief is based on historical precedent. Hmmm... well, I think that once the population at large becomes AGI-aware, then the collective mind of the first-world business and scientific community will start thinking of all sorts of AGI applications and working really hard to make them happen. And the speed of dissemination of AGI-awareness through society will depend a lot on the mode of dissemination. For example, suppose one launched an AGI in the context of a popular online multiplayer game, say the next Everquest (whatever that may be) Then a big sector of the population will get what the AGI is like very quickly. The game's popularity will grow because the AGI is involved with the game, and then a huge percentage of the teenage boys in the world will be highly AGI-savvy What if an AGI scientist with rudimentary English conversation skills makes a significant discovery?... and the AGI is interviewed on every popular TV talk show (together with its dubiously photogenic creator ;)? It doesn't even have to be a world-shattering discovery, just something moderately original and important, but conceived by a software system that can talk in rudimentary English about what it discovered and why. (Bear in mind that some kinds of scientific discovery will in a sense be easy for AGI's, compared to a lot of everyday tasks that seem easier to humans.) These are just two examples of how broad AGI awareness may be quickly raised -- there are many more... -- Ben G --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
RE: [agi] AGI's and emotions
Emotion is not sensory data but rather a product of it, from the machine point of view emotion is another reasoning faculty invoked from archetypal imprints adjusting to a sensory cognitive pattern (the resolution process). Emotion is the steering heuristic encapsulating the resolution domain. One may say its the seed of reason or at least the path it traces. Gus -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Ben Goertzel Sent: Wednesday, February 25, 2004 11:25 AM To: [EMAIL PROTECTED] Subject: RE: [agi] AGI's and emotions Mike, Regarding your definition of emotion. Ialmost agree with what you say -- BUT, I think you're missing a basic point. Emotions do involve data coming into the cognitive centers, vaguely similarly to how perceptual data comes into the cognitive centers. And, as with perception, emotions involve processing that goes on in areas of the brain that are mostly opaque to the cognitive centers. But in the case of emotion, the data comes in from a broadly distributed set of physiological and kinesthetic indicators -- AND from parts of the brain that are concerned with reaction to stimuli and goal-achievement rather than just perceiving. This is qualitatively different than data feeding in from sensors Emotions are more similar to unconscious reflex actions than to sensation per se -- but they last longer and are more broadly-based than simple reflex actions... ben g -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]On Behalf Of deering Sent: Wednesday, February 25, 2004 2:19 AM To: [EMAIL PROTECTED] Subject: Re: [agi] AGI's and emotions Bill, I agree with you that emotions are tied to motivation of behavior in humans. Humans prefer the experience of some emotions and avoid the experience of others, and therefore generate their behavior to maximize these goals. I think this is a peculiarly biological situation and need now be replicated in AI's. I think in AI's we have the design option to base the motivation of behavior on more rational grounds. Ben, I don't know if my personal definition of emotions will be of much help as it may not be shared by a very large community. but for what it's worth, here it is. MIKE DEERING'S PERSONAL DEFINITION OF EMOTIONS: Emotions are a kind of sensory data. The sensory organ that perceives this data is the conscious mind alone. The physical reality which generates this raw data are selected concentrations of neurotransmitters in the brain. Their effects vary with different types of neurons in different locations. Some types of neurons produce more of certain kinds of neurotransmitter than other types of neurons. Those that generate the neurotransmitters are not necessarily the same as those that are more affected. They are also affected by other chemicalsproduced by glands. It's complicated. These neurochemical phenomena are by evolutionary design causally linked to environmental circumstances and divided into positive and negative type. They are used, by evolutionary design, to positively and negatively reinforce behaviors to maximize and minimize the related circumstances. Emotions are not products of cognitive processes but are rather perceptions of neurochemical states and states of activation of selected regions of the brain. Because of the complicated feedback arrangements in the generation of neurotransmitters and hormones, and the neurons role in this feedback, some limited conscious influence can be exercised in the management of emotions. Emotions can be generated artificially by the introduction of various chemicals to the brain, the direct electrical stimulation of certain neuron clusters, or direct control of environmental circumstances. Certain physical bodily sensations are closely related to emotions: pain to sadness, pleasure to happiness. To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED] To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED] To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]