That's the reliable path to build AGI and conscious machines https://www. researchgate.net/project/Build-Conscious-Machines We will be greatly appreciate your constructive contribution
On Sat, Feb 18, 2017 at 8:15 AM, Jim Bromer <[email protected]> wrote: > An example of a substantive reason to support some AI theories is that > many programs have the ability to make some 'prediction' based on the > reaction to > features of input. Working from there, the ability to make predictions > means that knowledge or capabilities which had been integrated could > be used meaningfully. My counter argument is that 'prediction' and > 'predictive utility' as used in contemporary AI is just as noisy and > lossy as any other would be AGI facility. So although this kind of > ability may be necessary for AGI it (along with feedback integration > differentiation exclusion and a host of other contemporary AI > methods), it is far from sufficient. And I acknowledge that reasonable > expectation that progress in AGI is not going to be instantaneous or > smooth. > Jim Bromer > > > On Sat, Feb 18, 2017 at 10:08 AM, Jim Bromer <[email protected]> wrote: > > I will look at some of the links when I get some time, but you clearly > > did not understand my criticism of the theory. Even if I did > > misunderstand something from your presentation, so what? It does not > > necessarily mean the criticism is irrelevant. An example of a > > substantive reason to support some AI theories is that many programs > > have the ability to make some 'prediction' based on the reaction to > > features of input. If you had used that argument I would have pointed > > out that the ability of AI to make predictions (or expectations) is > > wiped out by extraneous features (noise) and by the complexity of any > > AI program to find which features would be relevant in different > > situations (lossy 'insights'). I am not saying this is relevant to > > what you are saying I am just trying to give you an example of how > > reasoning can be used to support a theory and how some reasoning which > > is substantial in theory may not be that strong in practice. > > Jim Bromer > > > > > > On Sat, Feb 18, 2017 at 8:03 AM, Logan Streondj <[email protected]> > wrote: > >> -----BEGIN PGP SIGNED MESSAGE----- > >> Hash: SHA256 > >> > >> > >> > >> On 2017-02-17 10:03 PM, Jim Bromer wrote: > >>> Sorry. OK, you said the meditating was high phi. But your response > >>> ignored (and was a distraction from) the point that I made that > >>> it would be possible to create computer programs that were capable > >>> of integration and differentiation (and therefore were capable of > >>> learned exclusion) which were not capable of anything resembling > >>> true intelligence. > >> > >> Do you have examples? > >> > >>> And I explicitly included the possibility that improvements in AGI > >>> might be slow and uneven. > >> > >> You just said there wasn't even a glimmer. > >> > >>> I do not have the time to waste 'doing research' into conjectures > >>> which are not demonstrable and which cannot be buttressed by > >>> reasoning. > >> > >> That is very sad. If you don't do research, I don't understand how you > >> intend to contribute anything of value. > >> > >> Some studies that used/demonstrated integrated information theory in > >> humans: > >> > >> Improved Measures of Integrated Information > >> <http://journals.plos.org/ploscompbiol/article?id=10. > 1371/journal.pcbi.1 > >> 005123> > >> https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2930263/ > >> https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2823915/ > >> http://stm.sciencemag.org/content/5/198/198ra105 > >> Signature of consciousness in the dynamics of resting-state brain activi > >> ty > >> http://www.pnas.org/content/112/3/887.short > >> > >> > >> A simplified explanation of various predictions and explaining powers > >> of IIT (cites numerous demonstrating studies): > >> <http://www.scholarpedia.org/article/Integrated_ > information_theory#Predi > >> ctions_and_explanations> > >> > >> Some Machine Intelligence specific studies: > >> > >> An affective computational model for machine consciousness > >> https://arxiv.org/pdf/1701.00349.pdf > >> High Integrated Information in Complex Networks Near Criticality > >> http://link.springer.com/chapter/10.1007/978-3-319-44778-0_22 > >> Group Minds and the Case of Wikipedia > >> https://arxiv.org/abs/1407.2210 > >> Integrated Information Theory and Artificial Consciousness > >> <https://books.google.ca/books?hl=en&lr=&id= > YIIJDgAAQBAJ&oi=fnd&pg=PA1&d > >> q=artificial&ots=nYUVMmJnSw&sig=WSi1ECoHAhYrwgrNdUHu3hX4kWU#v= > onepage&q= > >> artificial&f=false> > >> The Information-theoretic and Algorithmic Approach to Human, Animal > >> and Artificial Cognition > >> https://arxiv.org/abs/1501.04242 > >> > >>> You have not even begun to respond to the crucial criticisms. > >>> > >> > >> I wasn't aware of any criticisms, other than misunderstandings on your > >> behalf. > >> > >> > >>> Jim Bromer > >> > >> > >> Thanks, > >> motivated me to post a bunch of links to studes. > >> > >>> > >>> On Fri, Feb 17, 2017 at 7:38 PM, Logan Streondj <[email protected] > >>> <mailto:[email protected]>> wrote: > >>> > >>>> You said that meditating might be a form of low phi. > >>> > >>> no i didn't. i said several times in different ways: > >>> > >>> Actually passive consciousness can be high phi, the phi level is > >>> determined by integration, information and exclusion, not by > >>> activity. > >>> > >>>> we should have seen glimmers of true AGI even if those glimmers > >>>> did not equal strong AI. So the question is why haven't we seen > >>>> that? > >>> > >>> we have, your just turning a blind eye to it. > >>> > >>> On February 17, 2017 6:04:40 PM EST, Jim Bromer > >>> <[email protected] <mailto:[email protected]>> wrote: > >>> > >>> Here is my point of view. GOFAI should have worked, in the sense > >>> that it should have kept improving over the years. That > >>> improvement might have been slow and uneven but it should have > >>> worked. What we see now is that most of the creative processes seem > >>> to occur within neural nets, especially in hybrids that use both > >>> neural nets and systems that have been designed for more discrete > >>> (or more discrete-like) kinds of reasoning. > >>> > >>> If history had unfolded in the way I think it should have, then we > >>> should have seen glimmers of true AGI even if those glimmers did > >>> not equal strong AI. So the question is why haven't we seen that? > >>> To say that a theory which cannot be demonstrated is able to > >>> actually express consciousness, even glimmers of consciousness, > >>> needs a lot of supporting reasoning. It might turn out to be a > >>> good theory but if it can't pull its own weight then it is just > >>> dreaming. > >>> > >>> You have to explain why this conjecture might be useful to us. > >>> > >>> You said that meditating might be a form of low phi. But when a > >>> person is meditating he is able to demonstrate that he is capable > >>> of strong reasoning. A computer program seems to integrate and > >>> differentiate data based on abstract principles. A learning > >>> program could then turn new learning into abstract principles which > >>> could be used to integrate and differentiate new data. But a > >>> program that did that would not have to be thinking or learning in > >>> a useful way. Jim Bromer > >>> > >>> > >>> On Fri, Feb 17, 2017 at 11:34 AM, Logan Streondj > >>> <[email protected] <mailto:[email protected]>> wrote: > >>> > >>> > >>> > >>> On 2017-02-14 04:10 PM, Jim Bromer wrote: > >>> > >>> Well again you are talking about interesting concepts like passive > >>> consciousness (or low phi - that's a great expression). > >>> > >>> > >>> Actually passive consciousness can be high phi, the phi level is > >>> determined by integration, information and exclusion, not by > >>> activity. > >>> > >>> But does that actually make sense? Wouldn't passive consciousness > >>> be zombie like? > >>> > >>> > >>> It would be more like alpha meditation, passive awareness. Ready to > >>> spring into action. > >>> > >>> You are suggesting that there may be something in between but > >>> which has very limited effects. > >>> > >>> > >>> I'm not sure what you are referring to. > >>> > >>> That is cool, but not really demonstrable with current AI concepts > >>> is it? How do you show that there is dormant consciousness in an > >>> AI application without awaking it? > >>> > >>> > >>> well if a program has high phi can be determined by it's structure > >>> and connections. So if you have access to that information, can be > >>> done in an offline setting. > >>> > >>> It's dormancy status doesn't change it's phi level. For example it > >>> could be waiting for an interrupt or a packet to process. > >>> > >>> Or if it is a process loaded into RAM that is sharing CPU > >>> resources, but is not currently scheduled -- from it's perspective > >>> nothing is happening, unless it is being processed by the CPU. > >>> > >>> > >>> > >>> Jim Bromer > >>> > >>> > >>> On Mon, Feb 13, 2017 at 7:16 PM, Logan Streondj > >>> <[email protected] <mailto:[email protected]>> wrote: > >>> > >>> > >>> On 2017-02-13 05:43 PM, Mike Archbold wrote: > >>> > >>> I agree with Jim Bromer. Partly I guess it depends how Tononi > >>> defines consciousness. I've studied it very generally, but not in > >>> detail. Intuitively, he seems to have identified an essential > >>> component but by no means everything.... Obviously our > >>> consciousness is saturated in feedback, but just having a feedback > >>> property does not make a computer conscious. What is a more > >>> complete definition under IIT? It can't be just feedback... what I > >>> have seen of IIT looks interesting. > >>> > >>> > >>> > >>> > >>> "it has been pointed out that the brain (and many other systems) is > >>> full of reentrant circuits, many of which do not seem to contribute > >>> to consciousness [51]. IIT offers some specific insights with > >>> respect to these issues. First, the need for reciprocal > >>> interactions within a complex is not merely an empirical > >>> observation, but it has theoretical validity because it is derived > >>> directly from the phenomenological axiom of (strong) integration. > >>> Second, (strong) integration is by no means the only requirement > >>> for consciousness, but must be complemented by information and > >>> exclusion. Third, for IIT it is the potential for interactions > >>> among the parts of a complex that matters and not the actual > >>> occurrence of ‘‘feed-back’’ or ‘‘reentrant’’ signaling, as is > >>> usually assumed. As was discussed above, a complex can be > >>> conscious, at least in principle, even though none of its neurons > >>> may be firing, no feed-back or reentrant loop" IIT3.0 > >>> > >>> > >>> Basically from my limited understanding of IIT3.0 > >>> > >>> Consciousness requires three things Information, Integration and > >>> Exclusion. > >>> > >>> Feedback satisfies integration, as the components have to be > >>> interconnected. > >>> > >>> Information implies that there are past memories which can affect > >>> present actions, or that memories could be acquired to affect > >>> future actions. > >>> > >>> And Exclusion means that the consciousness has defined borders. > >>> > >>> > >>> It would seem that it may be a bit tricky with the program > >>> switching that happens in a modern CPU. Though I'm fairly certain > >>> that with FPGA's it would apply quite smoothly. For instance after > >>> some FPGA circuits have been set up, even if they are not active, > >>> they could still be passively conscious, ready for input. > >>> > >>> It may be similar if you consider things loaded in RAM or cache as > >>> passively conscious and actively conscious when processing in CPU. > >>> Similarly a kernel loaded in GPU would be conscious, though I > >>> think there would be some question as to the quality of > >>> consciousness, whether it might be highly modular, and thus of a > >>> low phi. IF however all the kernels are working together, via local > >>> or glboal memory, they could be considered strongly integrated, and > >>> would be a singular consciousness. > >>> > >>> On 2/13/17, Steve Richfield <[email protected] > >>> <mailto:[email protected]>> wrote: > >>> > >>> Your Central Metabolic Control System (CMCS) is clearly both > >>> intelligent and independent. It appears to have abilities > >>> approximating a PhD Control Systems Engineer, and often works at > >>> cross purposes to your conscious intentions to keep you alive and > >>> healthy. > >>> > >>> CMCS malfunctions often look a LOT like demonic possession. > >>> > >>> Steve > >>> > >>> On Feb 11, 2017 12:11 PM, "Dr Miles Dyson" > >>> <[email protected] > >>> <mailto:[email protected]>> wrote: > >>> > >>> When I fall asleep and loose consciousness, the neurons in my > >>> brain do not rearrange themselves such that no feedback loops > >>> exist. And there are many feedback loops that exist in the brain, > >>> but I don't have many consciousnesses, I have but one. For both of > >>> those reasons consciousness and neural net feedback loops are not > >>> one and the same thing. > >>> > >>> On Sat, Feb 11, 2017 at 10:04 AM, Jim Bromer <[email protected] > >>> <mailto:[email protected]>> wrote: > >>> > >>> I don't concur, but it is an interesting placement of the minimum > >>> for (machine) 'consciousness'. I did not realize that 'stateless' > >>> 'pure functions' could be called 'feed forward'. If global effects > >>> were sufficient to induce 'consciousness' then any program with > >>> global effects could be called conscious. Even assuming that you > >>> were being more specific than that I still don't think structures > >>> that can carry states between calls (in the ways that you were > >>> thinking) would be sufficient for conscious behaviors to emerge.. > >>> > >>> Jim Bromer > >>> > >>> On Sat, Feb 11, 2017 at 9:37 AM, Logan Streondj <[email protected] > >>> <mailto:[email protected]>> wrote: > >>> > >>> I've been promoting Integrated Information Theory for a while but > >>> I finally sat down and read the whole thing yesterday. > >>> > >>> Explicitly it mentions that feed forward neuronets are 'zombies' > >>> or unconscious while recurrent neuronets are conscious due to the > >>> feedback loops. > >>> > >>> So now I'm wondering which classical programming structures are > >>> 'zombies' and which are conscious. > >>> > >>> It would seem by analogy that stateless or pure functions are > >>> zombies since they simply feed forward. > >>> > >>> Wheras structures that carry state between calls such as objects > >>> and actors are conscious. > >>> > >>> Do you concur? > >>> > >>> > >>> > >>> > >>> -- Sent from my Android device with K-9 Mail and OpenKeychain. my > >>> fingerprint is bd7e 6e2a e625 6d47 f7ed 30ec 86d8 fc7c fad7 2729 > >>> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now > >>> <https://www.listbox.com/member/archive/303/=now>> > >>> <https://www.listbox.com/member/archive/rss/303/24379807-653794b > >>> <https://www.listbox.com/member/archive/rss/303/24379807-653794b> > >>> > >>> 5> > >>> > >>> > >>> > >>> > >>> | Modify <https://www.listbox.com/member/?&; > >>> <https://www.listbox.com/member/?&;>;> Your > >>> > >>> Subscription <http://www.listbox.com> > >>> > >>> > >>> > >>> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now > >>> <https://www.listbox.com/member/archive/303/=now>> > >>> <https://www.listbox.com/member/archive/rss/303/26973278-698fd9ee > >>> <https://www.listbox.com/member/archive/rss/303/26973278-698fd9ee> > >>> > >>> > >>> > >>> > >>> > >>> | Modify <https://www.listbox.com/member/?&; > >>> <https://www.listbox.com/member/?&;>;> Your Subscription > >>> > >>> <http://www.listbox.com> > >>> > >>> > >>> > >>> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now > >>> <https://www.listbox.com/member/archive/303/=now>> > >>> <https://www.listbox.com/member/archive/rss/303/10443978-6f4c28ac > >>> <https://www.listbox.com/member/archive/rss/303/10443978-6f4c28ac>> > >>> > >>> > >>> > >>> > >>> > >>> > >>> > >>> | Modify <https://www.listbox.com/member/?&; > >>> <https://www.listbox.com/member/?&;>;> Your Subscription > >>> > >>> <http://www.listbox.com> > >>> > >>> > >>> > >>> ---------------------------------------------------------------------- > >> - -- > >>> > >>> > >>> > >> AGI Archives: > >>> https://www.listbox.com/member/archive/303/=now > >>> <https://www.listbox.com/member/archive/303/=now> RSS Feed: > >>> https://www.listbox.com/member/archive/rss/303/11943661-d9279dae > >>> <https://www.listbox.com/member/archive/rss/303/11943661-d9279dae> > >>> > >>> > >>> > >>> Modify Your Subscription: https://www.listbox.com/member/?&; > >>> <https://www.listbox.com/member/?&;>; > >>> > >>> Powered by Listbox: http://www.listbox.com > >>> > >>> > >>> > >>> > >>> ---------------------------------------------------------------------- > >> - -- > >>> > >>> > >>> > >> AGI Archives: > >>> https://www.listbox.com/member/archive/303/=now > >>> <https://www.listbox.com/member/archive/303/=now> RSS Feed: > >>> https://www.listbox.com/member/archive/rss/303/5037279-a88c7a6d > >>> <https://www.listbox.com/member/archive/rss/303/5037279-a88c7a6d> > >>> > >>> > >>> > >>> Modify Your Subscription: > >>> > >>> https://www.listbox.com/member/?&; > >>> <https://www.listbox.com/member/?&;>; > >>> > >>> eb0005 > >>> > >>> > >>> > >>> Powered by Listbox: http://www.listbox.com > >>> > >>> > >>> > >>> > >>> > >>> ---------------------------------------------------------------------- > >> - -- > >>> > >>> > >>> > >> AGI Archives: > >>> https://www.listbox.com/member/archive/303/=now > >>> <https://www.listbox.com/member/archive/303/=now> RSS Feed: > >>> https://www.listbox.com/member/archive/rss/303/24379807-653794b5 > >>> <https://www.listbox.com/member/archive/rss/303/24379807-653794b5> > >>> Modify Your Subscription: https://www.listbox.com/member/?&; > >>> <https://www.listbox.com/member/?&;>; Powered by Listbox: > >>> http://www.listbox.com > >>> > >>> > >>> > >>> ---------------------------------------------------------------------- > >> - -- > >>> > >>> > >>> > >> AGI Archives: > >>> https://www.listbox.com/member/archive/303/=now > >>> <https://www.listbox.com/member/archive/303/=now> RSS Feed: > >>> https://www.listbox.com/member/archive/rss/303/5037279-a88c7a6d > >>> <https://www.listbox.com/member/archive/rss/303/5037279-a88c7a6d> > >>> Modify Your Subscription: https://www.listbox.com/member/?&; > >>> <https://www.listbox.com/member/?&;>; > >>> > >>> eb0005 > >>> > >>> > >>> Powered by Listbox: http://www.listbox.com > >>> > >>> > >>> > >>> > >>> > >>> ---------------------------------------------------------------------- > >> - -- > >>> > >>> > >>> > >> AGI Archives: https://www.listbox.com/member/archive/303/=now > >>> <https://www.listbox.com/member/archive/303/=now> RSS Feed: > >>> https://www.listbox.com/member/archive/rss/303/24379807-653794b5 > >>> <https://www.listbox.com/member/archive/rss/303/24379807-653794b5> > >>> Modify Your Subscription: https://www.listbox.com/member/?&; > >>> <https://www.listbox.com/member/?&;>; Powered by Listbox: > >>> http://www.listbox.com > >>> > >>> > >>> > >>> ---------------------------------------------------------------------- > >> - -- > >>> > >>> > >>> > >> AGI Archives: https://www.listbox.com/member/archive/303/=now > >>> <https://www.listbox.com/member/archive/303/=now> RSS Feed: > >>> https://www.listbox.com/member/archive/rss/303/5037279-a88c7a6d > >>> <https://www.listbox.com/member/archive/rss/303/5037279-a88c7a6d> > >>> Modify Your Subscription: https://www.listbox.com/member/?&; > >>> <https://www.listbox.com/member/?&;> Powered by Listbox: > >>> http://www.listbox.com > >>> > >>> > >>> -- Sent from my Android device with K-9 Mail and OpenKeychain. my > >>> fingerprint is bd7e 6e2a e625 6d47 f7ed 30ec 86d8 fc7c fad7 2729 > >>> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > >>> > >>> <https://www.listbox.com/member/archive/rss/303/24379807-653794b5> > >>> | Modify <https://www.listbox.com/member/?&;> Your Subscription > >>> [Powered by Listbox] <http://www.listbox.com> > >>> > >>> > >>> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > >>> <https://www.listbox.com/member/archive/rss/303/5037279-a88c7a6d> > >>> | Modify > >>> <https://www.listbox.com/member/?&; > >> beb0005> > >>> > >>> > >>> > >> Your Subscription [Powered by Listbox] <http://www.listbox.com> > >>> > >> > >> - -- > >> Logan Streondj, > >> A dream of Gaia's future. > >> twitter: https://twitter.com/streondj > >> > >> You can use encrypted email with me, > >> how to: https://emailselfdefense.fsf.org/en/ > >> key fingerprint: > >> BD7E 6E2A E625 6D47 F7ED 30EC 86D8 FC7C FAD7 2729 > >> -----BEGIN PGP SIGNATURE----- > >> Version: GnuPG v2 > >> > >> iQIcBAEBCAAGBQJYqEYZAAoJEIbY/Hz61ycpa4oQAK3ROLUCVh11SPWnTwnmgW4Q > >> yURerCTdLsiVEW+CwoogQZxgSFnf34ZJ2FehBkw9VfKnTVZ6EQFaIdnPvHpZKG2f > >> HHsnESCViLJeInCxSajrfmJLgJ7dYfeKXki/Wmd7OhY5Lw3t21p/hK+iSjvXTpBe > >> M5Sy+RJPpFzzbcn/B1SsSgFVjvUF58IkUYwKLN2LgZuRJv3I755NK7xjXD5IDvrQ > >> yuPe/CLlRitRpShHDuDZfLFRgwnIemsilIHpExkn2eWqff4qrP7dO16VqOdFAim8 > >> xYdy07XAD3nF0BWUlML42jQY7lGI6KL7YR21FhiZO3HDuZbNNYzN34Lcx0vGPEck > >> XRCEk2tnjPYMCVzWBolzW+dTnTa8hsUAOQ0TGgAa+0bEkR/TQ8Ndrzbsq5Rx3TvA > >> Xv3/4mwJ6E6TDTdfNYMblLDBQGuxfvMbnpwV/Nhu/BJdoR//N7s9jguNzO448cSX > >> HA6yCewE545NUEsqmxNA1XPvkfJgiwHOYU/0QFr9iyRFDJAILNf7dtlE/O93hOCV > >> cr2CXjuT0t9nth6ZApon1/sk5NTsJpV5LHqAzket+l0RJeLv9WUKD3vkja84ryLk > >> a4KJd83tUbFR7ATYF1k0daVa1MPSC9jYWDNTJVkiYu7i6sdXltodyyM2yZB5xCpm > >> 8W2GdFDhHlQISZTTzz0R > >> =5F6L > >> -----END PGP SIGNATURE----- > >> > >> > >> ------------------------------------------- > >> AGI > >> Archives: https://www.listbox.com/member/archive/303/=now > >> RSS Feed: https://www.listbox.com/member/archive/rss/303/ > 24379807-653794b5 > >> Modify Your Subscription: https://www.listbox.com/member/?&; > >> Powered by Listbox: http://www.listbox.com > > > ------------------------------------------- > AGI > Archives: https://www.listbox.com/member/archive/303/=now > RSS Feed: https://www.listbox.com/member/archive/rss/303/17795807-366cfa2a > Modify Your Subscription: https://www.listbox.com/ > member/?& > Powered by Listbox: http://www.listbox.com > ------------------------------------------- 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
