Re: [opencog-dev] Openness of knowledge-base of Hanson robotics
On Mon, Oct 30, 2017 at 9:01 PM, Nil Geisweiller wrote: > On 10/31/2017 02:52 AM, Mark Nuzz wrote: >> >> I am involved - I have taken time to respond to questions by a number >> of new posters to the group, to the best of my ability, in good faith. > > > That is greatly appreciated. > >> You've expressed, a few posts back, clear frustration with the fact >> that people have close to zero interest in knowledge-bases. That >> doesn't make much sense to me, and the way you put it, shows that it >> is worth checking into. Knowledge bases, being a key component of any >> AGI system, are pretty damned mandatory if you want to experiment with >> an AI system that isn't a blank slate. So when you express an > > > Yes, just note that OpenCog's primary goal isn't to provide pre-built KBs, > other projects do that like SUMO for instance http://www.adampease.org/OP/. > OpenCog's primary goal is to provide the tools that can build such KBs from > textual, sensory or whatever data. That said pre-built KBs can be very > useful, fortunately it's usually not too much work to import them to > atomese. Here's for instance a SUMO importer > > https://github.com/opencog/external-tools/tree/master/SUMO_importer > > and an example using it > > https://github.com/opencog/opencog/tree/master/examples/pln/sumo > > Nil > Thanks for the responses. This helps my understanding a lot. I wish I could devote more time to this project, Linas did have a point in a sense that it is hard to contribute casually. -- You received this message because you are subscribed to the Google Groups "opencog" group. To unsubscribe from this group and stop receiving emails from it, send an email to opencog+unsubscr...@googlegroups.com. To post to this group, send email to opencog@googlegroups.com. Visit this group at https://groups.google.com/group/opencog. To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/CAMyYmr9QTxGcF0XNKhRuXp5TKfP80-1L890DrOmJFfV%3D%2B2gvog%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
Re: [opencog-dev] Openness of knowledge-base of Hanson robotics
A side note regarding Linas's comments on the Sophia robot... I note that from an AI view the robot is a platform, that can be used to run many different software systems... We use Sophia and the other Hanson robots to experiment with OpenCog sometimes... But for public theatrical performances, currently a non-OpenCog-based chat system as used, which is mostly but not entirely open source at the moment Linas said some things about this non-OpenCog-based system in a message above; though I feel his comments were overly negative and did not do this system justice, indeed it's not an AGI system and is not incrementally evolvable into one... We aim to shift to using OpenCog as the main theatrical control system for Sophia and the other Hanson robots sometime soon, most likely in Q1 of 2018. This won't immediately bring amazing AGI to the robots, but it will start a process of steady incremental improvement of the robots' intelligence... -- Ben G On Mon, Oct 30, 2017 at 9:01 PM, 'Nil Geisweiller' via opencog wrote: > On 10/31/2017 02:52 AM, Mark Nuzz wrote: >> >> I am involved - I have taken time to respond to questions by a number >> of new posters to the group, to the best of my ability, in good faith. > > > That is greatly appreciated. > >> You've expressed, a few posts back, clear frustration with the fact >> that people have close to zero interest in knowledge-bases. That >> doesn't make much sense to me, and the way you put it, shows that it >> is worth checking into. Knowledge bases, being a key component of any >> AGI system, are pretty damned mandatory if you want to experiment with >> an AI system that isn't a blank slate. So when you express an > > > Yes, just note that OpenCog's primary goal isn't to provide pre-built KBs, > other projects do that like SUMO for instance http://www.adampease.org/OP/. > OpenCog's primary goal is to provide the tools that can build such KBs from > textual, sensory or whatever data. That said pre-built KBs can be very > useful, fortunately it's usually not too much work to import them to > atomese. Here's for instance a SUMO importer > > https://github.com/opencog/external-tools/tree/master/SUMO_importer > > and an example using it > > https://github.com/opencog/opencog/tree/master/examples/pln/sumo > > Nil > > >> apparently strong belief that this is not something people are >> interested in (even if you yourself are), then as a core developer >> you're less likely to put any work or polish into that feature. >> However, if the reason that people aren't interested is because there >> isn't enough work or polish into that feature, then perhaps if I can >> help you understand this, it would be very good for the project, so my >> efforts here would have some chances of adding a non-trivial amount of >> value, with the limited time that I am able to spare toward the cause. >> >> The open-source paradigm was designed precisely so that decisions, >> assumptions, beliefs, and implementations could be questioned openly, >> even "from the sidelines" as you put it. A component of the theory, >> which has not been proven false to my knowledge, is that it usually >> yields net positive value for the project, over the lifetime of the >> project. You can disagree with the points that I make, and I could be >> wrong, and my specific suggestions might even be a waste of time, but >> if that were the case, then I'd learn something new from it. And this >> has happened a few times already, resulting in my better understanding >> to the point of being more able to contribute or discuss in more >> meaningful ways. But in this case, it sounds like you're unhappy with >> the very idea that your ideas are being questioned in the first place. >> >> Besides, if it really was personal, any relevant questions I had would >> already be answered at that point, and it would be actions being done, >> not questions. You are probably spending too much time around >> passive-aggressive people or people who aren't very nice to you. >> Lighten up a bit and know who your real allies are in this game. >> > > -- > You received this message because you are subscribed to the Google Groups > "opencog" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to opencog+unsubscr...@googlegroups.com. > To post to this group, send email to opencog@googlegroups.com. > Visit this group at https://groups.google.com/group/opencog. > To view this discussion on the web visit > https://groups.google.com/d/msgid/opencog/d2fc96b3-0364-31dd-c8c3-2616a7aad57b%40gmail.com. > > For more options, visit https://groups.google.com/d/optout. -- Ben Goertzel, PhD http://goertzel.org "I am God! I am nothing, I'm play, I am freedom, I am life. I am the boundary, I am the peak." -- Alexander Scriabin -- You received this message because you are subscribed to the Google Groups "opencog" group. To unsubscribe from this group and stop receiving emails from it, send an email to open
Re: [opencog-dev] Openness of knowledge-base of Hanson robotics
On 10/31/2017 02:52 AM, Mark Nuzz wrote: I am involved - I have taken time to respond to questions by a number of new posters to the group, to the best of my ability, in good faith. That is greatly appreciated. You've expressed, a few posts back, clear frustration with the fact that people have close to zero interest in knowledge-bases. That doesn't make much sense to me, and the way you put it, shows that it is worth checking into. Knowledge bases, being a key component of any AGI system, are pretty damned mandatory if you want to experiment with an AI system that isn't a blank slate. So when you express an Yes, just note that OpenCog's primary goal isn't to provide pre-built KBs, other projects do that like SUMO for instance http://www.adampease.org/OP/. OpenCog's primary goal is to provide the tools that can build such KBs from textual, sensory or whatever data. That said pre-built KBs can be very useful, fortunately it's usually not too much work to import them to atomese. Here's for instance a SUMO importer https://github.com/opencog/external-tools/tree/master/SUMO_importer and an example using it https://github.com/opencog/opencog/tree/master/examples/pln/sumo Nil apparently strong belief that this is not something people are interested in (even if you yourself are), then as a core developer you're less likely to put any work or polish into that feature. However, if the reason that people aren't interested is because there isn't enough work or polish into that feature, then perhaps if I can help you understand this, it would be very good for the project, so my efforts here would have some chances of adding a non-trivial amount of value, with the limited time that I am able to spare toward the cause. The open-source paradigm was designed precisely so that decisions, assumptions, beliefs, and implementations could be questioned openly, even "from the sidelines" as you put it. A component of the theory, which has not been proven false to my knowledge, is that it usually yields net positive value for the project, over the lifetime of the project. You can disagree with the points that I make, and I could be wrong, and my specific suggestions might even be a waste of time, but if that were the case, then I'd learn something new from it. And this has happened a few times already, resulting in my better understanding to the point of being more able to contribute or discuss in more meaningful ways. But in this case, it sounds like you're unhappy with the very idea that your ideas are being questioned in the first place. Besides, if it really was personal, any relevant questions I had would already be answered at that point, and it would be actions being done, not questions. You are probably spending too much time around passive-aggressive people or people who aren't very nice to you. Lighten up a bit and know who your real allies are in this game. -- You received this message because you are subscribed to the Google Groups "opencog" group. To unsubscribe from this group and stop receiving emails from it, send an email to opencog+unsubscr...@googlegroups.com. To post to this group, send email to opencog@googlegroups.com. Visit this group at https://groups.google.com/group/opencog. To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/d2fc96b3-0364-31dd-c8c3-2616a7aad57b%40gmail.com. For more options, visit https://groups.google.com/d/optout.
Re: [opencog-dev] Openness of knowledge-base of Hanson robotics
I'm sure we're all here in favor of open KB. But we can't force companies to open their KBs. The way I see it is that they would tend to keep a part private, as long as they feel the market can take it, but eventually everything would go public. Maybe peer-to-peer technology could be used to counter that, etc. Nil On 10/31/2017 02:52 AM, Mark Nuzz wrote: On Mon, Oct 30, 2017 at 5:24 PM, Linas Vepstas wrote: On Mon, Oct 30, 2017 at 6:48 PM, Mark Nuzz wrote: Hi Mark, I feel like you are goading me, trying to get me to loose my temper, and get angry. And you are being successful. My reply down below is WTF, we've been laying this stuff at your feet for a decade. All you have to do is get involved, instead of sniping from the sidelines. --linas I am involved - I have taken time to respond to questions by a number of new posters to the group, to the best of my ability, in good faith. You've expressed, a few posts back, clear frustration with the fact that people have close to zero interest in knowledge-bases. That doesn't make much sense to me, and the way you put it, shows that it is worth checking into. Knowledge bases, being a key component of any AGI system, are pretty damned mandatory if you want to experiment with an AI system that isn't a blank slate. So when you express an apparently strong belief that this is not something people are interested in (even if you yourself are), then as a core developer you're less likely to put any work or polish into that feature. However, if the reason that people aren't interested is because there isn't enough work or polish into that feature, then perhaps if I can help you understand this, it would be very good for the project, so my efforts here would have some chances of adding a non-trivial amount of value, with the limited time that I am able to spare toward the cause. The open-source paradigm was designed precisely so that decisions, assumptions, beliefs, and implementations could be questioned openly, even "from the sidelines" as you put it. A component of the theory, which has not been proven false to my knowledge, is that it usually yields net positive value for the project, over the lifetime of the project. You can disagree with the points that I make, and I could be wrong, and my specific suggestions might even be a waste of time, but if that were the case, then I'd learn something new from it. And this has happened a few times already, resulting in my better understanding to the point of being more able to contribute or discuss in more meaningful ways. But in this case, it sounds like you're unhappy with the very idea that your ideas are being questioned in the first place. Besides, if it really was personal, any relevant questions I had would already be answered at that point, and it would be actions being done, not questions. You are probably spending too much time around passive-aggressive people or people who aren't very nice to you. Lighten up a bit and know who your real allies are in this game. -- You received this message because you are subscribed to the Google Groups "opencog" group. To unsubscribe from this group and stop receiving emails from it, send an email to opencog+unsubscr...@googlegroups.com. To post to this group, send email to opencog@googlegroups.com. Visit this group at https://groups.google.com/group/opencog. To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/e28ce637-9a80-cfd4-fb56-abf1157f205c%40gmail.com. For more options, visit https://groups.google.com/d/optout.
Re: [opencog-dev] Openness of knowledge-base of Hanson robotics
On Mon, Oct 30, 2017 at 5:24 PM, Linas Vepstas wrote: > > > On Mon, Oct 30, 2017 at 6:48 PM, Mark Nuzz wrote: > > Hi Mark, I feel like you are goading me, trying to get me to loose my > temper, and get angry. And you are being successful. My reply down below is > WTF, we've been laying this stuff at your feet for a decade. All you have to > do is get involved, instead of sniping from the sidelines. > > --linas I am involved - I have taken time to respond to questions by a number of new posters to the group, to the best of my ability, in good faith. You've expressed, a few posts back, clear frustration with the fact that people have close to zero interest in knowledge-bases. That doesn't make much sense to me, and the way you put it, shows that it is worth checking into. Knowledge bases, being a key component of any AGI system, are pretty damned mandatory if you want to experiment with an AI system that isn't a blank slate. So when you express an apparently strong belief that this is not something people are interested in (even if you yourself are), then as a core developer you're less likely to put any work or polish into that feature. However, if the reason that people aren't interested is because there isn't enough work or polish into that feature, then perhaps if I can help you understand this, it would be very good for the project, so my efforts here would have some chances of adding a non-trivial amount of value, with the limited time that I am able to spare toward the cause. The open-source paradigm was designed precisely so that decisions, assumptions, beliefs, and implementations could be questioned openly, even "from the sidelines" as you put it. A component of the theory, which has not been proven false to my knowledge, is that it usually yields net positive value for the project, over the lifetime of the project. You can disagree with the points that I make, and I could be wrong, and my specific suggestions might even be a waste of time, but if that were the case, then I'd learn something new from it. And this has happened a few times already, resulting in my better understanding to the point of being more able to contribute or discuss in more meaningful ways. But in this case, it sounds like you're unhappy with the very idea that your ideas are being questioned in the first place. Besides, if it really was personal, any relevant questions I had would already be answered at that point, and it would be actions being done, not questions. You are probably spending too much time around passive-aggressive people or people who aren't very nice to you. Lighten up a bit and know who your real allies are in this game. -- You received this message because you are subscribed to the Google Groups "opencog" group. To unsubscribe from this group and stop receiving emails from it, send an email to opencog+unsubscr...@googlegroups.com. To post to this group, send email to opencog@googlegroups.com. Visit this group at https://groups.google.com/group/opencog. To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/CAMyYmr93xMd66vyk%3DFmz_iE3adOOSxyh06YTjycXr95FkTcQPQ%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
Re: [opencog-dev] Openness of knowledge-base of Hanson robotics
On Mon, Oct 30, 2017 at 6:48 PM, Mark Nuzz wrote: Hi Mark, I feel like you are goading me, trying to get me to loose my temper, and get angry. And you are being successful. My reply down below is WTF, we've been laying this stuff at your feet for a decade. All you have to do is get involved, instead of sniping from the sidelines. --linas On Mon, Oct 30, 2017 at 4:36 PM, Linas Vepstas > wrote: > > Hi Mark, > > > > > > The answer is "yes", but perhaps not the way you are expecting it. Its > not > > like we have some defined format for "semantic triples" or whatever. > There > > are a large number of rich data representation styles that have been > used in > > a variety of projects. The commonalities in all of these are expressed as > > "atomese". That is, the standard protocol is "atomese". > > That's exactly the type of thing I had in mind, actually (hence > serialization). Have you validated your assumption that the lack of > interest in the downloading of knowledge-bases is due to "nearly-zero > interest in knowledgebases", as you claim? Well, did you download any of them? > Or is it possible that the > atomese json files They are not in json. > are not providing enough expected utility? Utility to whom? For what purpose? What do you have in mind? > What can > one do with those files, currently? Well, 10 years ago I provided files that gave semantic annotations (wordnet) to parsed sentences. If that's what you needed to do, then I had data for you. More recently I provided files that gave a provisional grammar for Chinese (mandarin). Did you need a Chinese grammar for anything? I'm guessing you didn't. More abstractly, there is a docker file that allows you to run an earlier version of the Hanson Robtics software stack, called "Eva" back then (before the Eva movie came out) - you could interact with her on-screen, chat to her by irc, using your webcam on your laptop. Did anyone ever run this besides me? As far as I can tell, the answer is "no". Its not just about me me me -- other teams working with opencog have published various datasets, including some genomic data. Its still sitting there in some github repo somewhere. Last I looked, there were like 5 clones of this dataset. The loving AI project has maybe 9 clones, of which I think all 9 of which are actively involved project participants. No one outside of the project is interested in this data. > Can I simply drop them into a data > directory and have them be automatically consumed by OpenCog Yes. > or is > there a difficult process to get them to work? Yes. > Is atomspace able to > deduplicate imported atoms that are conceptually equivalent to atoms > already in the database, Yes. > but not exactly equal? > Ah hah! Trick question! "conceptually equivalent but not equal" Yes, the code can be found here: https://github.com/opencog/atomspace/tree/master/opencog/sheaf and here: https://github.com/opencog/opencog/blob/master/opencog/nlp/learn/scm/gram-class.scm this code is in active development. --linas -- *"The problem is not that artificial intelligence will get too smart and take over the world," computer scientist Pedro Domingos writes, "the problem is that it's too stupid and already has." * -- You received this message because you are subscribed to the Google Groups "opencog" group. To unsubscribe from this group and stop receiving emails from it, send an email to opencog+unsubscr...@googlegroups.com. To post to this group, send email to opencog@googlegroups.com. Visit this group at https://groups.google.com/group/opencog. To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/CAHrUA34tsGKL3o6%2BW0RXq8g7n%2B2haeM_0Qb4pvD1ZUZX2MWb4g%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
Re: [opencog-dev] Openness of knowledge-base of Hanson robotics
On Mon, Oct 30, 2017 at 4:36 PM, Linas Vepstas wrote: > Hi Mark, > > > The answer is "yes", but perhaps not the way you are expecting it. Its not > like we have some defined format for "semantic triples" or whatever. There > are a large number of rich data representation styles that have been used in > a variety of projects. The commonalities in all of these are expressed as > "atomese". That is, the standard protocol is "atomese". That's exactly the type of thing I had in mind, actually (hence serialization). Have you validated your assumption that the lack of interest in the downloading of knowledge-bases is due to "nearly-zero interest in knowledgebases", as you claim? Or is it possible that the atomese json files are not providing enough expected utility? What can one do with those files, currently? Can I simply drop them into a data directory and have them be automatically consumed by OpenCog, or is there a difficult process to get them to work? Is atomspace able to deduplicate imported atoms that are conceptually equivalent to atoms already in the database, but not exactly equal? -- You received this message because you are subscribed to the Google Groups "opencog" group. To unsubscribe from this group and stop receiving emails from it, send an email to opencog+unsubscr...@googlegroups.com. To post to this group, send email to opencog@googlegroups.com. Visit this group at https://groups.google.com/group/opencog. To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/CAMyYmr_5hc0KXH03fmcFina5hur76M2t2K1xmUV4e%2BsFO%3DsSMg%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
Re: [opencog-dev] Openness of knowledge-base of Hanson robotics
Hi Mark, The answer is "yes", but perhaps not the way you are expecting it. Its not like we have some defined format for "semantic triples" or whatever. There are a large number of rich data representation styles that have been used in a variety of projects. The commonalities in all of these are expressed as "atomese". That is, the standard protocol is "atomese". Think of it this way: in your mind, substitute in "json" for "atomese". If you can represent it in json, you can represent it in atomese. (atomese is richer) We are not in the standards business of saying "here's a predefined format that you can use to represent airplane part numbers". The SQL vendors don't do this, the no-sql vendors don't do this, the json maintainers don't do this, and opencog doesn't do this either. Airplane part numbers are interesting, but opencog is not going to write standards for that kind of knowledge representation. Substitute any other kind of "knowledge" you are interested in, for "airplane part number". That said, you can look at some specific examples. One that I am currently working on is here: https://github.com/opencog/opencog/blob/master/opencog/nlp/learn/scm/gram-class.scm --linas On Mon, Oct 30, 2017 at 6:07 PM, Mark Nuzz wrote: > On Mon, Oct 30, 2017 at 12:31 PM, Linas Vepstas > wrote: > > Hi Ivan, > > > > Let me top-post answers to your questions. First, one needs to clarify: > > "what is a knowledgebase?" There are multiple components that can be > called > > "a knowledgebase", and they are all quite very different. > > > > Hi Linas, > > Does OpenCog define or use a standard protocol for serialization of a > knowledge base? Is there a schema or format? Is there a means to > deduplicate knowledge imported through these scripts? It sounds like > this is not the case but I'd love to check out any Wiki pages that > describe the state of this feature in OpenCog, as well as a roadmap. > Do you have any suggestions for pages to lookup to find more about it? > Thanks much. > > Mark > > -- > You received this message because you are subscribed to the Google Groups > "opencog" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to opencog+unsubscr...@googlegroups.com. > To post to this group, send email to opencog@googlegroups.com. > Visit this group at https://groups.google.com/group/opencog. > To view this discussion on the web visit https://groups.google.com/d/ > msgid/opencog/CAMyYmr8sz-ZMjsRWOiGLAoeh-D%2BWcAyEgyyoFhNAP%3DX0aJniWw% > 40mail.gmail.com. > For more options, visit https://groups.google.com/d/optout. > -- *"The problem is not that artificial intelligence will get too smart and take over the world," computer scientist Pedro Domingos writes, "the problem is that it's too stupid and already has." * -- You received this message because you are subscribed to the Google Groups "opencog" group. To unsubscribe from this group and stop receiving emails from it, send an email to opencog+unsubscr...@googlegroups.com. To post to this group, send email to opencog@googlegroups.com. Visit this group at https://groups.google.com/group/opencog. To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/CAHrUA37YeqRNobZ1x6wFP7qmbD8ocHoTUhu_g7-kUPX%3DT6Ybmw%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
Re: [opencog-dev] Openness of knowledge-base of Hanson robotics
On Mon, Oct 30, 2017 at 12:31 PM, Linas Vepstas wrote: > Hi Ivan, > > Let me top-post answers to your questions. First, one needs to clarify: > "what is a knowledgebase?" There are multiple components that can be called > "a knowledgebase", and they are all quite very different. > Hi Linas, Does OpenCog define or use a standard protocol for serialization of a knowledge base? Is there a schema or format? Is there a means to deduplicate knowledge imported through these scripts? It sounds like this is not the case but I'd love to check out any Wiki pages that describe the state of this feature in OpenCog, as well as a roadmap. Do you have any suggestions for pages to lookup to find more about it? Thanks much. Mark -- You received this message because you are subscribed to the Google Groups "opencog" group. To unsubscribe from this group and stop receiving emails from it, send an email to opencog+unsubscr...@googlegroups.com. To post to this group, send email to opencog@googlegroups.com. Visit this group at https://groups.google.com/group/opencog. To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/CAMyYmr8sz-ZMjsRWOiGLAoeh-D%2BWcAyEgyyoFhNAP%3DX0aJniWw%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
Re: [opencog-dev] Openness of knowledge-base of Hanson robotics
Hello Linas, Thank you for such comprehensive answer. That helped me to understand the current state of development and gaining the knowledge. And that Sopia is still far from three-years-kid. I'm just getting the knowledge about AGI and OpenCog, and filling in the technical gaps w/o hands on experience with OpenCog. And one of the problem that I've already read was the exactly the complexity of getting starting with opencog for the new members. I don't know the problems in the details : probably lackng of simple deployment ( as ubuntu package as an example) , some demo embodiment, tutorial and so on. But, probably growing community would exactly help with the training the robots and sharing the knowledge through the cloud. So this would be very impressive if easily deployed and started demo robot could sync the knowledge with the cloud and send the newly obtained knowledge back. This would be a kind of collaboration in teaching the robot and creating world-knowledge-base. To make this happen, you probably would need to solve the problems that hinders the new members to easily join to the community. But this is only my comments, I don't know how this fits into your visions. Ivan 30 Окт 2017 г. 20:32 пользователь "Linas Vepstas" написал: Hi Ivan, Let me top-post answers to your questions. First, one needs to clarify: "what is a knowledgebase?" There are multiple components that can be called "a knowledgebase", and they are all quite very different. Lets start with the Hanson Robotic Sophia Personality. This is not one knowledgebase, but several: A) A collection of (partly-hand-authored) face and arm animations. Not terribly useful, if you don't have the physical robot. (and there are many robots, with slightly different motors and mechanical details) You CAN also use these on the blender model, without the physical robot. The older animations were open-source, not sure if the current ones are. Based on interactions on the mailing list, exactly ZERO people (outside the HR circle) downloaded and used these. I used to wish that this could become a viable open-source project on its own, with active developers and hobbyists using it. No such luck. Basically, animators don't create animations as open-source projects. Hollywood has instilled a highly proprietary, closed-source, intellectual-property anti-piracy mindset in this class of people. What might it take to change this? HR did hire a professional animation company, and paid oodles of money, to expand the animation set. I don't know if these are public or private. Doesn't much matter, if no one cares. B) A collection of (hand-authored) monologues tailored for various public appearances. Stuff like "Be sure to visit booth B42, where Gadget Corp is displaying their latest gizmo." These are proprietary, but you wouldn't want these anyway. C) general personality. This is scattered across multiple databases. One of them is https://github.com/opencog/loving-ai which is public. Other bits are private. Its her personality, how she reacts to things, the words that fall out of her mouth. D) Knowledge about English and Russian grammar. This is in link-grammar, its public. The robots more or less do not use this subsystem, expect in a few half-baked, half-operational demos that have been given over the years. The demos can and will be improved, but its slow going. E) knowledge about the world. There is no such database at this time. The robot is as dumb as a rock. Its not a whole lot more than a glorified chatbot, at this time. (I actually counted -- the robot has somewhere between 5 and 8 bits of knowledge about the external world. That's "bits", not bytes.) The goal is, of course, to have the robots observe, read, learn. We are taking baby-steps in these directions. However, from practical experience, whenever I actually publish an actual dataset of knowledge, no one EVER downloads it. I can't even get my collaborators to download it -- I have to nag them, repeatedly. The interest in, the thirst for "knowledgebases" is very nearly zero. No one cares. Anyway, even if people were asking for knowledgebases, we have almost zero tools for creating and curating them. I create them by hand and manage them by hand with a motley collection of scripts, all in public github repos. You can help by improving these scripts. --linas On Mon, Oct 30, 2017 at 1:21 PM, Ivan Bludov wrote: > Hello all, > > Can you please clarify for me one ethical points behind OpenCog? > I have heard from the last RISE conference that Hanson's robots are > connected to the cloud, which stores the knowledge gained by the robots and > share this knowledge among them. > > My straight question: is this knowledge base planned to be open or close? > Or to be intellectual property of Hanson robotics? > I mean, you are talking about the openness and open architecture, what is > nice. But what about knowledge base? Who owns that? > Would Hanson robotics be another ki
Re: [opencog-dev] Openness of knowledge-base of Hanson robotics
Hi Ivan, Let me top-post answers to your questions. First, one needs to clarify: "what is a knowledgebase?" There are multiple components that can be called "a knowledgebase", and they are all quite very different. Lets start with the Hanson Robotic Sophia Personality. This is not one knowledgebase, but several: A) A collection of (partly-hand-authored) face and arm animations. Not terribly useful, if you don't have the physical robot. (and there are many robots, with slightly different motors and mechanical details) You CAN also use these on the blender model, without the physical robot. The older animations were open-source, not sure if the current ones are. Based on interactions on the mailing list, exactly ZERO people (outside the HR circle) downloaded and used these. I used to wish that this could become a viable open-source project on its own, with active developers and hobbyists using it. No such luck. Basically, animators don't create animations as open-source projects. Hollywood has instilled a highly proprietary, closed-source, intellectual-property anti-piracy mindset in this class of people. What might it take to change this? HR did hire a professional animation company, and paid oodles of money, to expand the animation set. I don't know if these are public or private. Doesn't much matter, if no one cares. B) A collection of (hand-authored) monologues tailored for various public appearances. Stuff like "Be sure to visit booth B42, where Gadget Corp is displaying their latest gizmo." These are proprietary, but you wouldn't want these anyway. C) general personality. This is scattered across multiple databases. One of them is https://github.com/opencog/loving-ai which is public. Other bits are private. Its her personality, how she reacts to things, the words that fall out of her mouth. D) Knowledge about English and Russian grammar. This is in link-grammar, its public. The robots more or less do not use this subsystem, expect in a few half-baked, half-operational demos that have been given over the years. The demos can and will be improved, but its slow going. E) knowledge about the world. There is no such database at this time. The robot is as dumb as a rock. Its not a whole lot more than a glorified chatbot, at this time. (I actually counted -- the robot has somewhere between 5 and 8 bits of knowledge about the external world. That's "bits", not bytes.) The goal is, of course, to have the robots observe, read, learn. We are taking baby-steps in these directions. However, from practical experience, whenever I actually publish an actual dataset of knowledge, no one EVER downloads it. I can't even get my collaborators to download it -- I have to nag them, repeatedly. The interest in, the thirst for "knowledgebases" is very nearly zero. No one cares. Anyway, even if people were asking for knowledgebases, we have almost zero tools for creating and curating them. I create them by hand and manage them by hand with a motley collection of scripts, all in public github repos. You can help by improving these scripts. --linas On Mon, Oct 30, 2017 at 1:21 PM, Ivan Bludov wrote: > Hello all, > > Can you please clarify for me one ethical points behind OpenCog? > I have heard from the last RISE conference that Hanson's robots are > connected to the cloud, which stores the knowledge gained by the robots and > share this knowledge among them. > > My straight question: is this knowledge base planned to be open or close? > Or to be intellectual property of Hanson robotics? > I mean, you are talking about the openness and open architecture, what is > nice. But what about knowledge base? Who owns that? > Would Hanson robotics be another kind of Google in this case? When every > data, that you uploaded, belongs to Google. > > I understand that this is the way how the companies usually protect their > business from the competitors. And you have done a lot of work to teach > your robots. But for me this would not be a beautiful future, when all > gained knowledge would belong to one company. > > Can you please comment your visions about this? > > Thanks, > Ivan Bludov > > -- > You received this message because you are subscribed to the Google Groups > "opencog" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to opencog+unsubscr...@googlegroups.com. > To post to this group, send email to opencog@googlegroups.com. > Visit this group at https://groups.google.com/group/opencog. > To view this discussion on the web visit https://groups.google.com/d/ > msgid/opencog/bcdbc967-76f9-4dc4-9cd7-0d63e02a3d72%40googlegroups.com. > For more options, visit https://groups.google.com/d/optout. > -- *"The problem is not that artificial intelligence will get too smart and take over the world," computer scientist Pedro Domingos writes, "the problem is that it's too stupid and already has." * -- You received this message because you are subscribed to the
[opencog-dev] Openness of knowledge-base of Hanson robotics
Hello all, Can you please clarify for me one ethical points behind OpenCog? I have heard from the last RISE conference that Hanson's robots are connected to the cloud, which stores the knowledge gained by the robots and share this knowledge among them. My straight question: is this knowledge base planned to be open or close? Or to be intellectual property of Hanson robotics? I mean, you are talking about the openness and open architecture, what is nice. But what about knowledge base? Who owns that? Would Hanson robotics be another kind of Google in this case? When every data, that you uploaded, belongs to Google. I understand that this is the way how the companies usually protect their business from the competitors. And you have done a lot of work to teach your robots. But for me this would not be a beautiful future, when all gained knowledge would belong to one company. Can you please comment your visions about this? Thanks, Ivan Bludov -- You received this message because you are subscribed to the Google Groups "opencog" group. To unsubscribe from this group and stop receiving emails from it, send an email to opencog+unsubscr...@googlegroups.com. To post to this group, send email to opencog@googlegroups.com. Visit this group at https://groups.google.com/group/opencog. To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/bcdbc967-76f9-4dc4-9cd7-0d63e02a3d72%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.