" to develop AGI, one must first become it - somehow. " -- What does "becoming an AGI" mean to you?
On Fri, Jan 1, 2021 at 10:18 AM Nanograte Knowledge Technologies < [email protected]> wrote: > I appreciate your response. > > A purely, policy-driven AGI may satisfy some of those requirements. It > depends on the extent of operational control that was desired. Policy > management seems central to measurable, human intelligence. > > If I recall, the Component Architecture Management Framework (CAMF), which > I withdrew from IEEE publication in 2009, clearly indicated how intent - as > a theme - translated across various architectural boundaries of a > fully-recursive value chain. In this sense, 'Intent' would correlate with > policy management. > > Note: Not punting. Seriously not. My research interest is now focussed > elsewhere. > > https://www.researchgate.net/publication/281150415_Project_Success_and_the_Component_Architecture_Management_Framework_CAMF > > <https://www.researchgate.net/publication/281150415_Project_Success_and_the_Component_Architecture_Management_Framework_CAMF> > (PDF) Project Success and the Component Architecture Management Framework > (CAMF) - ResearchGate > <https://www.researchgate.net/publication/281150415_Project_Success_and_the_Component_Architecture_Management_Framework_CAMF> > COVID-19 presents as a sufficiently-complex case for the efficacy of a > regenerative pattern language to be tested against. Using public, COVID-19 > information, a regenerative pattern method would ... > www.researchgate.net > While not AGI enabling, it was developed with AGI in mind. > > My philosophical argument remains; to develop AGI, one must first become > it - somehow. There's a significant, cultural threat in that. > > I think, what might be needed is a fully-resourced tribe (perhaps called > AGIans), a tribe existing in its own dedicated space, bearing its own > nationality and passport. A tribe having its cultural and societal roots > founded on evolving AGIan adaptation. Perhaps, such a tribe already exists > somewhere? > > Indeed, the PRC could readily achieve such a tribe. However, so could any > visionary country, supertech, or significant investor group. The question I > have is this: "Why is this apparently not being done?" > > May it then be that the AGI race would eventually be won by the > most-intelligent nation after all? > > > ------------------------------ > *From:* Ben Goertzel <[email protected]> > *Sent:* Friday, 01 January 2021 21:13 > *To:* AGI <[email protected]> > *Subject:* Re: [agi] CCP as a model for AGI > > *** > However, by virtue of the anecdotal evidence I observed in this > context, I'm convinced the PRC are probably pursuing a version of AGI > with utmost urgency > *** > > It could be. I'm just saying, I looked for evidence for that > moderately hard at the top Chinese tech companies and universities and > didn't find it.... > > Should they want to turn that direction, they certainly have > tremendous resources with which to do so. But if they do, you can be > sure they will be seeking some species of AGI that has very clearly > defined and effective control levers. My suspicion (OK more than a > suspicion) is that seeking to create AGIs that are controllable in a > simple sense will be a factor slowing down a party's AGI progress. > > I would make the same argument in regard to US military and also big > Western tech companies, though w/ the latter to a moderately lesser > degree. All of these will want to create AGIs that have simply > defined, clearly tractable control levers relative to their > organizational goals. But there is very likely a tradeoff here btw > simplistic controllability and creative general intelligence. > > It may be those who are willing to give their AGIs more liberty to > self-organize (which does not necessarily mean bypassing human values > btw) will have sufficient advantage in the AGI race -- due to the > power of free self-organization for intelligence -- to overcome the > material-resources advantage of those seeking more controllable AGIs > > -- Ben > > On Fri, Jan 1, 2021 at 8:39 AM Nanograte Knowledge Technologies > <[email protected]> wrote: > > > > We're not discussing Chinese politics, but aspirations, specifically > their possible interest in winning the "AGI" race. We're also not running a > one-liner tit-for-tat campaign. Bear with me, if you will. Our arguments > require more than just a few, terse comments. > > > > Agreed, I'm raising a politics-of-economics argument, but weighted more > towards economic colonization than Communist politics. > > > > As such, the context of this discussion would then refer to some version > of "AGI" as a tool for the PRC to be advancing a globalization strategy > with. > > > > The Romans demonstrated how an empire only needed to be established > administratively - with a hold over local chieftains - for all resources > within that domain to be channelled at the behest of Rome. > > > > I think it would be considered unintelligent to try and "own" all > colonized peoples, when instead one could get the same benefit by simply > turning them into resources. > > > > The IMF demonstrated how their version of democracy could be enforced > via loaded loans. The PRC have followed suit. > > > > One example being masquerading as an innocent Congolese company who won > a national tender for iron ore in Australia. Once the Aussies woke up, more > than 1 billion tons of iron ore had landed in the PRC. It's fair to say > your Africa argument is pointless. The hard facts on the ground to the > contrary are simply overwhelming. > > > > While perhaps not copying the aforementioned globalization models > exactly, the PRC definitely exhibits tendencies to colonize globally. > Arterial-infrastructure-for-unpaid-loans is a dead giveaway. These are the > dominant patterns of deals the PRC have been forging with 3rd-world > countries for the past, 5 years. Always the airports, railways, ports, and > key weapons sites. > > > > While various globalization models are in use today, they all share > similar heuristics. It is synonymous with empire building, by any name. > > > > The PRC announced, circa 1998, that their boundary of influence and > method would in future strategy have no borders, as unrestricted warfare. > > > > In doing so, the PRC declared the world their theatre of operations, > using any method they chose to use. > > > > In addition, the PRC just announced how 2020 was a huge success for > them. I wonder why? > > > > The 2008/2009 post-economic crash radicalized the PRC's business > practices, and produced a razor-blade of intent. > > > > At the time, the PRC, as subject state, kicked in and many private > businesses went under while chanelling resources exclusively to the rescue > of the state. It worked, and a loss of over $2 trillion was mitigated - at > great cost to the private sector. > > > > Afterwards, a new aggressiveness to be number 1 in the world became > apparent. For example, the PRC started hoarding US dollars, building up to > having in excess stock to the value exceeding the total US debt to GDP. > > > > To be absolutely clear. I'm not referring to mainland China culture. I'm > not referring to the Chinese people, but to the CCP as the PRC and their > global strategy. Therefore, I'm not arguing about Chinese culture, but the > 300-year strategic plans of an empire on the march. > > > > Perhaps then, what you perceived as being a "fear" or "unreadiness" of > the PRC to pursue AGI, might've been ancient prudence. Why be the early > adopters when there are simpler ways to skin the cat? > > > > Still, the PRC know - as do any other country in the world - that if > they cannot nurture their supermassive population, they would be at risk of > not achieving their global objectives for stability and growth. Growth = > Relative stability and it may be achieved by all means possible. By ALL > MEANS possible. > > > > I acknowledge the point on narrow AI. I made a similar point in my prior > message. However, as you well know, the road towards AGI is not a quantum > leap from narrow AI. There are multiple developmental stages in between the > 2 poles. > > > > Further, you know what the technical difference is between a > controllable superAI and AGI. Initially, it's a design choice. As such, AGI > might just be the star to aim for in order to land on the moon. > > > > It seems, superpowers are now competing for 5th-wave technologies, with > R&D towards a 7th-wave. That includes the PRC. The star might well be AGI, > but the moon would be pervasive, adaptive AI (in the sense of near AGI). > The playing field is levelling off. The PRC would never be left behind. > > > > However, physics are making one breakthrough after another, all the > while looking for game changers. It might be, one such a discovery would > soon open the doorway to AGI. > > > > For example, suppose the 4 forces were unified? With quantum computers, > incredible force fields could be released with a workable theory of the > cosmos. The hadron collider could fulfill it's potential. > > > > AGI seems a likely testbed for the 5th state of matter, perhaps a future > 6th state, and even a 7th state. > > > > All I'm saying is that the scientific boundaries are now tumbling very > fast. Never say never. > > > > It looks to me as if you have a motivated sense of having to argue as > you do, to deny the PRC's having any such ambitions. > > > > Therefore, for the lack of evidence to the contrary, both our arguments > are equally valid and only time would prove their reliability. > > > > Alternatively, one may choose to understand the development of this > context and view these artifacts as building blocks within a strategic > architecture, as a system. Then, to simply read the results off the outputs. > > > > On the weight of it, one may be pursuaded, or at least left considering > its plausibility. > > > > The whole is greater than the sum of its parts. Again, if I were a > betting man, I'd strongly suggest keeping an eye out for a few, significant > surprises in the superAI category in networks directly beneficial to the > PRC. > > > > I'm not saying the PRC are actively developing near-AGI on home soil, > even if they may find application for its "stages" of product development. > > > > However, by virtue of the anecdotal evidence I observed in this context, > I'm convinced the PRC are probably pursuing a version of AGI with utmost > urgency. > > > > ________________________________ > > From: Ben Goertzel <[email protected]> > > Sent: Friday, 01 January 2021 10:45 > > To: AGI <[email protected]> > > Subject: Re: [agi] CCP as a model for AGI > > > > I don't have time to embark on an in-depth discussion of Chinese > > politics and history etc., though I do know something about it (not > > just due to living in HK for 9 yrs, but also my wife is a mainland > > Chinese AI PhD, etc.).... But I want to note a couple points, > > > > -- China has never sought global empire, they have sought empire > > within their borders and periphery. In line with this, they don't > > want to take over Africa, they want to get good deals on its natural > > resources so as to make China richer. Etc. > > > > -- Chinese culture habitually values stability over all else. AGI > > scares the Chinese gov't and that's part of why there's no Chinese > > analogue of Deep Mind or OpenAI at this point. They want AI they can > > control, and they are smart enough to know that AGI will not likely be > > controllable in any simple way. What they want is an armamentarium > > of narrow AIs they can use to achieve their various purposes in > > determinate ways. > > > > I have met with the AGI-oriented teams within Baidu, Tencent, Alibaba > > etc. as well as the top AI profs at Tsinghua, Fudan, BeiDa and other > > unis there.... Of course one can always hypothesize super-top-secret > > projects but without evidence this isn't so interesting to me.... Of > > course they have their top-secret projects in cybersecurity, hacking, > > weapons systems etc. but the pattern of recruiting/hiring w/in China > > does not support the hypothesis of a major top-secret AGI project. > > > > -- ben > > > > On Thu, Dec 31, 2020 at 2:21 AM Nanograte Knowledge Technologies > > <[email protected]> wrote: > > > > > > Why develop your own AGI when you can wait and eventually take it for > free? > > > > > > Have a look at what China is investing in heavily: communications > technologies and the space race, currency, natural resources, independent > Internet/Web technologies, 5th-gen weaponry, AI-enabled transport, > supermassive production (build, food, water, and now moving to regen > power), and so on. > > > > > > What does that tell us? Either nothing, or that China is preparing a > specific environment for it to be able to push suddenly and effectively > into competitive territories. China only play to win, not to share. I think > that's what they have in common with the USA and Russia. > > > > > > Ben, for AGI, it must be that your direct, or indirect contribution to > their future capabilities speak for itself? > > > > > > China's scaled activities resemble strategies of empire building, not > dissimilar to the Han and Ming dynasties. > > > > > > In further support, more competitive indicators for my argument: > > > > > > China produces 1 million engineers per annum (they dominate the > high-end labor force). China have 2 million free, skilled to highly-skilled > laborers in prison (lowest-cost labor), China's GDP is growing while the > rest of the world are mostly stagnating (more liquidity). China have > acquired masses of land in many continents and are actively colonizing > those areas with loyalists and officials (space of influence where to > establish colonies). Chinese children make up an estimated 90%+ of all > online English tutoring (preparing to be in the West). China won the > national health service contract for the AU, the whole African continent - > together with France (securing medical influence of the largest continent - > the breadbasket of the world). > > > > > > Furthermore, China have numerous young-adult citizens working as > professional and students on key projects in the West, and co publishing. > China have many, exact replicas of Western towns/city areas, assimilating > choice citizens. China's on a mission to be the first in the world, to > build the biggest, to be the smartest. I think they are succeeding. > > > > > > What does that tell you then? Is China going to compete, or are they > already in advanced phases of globalization? Are these symbols of a nation > unsure of its capabilities, bedding itself down to wait things out, or of a > nation on the march? > > > > > > Ask Taiwan, Australia, Japan, America and Europe then, consider the > South and East China Sea confrontations, growing incidents of war, and > these might agree with China's most-aggressive advancing tactics. China > now, seems ready to advance, while the West are reeling. > > > > > > To imagine and suggest that China are not going to effectively compete > in the AGI space is implausible. They are already, and I think they are > just doing it very smartly, most unobtrusively, and in a highly-militarized > fashion. > > > > > > China are collaborating with many 3rd-party countries on emerging > technologies, smart countries, by all accounts. They have obtained deals > for new-gen nuclear plants in many countries, Namibia being the latest. > > > > > > Over the past, 5 years, there's been a trove of white papers and > hi-tech patents blossoming from China. I was looking for some of these > products, but they are not on the market yet. > > > > > > I predict we would soon see a massive unveiling of Chinese-owned > general AI technologies, as well as hordes of narrow AI applications. One > thing is for sure; where the West misses a step, China almost certainly > steps in. > > > > > > Are Chinese persons more intelligent than their competitors, than > non-Chinese? It's still too early to tell. > > > > > > Can the CCP be studied as a model for AGI? Given the hi-tech impetus > obtained from the covid-pandemic, to my mind they are appearing to be > rapidly moving towards becoming the first nation with a citified > singularity. Certainly, we have much to learn from the Chinese. > > > > > > > > > ________________________________ > > > From: Ben Goertzel <[email protected]> > > > Sent: Thursday, 31 December 2020 11:19 > > > To: AGI <[email protected]> > > > Subject: Re: [agi] CCP as a model for AGI > > > > > > > > > I don't think it's very useful to model complex systems like major > nations as one-dimensional utility-maximizers. Asking "whose utility > function" about a complex system of that nature -- which has a large number > of shiftingly-weighted, imprecisely-and-shiftingly-defined "objectives" and > also largely self-organizes in a non-goal-directed way -- is probably the > wrong framing.... But asking who will exert a more major influence (e.g. > the West versus China, or corporate shareholders vs. the scientific > community) certainly has meaning.... > > > > > > And I don't currently see evidence that China will exert more > influence on AGI than the West. Things could evolve that way. But I note > there is not yet a China analogue of Deep Mind or OpenAI, let alone say > OpenCog or SingularityNET or whatever. OpenNARS is founded by Pei Wang, > who is mainland Chinese originally, but is centered in the West, etc. > > > > > > I truly don't understand why folks believe the Chinese gov't is going > to be able to assimilate the US to its goals and thus achieve a dominant > role in shaping AGI .... China does have a larger population than the US > and has an extraordinary capability for mass-manufacture of electronics, > and plenty of other interesting advantages, but the AGI advantage seems > clearly to US/UK ... > > > > > > I'd like to understand if there are better arguments though... > > > > > > ben > > > > > > ben > > > > > > On Wed, Dec 30, 2020 at 8:58 AM James Bowery <[email protected]> > wrote: > > > > > > > > > > > > On Wed, Dec 30, 2020 at 12:17 PM Ben Goertzel <[email protected]> > wrote: > > > > > > > > > Regarding the CCP as a general intelligence(*), I would say all > societies and large corporations can be viewed that way, but I don't see > evidence that the corporate-government complex of China is more generally > intelligent than the corporate-government complexes of US or Western > Europe. What is the evidence or argument in that regard?... > > > > > > > > > If the CCP is more capable of assimilating ("Turking") the US to the > CCP's utility function than vis versa then any claims as to the US being > "more generally intelligent" become superfluous. That's what I meant when > I said: > > > > The CCP-as-AGI is more capable of "Turking" the US-as-AGI than is > the US-as-AGI of "Turking" the CCP-as-AGI. > > > > > > > > > (*) to me calling a country or corporation an "AGI" feels needlessly > confusing, since these are systems largely composed of humans, and not > engineered from human parts but evolved from human social interactions. > But whatever, I understand what is meant. > > > > > > > > > The Future of Humanity Institute is an exemplar for why the question > of "Whose utility function?" cannot be swept under the rug with regards to > "systems largely composed of humans...evolved from human social > interactions". Indeed "artificial" means humans had agency in the creation > of the artifact. The concern of "Friendly Artificial General Intelligence" > hence "The Future of Humanity" is all about the proper application of that > agency in selecting the utility function of aid artifact. What future is > there for "humanity" under the wrong utility function of _any_ notion of > AGI? > > > > > > > > > > > > On Wed, Dec 30, 2020 at 6:54 AM James Bowery <[email protected]> > wrote: > > > > > > As with "AI debates" in general, people can easily talk past each > other by failing to acknowledge they are addressing different questions. > Ben Goertzel is addressing China's in/ability to create an "AGI" in the > sense of Legg, Hutter, et al. Steve Richfield is positing the CCP _is_ an > "AGI" in a more vague sense that might, if "black boxed" also fit with > "AGI" in the sense of Legg, Hutter, et al. Now, it may certainly be argued > that _if_ Steve is right, _then_ it is capable of _creating_ an AGI: "The > Singularity" occurs when some AI achieves the ability to create a more > intelligent AI, and this threshold of "AI" is the most general notion of > AGI. > > > > > > My approach, respecting Steve's original question, is from a position > that what we call "The Global Economy" _is_ an AGI that is already > operating with an "unfriendly" utility function, seeing individual human > beings as raw materials in its environment to refine into "Mechanical > Turks". The only extent to which human quality of life, or even the > quality of the biosphere, is relevant to this AGI is the extent to which it > can provide resources to replicate its incorporations (corporations/NGOs, > governments, etc.) wielding hive-like power over, and ultimately > disintermediating life in seeking access to energy and matter. The CCP is > merely among the more conspicuous cases of evolution toward such an > incipient AGI hive incorporation. > > > > > > Now, having clarified the question I am addressing (Steve's in the OP): > > > > > > Hive specialization in eusocial species recapitulates, in a less > effective way, the clone-army specialization seen in sexual organism > stem-cell differentiation (modulating SC clone gene expression) into > various organs of the organism. The brain is an organ. The CCP constructs > its "brain" not so much by altering gene expression of clones but by > utilizing its long history of civil service examination to mine the > population for "neurons". THAT is where the math comes in to compare the > CCP to the US government's intelligence agencies. Having said that, Ben is > correct that the CCP's structure is more amenable to this mining operation, > and one should see the "private sector" coddled by the CCP as an updated > form of its civil service examination tradition. While it may be true that > the resulting "brain" is not going to be as capable of producing a silicon > AGI as the US, this misses Steve's, or at least my point: > > > > > > The CCP-as-AGI is more capable of "Turking" the US-as-AGI than is the > US-as-AGI of "Turking" the CCP-as-AGI. > > > > > > Why do I say this? > > > > > > See my prior post describing all the ways the US has inhibited its own > intelligence agencies from mining the population for intelligence that > those intelligence agencies can "Turk". Indeed, it is my working > hypothesis that this inhibition was the result of the CCP engaging in the > _real_ "Unrestricted Warfare" that the document by that name represents as > something far more benign. > > > > > > On Wed, Dec 30, 2020 at 5:03 AM Ben Goertzel <[email protected]> wrote: > > > > > > > > > I don't think China's slightly higher average IQ is a big advantage > for them... > > > > > > However, their governmental organization obviously has some practical > advantages. As one example, they can get their intel/ military work done > directly within their big internet tech companies, rather than via sluggish > military contractors and limited-scope awkward back-channel-ish alliances > with big internet tech companies like happens in the US. This means they > are getting on average cleverer and harder working folks working on their > gov't oriented tech, not due to IQ issues but due to organizational > issues... > > > > > > On the other hand they continue to have deep problems with radical > technical innovation due to a persistent culture of mistrust, and this will > cause them real issues, because there are significant differences btw US > and China contexts and copying/adapting Western innovations will probably > not allow them to overtake the West technologically... > > > > > > I predict AGI will emerge first via organizations that are centered in > the West, and China will then attempt to copy it, but will not be fast > enough ... because the org that first creates AGI will be very fast-moving > and agile and not that easy for creativity-phobic Chinese institutions to > catch up with > > > > > > Note I lived in HK for 9 yrs and made many dozens of trips to Beijing, > Shanghai, Xiamen etc. etc. ... I have met w/ folks at the highest levels in > Chinese tech companies and SOEs and fairly high up in gov't. There is a > lot to admire and a lot to fear there, but I don't think China is really in > the race as regards AGI and nor do they have the capacity to extremely > rapidly play catch-up > > > > > > Of course all this could change in 10 yrs, so these comments are most > relevant if AGI is achieved in the next say 7 yrs... > > > > > > ben > > > > > > On Tue, Dec 29, 2020 at 4:22 PM James Bowery <[email protected]> > wrote: > > > > > > It's "Unrestricted Warfare" and as I've pointed out on numerous > occasions, that document strikes me as a limited hangout disinformation. > Keep in mind the Chinese have a higher average IQ than Europeans, their > population is several times larger and they have a _very_ long history of > civil service examinations. Extrapolate that mean advantage out to the > high IQ tail where the ratios explode and it's hard to imagine how great an > advantage they have when it comes to "peacetime" strategy. Add to that the > belly-full of the West with Sassoon's steamships delivering opium and Mao > calling it "a century of humiliation"... > > > > > > On Tue, Dec 29, 2020 at 6:48 PM Steve Richfield < > [email protected]> wrote: > > > > > > As you are reading this, doing the best you can to survive the > Pandemic, consider... > > > > > > The Chinese Communist Party (CCP) is a pretty good model for AGI, as > there are ~500 people working together to provide the best possible > management for China as it attempts to interact as well as possible with > the rest of the world. A rising tide usually floats all boats, but China > perceived an advantage to restrict information about COVID-19 to inflict it > on the rest of the world, which is consistent with their internal manual > Unconventional Warfare, which details LOTS of dirty tricks you might expect > an AGI to employ as it seeks its goals. This manual is a REALLY scary read. > > > > > > Why would anyone expect an AGI to be any "friendlier" than the CCP? > Why wouldn't anyone expect an AGI to be even nastier? > > > > > > This dirty deed WILL work for the CCP - unless worldwide revulsion > costs the CCP even more. I doubt whether an AGI would greatly consider > feelings that run counter to profit. We may all be paying dearly for not > reigning in the CCP long ago - and we might end up paying more if we turn > an AGI loose on the world - for exactly the SAME reasons. > > > > > > Thoughts? > > > > > > Steve Richfield > > > > > > > > > Virus-free. www.avg.com > > > > > > > > > > > > -- > > > Ben Goertzel, PhD > > > http://goertzel.org > > > > > > “Words exist because of meaning; once you've got the meaning you can > forget the words. How can we build an AGI who will forget words so I can > have a word with him?” -- Zhuangzhi++ > > > > > > > > > > > > -- > > > Ben Goertzel, PhD > > > http://goertzel.org > > > > > > “Words exist because of meaning; once you've got the meaning you can > forget the words. How can we build an AGI who will forget words so I can > have a word with him?” -- Zhuangzhi++ > > > > > > > > > > > > -- > > > Ben Goertzel, PhD > > > http://goertzel.org > > > > > > “Words exist because of meaning; once you've got the meaning you can > forget the words. How can we build an AGI who will forget words so I can > have a word with him?” -- Zhuangzhi++ > > > Artificial General Intelligence List / AGI / see discussions + > participants + delivery options Permalink > > > > > > -- > > Ben Goertzel, PhD > > http://goertzel.org > > > > “Words exist because of meaning; once you've got the meaning you can > > forget the words. How can we build an AGI who will forget words so I > > can have a word with him?” -- Zhuangzhi++ > > Artificial General Intelligence List / AGI / see discussions + > participants + delivery options Permalink > > -- > Ben Goertzel, PhD > http://goertzel.org > > “Words exist because of meaning; once you've got the meaning you can > forget the words. How can we build an AGI who will forget words so I > can have a word with him?” -- Zhuangzhi++ > *Artificial General Intelligence List <https://agi.topicbox.com/latest>* > / AGI / see discussions <https://agi.topicbox.com/groups/agi> + > participants <https://agi.topicbox.com/groups/agi/members> + delivery > options <https://agi.topicbox.com/groups/agi/subscription> Permalink > <https://agi.topicbox.com/groups/agi/Tf065676fd779dd5c-Ma86610acfc10e70c83bc366c> -- Ben Goertzel, PhD http://goertzel.org “Words exist because of meaning; once you've got the meaning you can forget the words. How can we build an AGI who will forget words so I can have a word with him?” -- Zhuangzhi++ ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Tf065676fd779dd5c-M1dc5e1b158efb8734c60ec3d Delivery options: https://agi.topicbox.com/groups/agi/subscription
