intro’d to Ben via top documentary site ‘s
singularity (2009) 48 min
What is singularity? A future period during which the pace of technological innovation will be so rapid, its impact so deep, that human life will be irreversibly transformed. Although neither utopian nor dystopian, this epoch will transform the concepts that we rely on to give meaning to our lives, from our modes of commerce to the cycles of human and non-human life, including death itself.
b: 10 yrs to a positive singularity.. *if the right amount of effort is expended in the right direction.. if we really really try
i say we could do it in 1-2 .. if the right amount of effort is expended in the right direction.. ie: disengage from irrelevant s
2 min – b: i focus mainly on the path to singularity via ai.. because that’s where my own work is.. i believe i know how to create an artificial gen intelligence.. that can lead us to a positive singularity.. but there’s a lot of other interesting work going on………any one of these paths could lead to a positive singularity way way sooner than most people think…. but the sad thing is.. the amount of resources/energy that our society devotes to these things is very very small
3 min – the human race is at a funny stage in our evolution.. we have spent far more energy/attention on things like making chocolatier chocolates.. or more attractive underpants.. then we do on creating new forms of matter.. improving human cognition.. extending human life.. ending scarcity.. ending human suffering.. creating advance artificial minds..
p 1: hong kong
5 min – on agi vs narrow ai
6 min – the human mind is complex and not that well understood.. so i think we have to draw info about it from every source we have.. brain scanning only tells us so much.. computer sci only tells us so much… *there’s a lot that can be gotten by introspecting..by observing own thoughts..and that’s a very structured discipline as well
imagine everything we really need/crave.. we can get thru introspecting.. ie: self-talk as data
7 min – people think it’s just a casual thing of looking in your own mind.. but there’s a number of traditions that have really refined it in a systematic way..
there’s some things you can’t get that way.. on the other hand.. right now.. there’s a lot of things in our mind.. we can only get through introspection.. a central processes of cognition/reflection/understanding
my grandmother and grandfather wrote a book called..cradles of eminence.. where they studied the childhoods of famous people.. one thing they found.. the great scientists almost always had very lonely childhoods.. no friends.. isolated.. many sick in their adolescence
8 min – i grew up in oregon.. early 1970s.. surrounded by peace loving.. hippie guys.. tossing around weird ideas… started college when i was 15.. studied mathematics.. until i was about 27-28 i was doing research… when i started my career as a scientist i was interested in a bunch of big problems.. spacecraft/time-machines/theoretical-physics/simulating-immune-systems/artificial-life/people-live-forever/thinking-machines…i figured i could solve all these at one time.. but that proved significantly more difficult than anticipated.. so .. late 20s.. focused 80% of my time.. on one thing.. to make thinking machines..
9 min – at conferences et al.. most ai scientists are materialistic in their outlook..most feel strongly that human consciousness comes about as a consequence of certain complex structures in the human brain… that’s one respected philosophy of human consciousness… it isn’t mine.. so when i’m around ai people discussing consciousness.. i’m almost always the outlier…
10 min – in the view i take.. consciousness/awareness is the ground.. it’s the base at which everything is formed.. and then diff structures manifest consciousness in diff ways.. so.. our human brain manifests universal consciousness in one way and this coin manifest universal consciousness in a diff way… from that point of view.. the question isn’t which systems are conscious.. the question is.. what kind of consciousness each system demonstrates.. how it manifests universal consciousness.. that’s a very eccentric view for an ai scientist and i’m usually in the minority of one if around ai people.. on the other hand.. if around buddhists or people from other analogous spiritual traditions.. they just take this for granted
11 min – i read some interviews between the dalai lama and a number of scientists.. and some of them asked him.. about ai and consciousness… can machines ever truly think.. truly feel/understand.. be self aware.. and he said in a sort of tongue and cheek way .. that perhaps one day when programmers die .. they will be reincarnated in ai programs..
12 min – from hong kong to xiamen china .. working with hugo on conscious robot project.. hugo works on genetic algo’s.. and i’ve worked more on adv cognition and learning systems.. so
p 2: xiamen
h: there’s tons of money here. the opps here are just amazing.. we’re hoping to tap into some of that money to make xiamen an ai center.. for china..and if we can attract enough foreign.. beyond chinese.. talent.. maybe even the world
h: to build.. for future industry.. massive industry.. that’s why i’m here
h: i voted to come here.. because i see china winning… my feeling is .. america peaked in the 60s… i see china on the rise and china seems willing to foot the bill
18 min – b: all the stuff we’re doing with virtual dogs now.. learning by imitation/reinforcement.. we want to plug that into these robots
20 min – b: the hard part of building ai.. is knowing what to program.. knowing how to make the software..understanding how the mind works
21 min – b: most of the time.. 20 yrs.. since my phd…i’ve spent.. just understanding cognition.. creating mathematical computational models.. of how thinking works..
kind of the same.. detox..
b: of course there’s a lot of detail here.. but at bottom.. i believe the mind is a patter recognition system.. a mind recognizes patterns in the world around it… patterns in its own actions….patterns regarding which of its own actions will lead to which outcomes.. most critically.. a mind recognizes patterns in itself…. the self each of us carries around… the self i have in my mind when i think of ben goertzel.. this self is itself a pattern that the mind is recognized.. creating an ai is about creating a pattern recognition system to allow the ai to recognize the patterns in the world and in itself.. that needs to demo a very high level of intelligence
24 min – b: asking robot.. am i a robot.. robot: i think a lot of people are
26 mi – b: one question i get asked a lot.. why do this.. why not be content with life the way it is.. *the way nature has given it to us.. the natural order has great beauty.. there’s also a lot of suffering in human life
*we’re so not there…like robot said.. we’re turned ourselves into robots..
27 min – b: i love human life.. i love the natural order of things.. i don’t love death/disease.. i don’t love all the limitations on my mind/life that are placed on my by the human body/mind.. by the limitations i was born with..we’re stuck in one body.. ai can have as many bodies as they want…*we don’t have to think about ai minds as imitations of the human mind.. the scope of the ai mind can be much much broader than that..
*yes.. ai and human.. two diff realms.. and right now.. we’re limiting them both with this imitation business
28 min – b: people have this idea that computers/robots.. somehow have to be cold/impersonal/uncaring.. i don’t see why that’s true.. i believe we can build an artificial minds that are more caring/compassionate.. better and more stable goal systems..than any human being.. once these machines can think better than us.. all these other techs will develop faster and faster.. instead of humans developing micro/nano/brain scanners.. artificial organism.. super human minds will be doing this science.. at that point.. human beings are no longer in control.. it’s a scary thought.. but it’s also an exciting one.. one thing is for certain.. it’s not going to be boring..
29 min – b: my research collaborator.. hugo also sees a tech singularity coming.. but he sees outcome a little differently
h: my ai/brain lab.. in a sense.. we are the problem.. because what are we doing.. we’re creating artificial brains that will get smarter and smarter every year
30 min – h: the most probably scenario.. in my view.. is probably the worst..
31 min – h: cosmists (want to build god like machines) vs terrans (main motive .. fear)…. to me this artilect wall is not between humans and machines.. it’s between humans and humans anticipating that if they don’t do anything.. then it really would be between humans and machines.. so in a sense.. the terminator movies are educating the terrans
32 min – h: so if.. if.. there’s a major war.. the most passionate war we’ve ever known.. with late 21st cent weaponry.. then we’re not talking millions getting killed.. like 20th cent.. it will be in the billions killed
33 min – h: i calculate my grandkids who will be caught up in this and i say.. thank god i won’t see it..
h: it’s a binary decision.. not fuzzy.. you build them or you don’t.. so everyone has to choose.. and i chose cosmist.. fully conscious that the price of that choice is ultimately.. maybe humanity gets wiped out
35 min – b: of course i want to help people… i don’t want harm to come to anyone.. i want good things for myself.. my children..i think i have the same desires for good things for humanity as everyone else.. but i don’t pretend to know what’s going to happen.. even though i know what i would like to happen
36 min – b: this is just going to be a completely revolutionary break from anything that came before.. i don’t think it’s comparable to fire/guns/writing/prior-rev-changes… this is the notion of the singularity… point beyond which we can’t predict what’s going to happen.. i think we can predict up to.. but not after singularity
37 min – i’m doing what i’m doing.. what feels natural for me to do.. i don’t think the first humans to use language.. that was a big change.. and the first people .. did not prognosticate everything that could come out of it.. i’m sure they could see good/bad aspects.. and personal gut intuition it was the right thing to do.. and i think most of us working on agi have that same gut instinct.. i think agi is not that diff from other pursuits.. driven by complex individual and social motivations…
38 min – the human race as a whole is creating super human intelligence.. we’re all a part of that.. and the stories we tell ourselves about our own motives.. are all a part of it..
p 3: home
b: i find working w chinese.. they definitely have a diff attitude on things
43 min – b: i feel i’m participating in the human condition.. married.. 3 kids.. i hike/backpack.. i feel i want to go beyond the human condition.. striking balance between participating in and going beyond is a tricky thing
44 min – before kids.. i wanted to be immortal.. and now.. having kids.. has reduced my fear of death .. dramatically..
My talk at the Global Leaders Forum in Seoul, on AGI and robotics,
4 min – the ai gaining the most traction today is what i would call narrow ai.. can do one there very well.. not a bad thing for the world.. but not same as a general ai.. that can think about new ideas… *imagination.. like humans can
although humans can be stupid in some contexts.. we can think/reason about things that didn’t exist before..
(we were born.. went to school.. ie: no internet before i was born.. no mobile phones…)
..because we can generalize
new topic.. we understand how that topic relates to the world at large.. and to ourselves.. and we can (im)port our previous knowledge to the new topic..
so too.. we can imagine.. isn’t that beyond importing previous knowledge to new topic..?
because we have a type of general intelligence that current ai systems don’t have..
5 min – in context of alpha go – what if you wanted alpha go to …. if change shape of board.. alpha go can’t do it.. because deep in alpha go is a repetition of the .. 19×19 square array… whereas.. human.. w a funny shaped go board.. we’re gonna adapt… generalize to new situation… of course you could modify the code on alpha go to play on diff shaped boards.. but then again.. that’s human being doing that generalization
7 min – in general.. what we need is a system that can recognize patterns in its environment and in itself.. *including patterns about what actions it should take.. get what things done in what situations..
*great for ai.. but we can’t limit human beings to this.. as ai.. it can augment humans.. but this is a limitation (supposed to ness) of humanity.. we can’t go there.. well.. we are there.. we need back out.. not deeper in..
and has to do this efficiently given limited resources in a complex ever changing environment
this isn’t an easy thing.. but you know.. building smart phones is not an easy thing… we’ve built many many complex things before.. and i and many others believe that now is the time for building the complex thing that is general intelligence.. in next 5-10 yrs
8 min – the preconditions are here.. we have computing hardware and software that has advanced tremendously… advanced cognitive science to understand parts of human mind.. more knowledge of neuroscience… and a huge amount of algo’s and representations that came out of narrow ai…. because even though narrow ai and agi are different.. there’s still a lot to learn from narrow ai’s tech.. on how to build general intelligence
building agi is complex.. my own detailed ideas about how to do so are contained in the book… engineering gen intelligence..
9 min – deep learning neural nets.. which are very fashionable these days.. i believe.. are only part of the picture..and open cogs’ knowledge representation.. that we’re working in hanson robotics.. our other ai work.. is based on the neural symbolic logic.. we have nodes/links inside the ai’s mind.. some of them are neural net type of links.. some of them are nodes/links representing symbolic logic expressions.. so we do logical inference.. and neural net based pattern recognition.. together in the same network.. we handle many kinds of knowledge together in the same network.. declarative (facts/beliefs); procedural (how to do things); eposodic (about life histories); sensory motor;…. these are handled by diff algo’s . . that work together on the common graph.. neural symbolic representation…
10 min – so we use probabilistic knowledge to do reasoning.. we use deep neural nets to perceptions.. we use evolutionary learning to do creativity.. and we use activation spreading .. like in a typical neural net.. to allocate attention throughout the system.. by putting all these diff things together.. we’re able to make a system that can come to grips with the world as an autonomous agent..
now.. our ai’s are not as smart as human beings.. but we believe *we have a detailed plan to get from where we are now to ais with human level and even greater general intelligence..
what if our *goal/detailed plan .. is instead.. to get to level of greater connectivity/awareness/eudaimonia/peace/love/humanity
kurzweil on record saying he believes.. we’ll get there by 2029.. and he is now working for google trying to do this.. and we want to beat them.. we believe we can get there in 9 years by 2025… we have a 9 yr plan broken in to 3 stages.. and we’re currently focusing on the first stage.. which is to get the hanson robot to have basic sentience (feeling)… i’d like to have robot.. up on stage.. answering questions just like everyone else..
ok.. cool.. but what if life isn’t about answering questions on a stage.. (i know that was just an ie..but that seems to be premise.. in mind.. looks like.. more schooling)
doesn’t need to be genius yet.. just basic sentient awareness (get off stage at right time.. realize you don’t need coffee.. rather electricity) of its environment.. who it is .. with what’s around it.. what’s going on.. it may not know how to build a phone.. but if it sees me with this it knows this is a device i use to talk to other people.. and look at pictures..
12 min – once you get there.. basic common sense knowledge.. basic sentience.. then you have a system that you can teach *much like you would teach a young child.. and the system can learn more and more..
*perhaps so.. perhaps the way we teach is more fitting for machine.. it hasn’t boded human/children well.. again .. difference of supposed to’s and whimsy/curiosity/antifragility..
of course won’t be smart enough to turn over the world govt soon but..
but eventually it can then be taught to take on *all the diff jobs that human beings can do
well..*just ones we choose not to do.. right.. or that are too much for us to do.. ie: facilitate the curiosity of 7 bn people everyday..
we’re at the v2 rocket stage.. *we need to build something that fully understands the principle..
a nother way.. let’s *build that..
v2 rocket showed.. you could go into space.. and *once you saw that.. you could go into space.. then people really understood what was possible..
*we need to show that we can build basic sentience.. **a system that really understands what’s going on
or perhaps.. *show that with tech augmenting/facilitating us.. we can be/live free.. **a people really grokking what matters…
and *isn’t on any level faking it or relying on human prepared rules or trending data.. but it’s **learning things about the world from its own experience.
h u g e
13 min – once you get there.. suddenly.. *ai is going to become way way way bigger than it is now.. will be the biggest industry on the earth
how about.. equity/humanity will become way better on the entire earth.. first time in history
my approach to building ai is integrative.. it’s more based on computer science and cognitive science than neuroscience.. not because i think you couldn’t imitate the brain.. but i think the brain is just too poorly understood right now.. to take a pure neuroscience based approach.. i also think that the brain is not that similar to the computing hardware we have now.. so i’m taking a non neuroscience approach in building ai that’s closely adapted to the computer hardware that we have available today..
14 min – this is a diagram of the components of the human brain.. 300 diff regions..with a very complex connectivity architecture.. if you look at current deep neural network.. it’s what underly alpha go.. facebook’s face recognition.. and a lot of things..in essence these are a model of the feedforward activity of the primary visual cortex.. which is a few diff regions of the brain.. and it’s very interesting that we can model a few regions of the brain to get that model to do an approximation of the vision processing that those brain regions do.. and it can be extended a bit.. so .. google deep mind.. which is an amazing company.. their recent work on differential neural computers.. that takes that model visual cortex and connects it to a simple model of the hippocampus.. a diff part of the brain which deals with spatial (mapping?) among other functions.. and that gets a deep neural network to be able to navigate on the map and do other things that deep neural networks can’t do on their own… again .. that’s simplified models of two brain regions.. there’s hundreds of them..
15 min – what ultimately we need to do.. not just seeing/mapping.. we need to be seeing/mapping/reasoning/emotions/short-term-memory/long-term-memory/reinforcement/action/imagination/language.. there’s a lot of diff parts there.. but i believe we can connect these all together..
what i’m seeing here is.. few interactions with robot (telling it to turn right/left).. so this is what we do all day at hanson robotics.. we try to get some of our intelligence into these robots (asked what smart means.. said.. i know defn.. but not experiential of it)
16 min – he doesn’t fully understand everything he’s saying.. but i don’t fully understand everything i’m saying either… he understand some of it .. and some he seams together from what people have said and stuff he found from the internet.. and as we teach him more and more.. more and more understanding..
17 min – showing other robot.. sofia..
18 min – it’s a lot of fun experimenting with these robots.. we can run them in a bunch of diff modes.. we can program the robots with knowledge by the specific domain area.. we can run the robots like a young child.. *where there’s no knowledge in them….or we can feed them with a diversity of knowledge and let it recombine it and see what comes out
? *that’s what we’re missing
w sofia.. what i’d done is filled the robot’s knowledge base with a whole lot of info from the novels of philip k dick.. a science fiction writer .. and then.. you ask a question and it looks in its knowledge from philip k dick’s novels.. and tries to figure out based on info gleaned from novels.. how should i answer… and .. she completely surprised me..ie: reality cannot be detected… that wasn’t something we fed her .. in novels or by us
20 min – from commercial standpoint…. (blah blah).. from research .. we can play with them.. and play is where creativity comes in.. this sort of creative experimentation with ai algo’s and robots based on a *solid gen intelligence architecture… as in open cog.. i think is how we’re going to make the next big great leap toward **thinking machines
thinking rather.. *sci of people.. fake.. (for humans) which is fine architecture for building robots.. so perhaps that will save us.. all we’ve tried to do to humans.. we can now do with robots.. and let humans be human (perhaps all this time we’ve been perfecting how to build a robot.. and now we can get back to being human)
i’d say **thinking.. as much as we’ve (humanity) been thinking over the last many years.. ie: not ie: voluntary compliance ie: fake.. which again is fine for a robot to augment su
21 min – the open cog platform for gen intelligence is as you’d expect.. quite general in its applications.. i think the robotic application is especially interesting in that it’s an ideal way to get common sense knowledge about the human world.. into the ai’s mind.. because a robot is in the human world acting as a human… on other hand.. a lot of other valuable things we can do with this ai system.. ie: we’ve used open cog to analyze dna data for people w parkinson’s disease… and the dna of flies..toward improving aging in human trials.. so now to dna of humans who live to 110+
22 min – we view all these applications as being under the same global mind cloud.. so can have ai controlled: robots; video games; biomedical analysis apps; self driving cars; .. and all are relying on same cloud based ai.. and whatever one of these ai agents learns is then fed into the overall mind cloud.. increasing the global knowledge that any ai can draw on… and that’s ultimately how we’re going to get to the singularity..
23 min – we’re going to take an ai design.. ai architecture.. like open cog.. which models all the key aspects of human intelligence.. we’re going to put this in the diversity of applications.. helping people.. learning from people.. interacting with people… over time.. this ai will learn to help people with every task that we need to do..
24 min – and eventually.. this ai will learn computer programming.. and so it can modify its own software code.. make itself more and more intelligent…
this is a similar vision that my friend ray kurzweil has invented.. i think what he’s doing.. what google’s deep mind is doing.. what folks at vidoo.. and other co’s are doing in this direction is all very exciting/interesting..
personally.. i’m hoping we can get there with our open source open cog.. and grow ai more like linux.. grace roots..
25 min – friends in brussels.. globe as brain.. phones in heads.. where we are now is amazing.. to what we’d have thought 20 yrs ago.. but the next few steps.. during the next 5-10-15-20 years are going to be even more exciting
Ben Goertzel (born December 8, 1966 in Rio de Janeiro, Brazil) is Chief Scientist of financial prediction firm Aidyia Holdings and robotics firm Hanson Robotics; Chairman of AI software company Novamente LLC, which is a privately held software company; Chairman of the Artificial General Intelligence Society and the OpenCog Foundation; Vice Chairman of futurist nonprofit Humanity+; Scientific Advisor of biopharma firm Genescient Corp.; Advisor to the Singularity University; Research Professor in the Fujian Key Lab for Brain-Like Intelligent Systems at Xiamen University, China; and general Chair of the Artificial General Intelligence conference series, an American author and researcher in the field of artificial intelligence. He was the Director of Research of the Machine Intelligence Research Institute (formerly the Singularity Institute).
Goertzel is the son of Ted Goertzel, a former professor of sociology at Rutgers University. He left high school after the tenth grade to attend Bard College at Simon’s Rock, where he graduated with a bachelor’s degree in Quantitative Studies. Goertzel went on to obtain a Ph.D. in mathematics from Temple University in 1989. Before entering the software industry, he served as a university faculty in several departments of mathematics, computer science and cognitive science, including the University of Nevada, City University of New York, the University of Waikato, and the University of Western Australia. He spends most of his time at a residence in the New Territories of Hong Kong.
His research work encompasses artificial general intelligence, natural language processing, cognitive science, data mining, machine learning, computational finance, bioinformatics, virtual worlds and gaming and other areas. He has published a dozen scientific books, more than 100 technical papers, and numerous journalistic articles.
He actively promotes the OpenCog project that he co-founded, which aims to build an open source artificial general intelligence engine. He is focused on creating benevolent superhuman artificial general intelligence; and applying AI to areas like financial prediction, bioinformatics, robotics and gaming.
In 1996 Goertzel together with Francis Heylighen founded the Global Brain Group to study the global brain emerging from an increasingly intelligent Internet, and in 2011 he joined the scientific board of the newly founded Global Brain Institute at the Vrije Universiteit Brussel. In 1997, he left academia, which he said made his life much more interesting, but a lot more stressful. In 1997, he started a software company, and relocated to New York City to launch Intelligenesis Corp. (later known as Webmind Inc.), a company with the mission of creating a truly intelligent AI system and making money along the way by productizing its components.
He defines intelligence as the ability to detect patterns in the world and in the agent itself. He tries to create a “baby-like” artificial intelligence first, and then raise and train this agent in a simulated or virtual world such as Second Life to produce a more powerful intelligence. Knowledge is represented in a network whose nodes and links carry probabilistic truth values as well as “attention values”, with the attention values resembling the weights in a neural network. Several algorithms operate on this network, the central one being a combination of a probabilistic inference engine and a custom version of evolutionary programming. He claimed that this combination is able to avoid the combinatorial explosions that both these algorithms suffer from when exposed to large problems. In 2008, he founded OpenCog, which was an open-source AGI software project.
In 2009, Ben Goertzel and Hugo DeGaris starred in a 45-minute documentary called Singularity or Bust. In 2014, Goertzel appeared on the American science documentary television series, Through the Wormhole, in episode 1 of season 5.
Hugo de Garis (born 1947, Sydney, Australia) is a retired researcher in the sub-field of artificial intelligence (AI) known as evolvable hardware. He became known in the 1990s for his research on the use of genetic algorithms to evolve neural networks using three-dimensional cellular automata inside field programmable gate arrays. He claimed that this approach would enable the creation of what he terms “artificial brains” which would quickly surpass human levels of intelligence.
He has more recently been noted for his belief that a major war between the supporters and opponents of intelligent machines, resulting in billions of deaths, is almost inevitable before the end of the 21st century. He suggests AIs may simply eliminate the human race, and humans would be powerless to stop them because of technological singularity.
The feature-length documentary film The Singularity by independent filmmaker Doug Wolens (released at the end of 2012), showcasing Goertzel’s deep vision and understanding of making general AI general thinking, has been acclaimed as “a large-scale achievement in its documentation of futurist and counter-futurist ideas” and “the best documentary on the Singularity to date.
neuroscience of imagination.. just after taking in Ben (and his thinking/hoping a machine can do imagination.. and me thinking more .. machine for govt/B stuff.. people for imagination).. my first watch of this.. w/o slowly taking it in et al.. does seem pattern/algo ish.. i don’t know
Ben Goertzel (@bengoertzel) tweeted at 6:55 AM – 28 Dec 2016 :
woops, try this link for the Korean TV out-take of me and the Hansonbots —
Recent Advances in AI — my interview with Adam Ford at the Beijing Global Innovators Conference, https://t.co/bqLAruJOPn
Original Tweet: https://twitter.com/bengoertzel/status/848880987375194113
apr 2017 Amazing Progress in Artificial Intelligence:
practically – alpha go
1\ deep mind’s alpha go software is a really a hybrid architecture.. you have deep neural network.. then monte carlo’s game research.. coming together to play go
2\ the core ideas came from some academic papers.. .but it was a tech co that really made the thing work…
deep neural nets around for ever.. i was teaching about them in mid 1990s.. others before that.. again it was the tech co to make the academia programs scale up
5 min – limitations are obvious to people like me in cog sci and cog neuro sci.. and to a lot of deep learning researchers.. but not that open to people reading reports in mass media
recurring problem in deep neural nets.. crude but interesting model of the feed forward activity in the visual and auditory cortex.. so it’s a decent model.. with a lot of changes to adapt the algos to current computer hardware
a decent model.. when you’re seeing/hearing first 1/2 second or so before the cognitive parts of the brain do their more advance/conceptual feedback to the perception process
6 min – what we’re not modeling there is hippocampus w various kinds of working memory and localized memory.. basal ganglia.. it’s influence on goals.. order activity.. cerebellum and sequences.. the thalamus’s interaction with the cortex.. amygdala.. the limbic system that’s the emotions.. and then the dozens of other important brain areas.. which has its own architecture/dynamics.. all of which act in a coordinated way to produce human intelligence
so i was thinking for a while.. the next step for those who like deep neural networks.. should be to take deep net architecture and connect them with other components.. which architecturally reflect other parts of the brain besides visual/auditory cortex in their connectivity structure and dynamics.. and then you would gradually build architectures with more and more diff neuro net components that reflect diff parts of the brain and what they do..
7 min – i said this to a bunch of people.. it’s not a direction i was pursuing because..
my own preferred approach to agi is less closely tied to how the brain works..
Ben Goertzel (@bengoertzel) tweeted at 4:33 AM – 3 Apr 2017 :
The ridiculous nasty mess of the US “suspected illegal immigrant prison system”, https://t.co/PdBAS99SrU(http://twitter.com/bengoertzel/status/848845864713895937?s=17)
“Why We Are Building SingularityNET” by @bengoertzel https://t.co/ARkVA6DDEB
Original Tweet: https://twitter.com/Innovandiamo/status/917841453744869376
Cooperation is still the holy grail of AI
The biggest gains, however, will be experienced by AIs that can learn from each other, leveraging external datasets and capabilities. *There is currently no platform that enables this type of interoperability and coordination. The launch of such a platform would be one of the biggest breakthroughs ever in the evolution of AI.
*currently no platform that enables this type of interop and coord for people.. the launch of such a platform would be **one of the biggest breakthrus ever in the evolution of human beings..
SingularityNET is the first and only platform that allows AIs to cooperate and coordinate at scale. This removes one of the major limiting factors to AI growth today — the lack of interoperability — which severely restricts the ability to leverage the strengths and capabilities of individual AIs.
With SingularityNET, companies, organizations, and individuals can empower their AI agents to participate in markets, buying and selling goods and services.
this seems limiting.. just about money/markets..?
*AIs can use SingularityNET markets to buy rare and valuable datasets or analytical capabilities that previously needed to be developed in-house from scratch. **AIs can also use SingularityNET markets to monetize their own assets, creating massive value from proprietary information, algorithms, and functionality.
*This is a major market opportunity ..
AI Can Help Create a Better World—If We Build it Right https://t.co/MkddBucXaM
Original Tweet: https://twitter.com/singularityhub/status/1029021526039973888
AI Can Help Create a Better World—If We Build it Right https://t.co/MkddBucXaM
Original Tweet: https://twitter.com/singularityhub/status/1029021526039973888
It is true that automation is contributing to the increasing inequality within most countries, with the income gap widening between each country’s educated middle and upper classes and their less-educated lower classes. However, what’s systematically overlooked is that inequality between different countries is decreasing, in large part due to a combination of global trade and advances in computing, communication, shipping, and manufacturing technologies.
monetary ineq is not the root of our problem..
This also means that most of the brilliant AI developers in the world have little opportunity to participate centrally in the rollout and application of AI technology, because they have limited opportunity to work for these tech companies or militaries. As a result, their cultures and voices are not included in the network of emerging AI tech. This is problematic because they are the ones who know best what their countries need.
We are teaching AI to see the world from only one perspective..t
yes.. and it’s grounded in money as os
How do we root culture-specific cognitive biases out of a globally-distributed, emergent general intelligence?..t
focus on cure ios city..
The safest approach, and the one most conducive to a high level of intelligence, creativity, and open-mindedness on the part of the AIs we are creating is for the AI’s process of learning and formation to be as globally inclusive and participatory as possible..t
as inclusive as possible: has to be all of us..
What is needed is creation of mechanisms that educate more people in the developing world with deep technical skills, and then enable them to monetize these skills..t .. to support themselves and their families.
rather.. a mech listen to and facil all the voices.. everyday.. as it could be..
ie: ai as augmenting interconnectedness
mufleh humanity law: we have seen advances in every aspect of our lives except our humanity
about AI and future
Ben Goertzel (@bengoertzel) tweeted at 5:15 AM – 27 Aug 2018 :
My interview with Alexey Turchin on AGI and longevity, at HLAI/AGI-2018 conference in Prague… https://t.co/JPxFVd4nOC(http://twitter.com/bengoertzel/status/1034036755211608064?s=17)
1 min – we don’t know what we need because we haven’t discovered the secret yet.. maybe you don’t need ai at all.. you just need some clever human biologist to discover a few things.. it may be tools w smart biologist provide the secret.. t
3 min – get there in 2-100 yrs.. my gut is that we’ll get there in 10 yrs
sooner if we try a nother way
4 min – there’s a small chance we’re just missing something fundamental about mind and universe.. but i doubt it.. t
7 min – it seems to be beyond human capability to control these dangers..
what it takes to become an agi programmer – 5 min video – aug 2018
agi is quite cross disciplinary and the ed system isn’t really that way.. so what that means .. if someone wants to work in some of these cutting edge topics that has the highest probability of transforming the world, if they want to work on these things in the core capacity, they have to take their own time to study a bunch of other fields that they didn’t learn in school.. and that also takes time.. you can’t do that by reading a blog post
cognitive synergy is a term i produced.. to refer to the process of multiple cognitive processes, acting on a common knowledge representation, interacting
Nice article by Yuval Noah Harari on how AI exacerbates inequality — he gets some major stuff right, but over-idealizes “liberal democracy”.Actually AI is merely amplifying imperialistic and inegalitarian dynamics intrinsic to large-scale capitalism..https://t.co/cc6hFcgqyF
Original Tweet: https://twitter.com/bengoertzel/status/1036258828596039680
In the past, machines competed with humans mainly in manual skills. Now they are beginning to compete with us in cognitive skills. And we don’t know of any third kind of skill—beyond the manual and the cognitive—in which humans will always have an edge…t
1 yr to be 5 et al
What’s more, AI enjoys uniquely nonhuman abilities, which makes the difference between AI and a human worker one of kind rather than merely of degree. Two particularly important nonhuman abilities that AI possesses are connectivity and updatability..t
indeed.. let’s focus on that.. go deeper than traffic, disease, et al ie: augmenting human interconnectedness
mufleh humanity law: we have seen advances in every aspect of our lives except our humanity – Luma Mufleh
People will need to retrain and reinvent themselves not just once, but many times..t
like everyday.. bravery to change mind
Just as in the 20th century governments established massive education systems for young people, in the 21st century they will need to establish massive reeducation systems for adults. But will that be enough? ..t
AI makes it possible to process enormous amounts of information centrally.
we need to place a much higher priority on understanding how the human mind works—particularly how our own wisdom and compassion can be cultivated..t
perhaps rather.. more a focus on listening-to/faciling us than understanding us
ie: as it could be
In the 21st century, data will eclipse both land and machinery as the most important asset, so politics will be a struggle to control data’s flow..the race to accumulate data is already on..t
matters little if data is illegit.. ie: from whales in sea world
time to try self-talk as data
So we had better call upon our scientists, our philosophers, our lawyers, and even our poets to turn their attention to this big question: How do you regulate the ownership of data?..t
the most important contribution you can make is to find ways to prevent too much data from being concentrated in too few hands, and also find ways to keep distributed data processing more efficient than centralized data processing. These will not be easy tasks. But achieving them may be the best safeguard of democracy.
we could do that by getting back to meadows undisturbed ecosystem.. but it would be a best safeguard for humanity rather than democracy (whatever that is)
NY Times writes about How the Blockchain Could Break Big Tech’s Hold on A.I. @singularity_net https://t.co/oZA9C5Z1sL
Original Tweet: https://twitter.com/bengoertzel/status/1053672897057681413
using blockchain as a common izer
SingularityNET (@singularity_net) tweeted at 3:16 AM – 3 Dec 2019 :
The Future of AI – @bengoertzel a world-leading AI scientist who is the CEO and founder of #SingularityNET relays his vision of the future and answers some very astute questions in a fireside chat with the CEO and Founder of @XccelerateSkill Lavine Hemlani https://t.co/Qq4jikpSSHhttps://t.co/82XNtEvGuX (http://twitter.com/singularity_net/status/1201807478784221184?s=17)
1 hr interview
2 min – i tend to trace it (ai) back to 1930s norbert wiener’s book cybernetics where he’s looking at *control and communication and the biology systems and gene systems.. and that was even a couple decades before the term ai was coined in the late 50s.. so these ideas have been around a while
last few years.. an explosion.. now applied in visible/everyday things.. what we need to get to is agi – general intelligence.. which can generalize/reason/transfer knowledge beyond its training and experience.. and what’s the best path to get from here to there is sort of a subtle point.. open ai thinking bet a bn taking 5 yrs to get to agi via large amts of data.. i’m guessing not that simple
4 min – one route i’m interested in.. i’d love to make a robot toddler and get it to preschool and teach it to do everything a 3 yr old kid can do in preschool because that’s sort of the crux of everyday common sense intelligence
oh my.. straight up training alongside .. to be whales in sea world
5 min – but probably not going to be the whole story from here to agi.. because not the most commercially lucrative thing.. i’m hoping i can pull together resources to do that project.. but in parallel there’s going to be huge resources spent on applying ai in all sorts of application areas.. that’s what i’ve been thinking about in narrow agi.. ie: medical agi.. what if you had one ai that could integrate all those.. that connection is what we’re trying to do with singularity net
5 min – it’s easy to talk about things we don’t understand
7 min – i intro’d the term agi out of desperation to find a title for a book.. was going to call it real ai.. seemed insulting.. because narrow ai is doing some good things.. you can never write out rules for everything.. full gen intelligence to transfer to all possible.. not possible. we’re just talking about ramping up the level of generality
45 min – the end game has to be some form of ubi.. as my friend andrew yang is promoting in the us.. (but) who give ubi to .. developing world.. some form of bi is going to happen.. but transition is less obvious.. i don’ think developed world is going to give ubi to developing world..
50 min – after bi problem solved.. then younger people who grew up with ai and ubi.. and won’t be able to imagine how shitty it was to have to work for a living and not be able to do just whatever you want..t
ie: ubi as temp placebo..