By understanding and adopting Active Inference AI, enterprises can overcome the limitations of deep learning models, unlocking smarter, more responsive,…
In an enlightening discussion on the Spatial Web AI Podcast, Dan Mapes, President of VERSES AI, shed light on the future of artificial intelligence (AI), particularly focusing on ‘Active Inference AI’ based on Karl Friston’s ‘Free Energy Principle’. This groundbreaking approach is set to revolutionize AI by enabling systems to learn and evolve, mimicking natural evolutionary processes.
Mapes discussed the evolution of the internet and the emergence of the Spatial Web, highlighting its potential to transform our digital interactions. The Spatial Web, leveraging new protocols like HSTP (HyperSpatial Transaction Protocol) and HSML (HyperSpatial Modeling Language), is designed to create a 3D internet, offering immersive and interactive experiences far beyond the capabilities of current web technologies.
A significant part of the conversation revolved around HSTP and HSML. These new protocols are the backbone of the Spatial Web, enabling the creation of detailed, interactive 3D environments. Mapes emphasized that these technologies would democratize 3D web creation, allowing individuals and organizations to build their virtual spaces, much like websites today.
Mapes paid homage to Karl Friston, whose work on the Free Energy Principle is a key influence on Active Inference AI. This approach marks a departure from traditional AI methods, focusing on a bottom-up approach where AI systems can evolve and adapt in a manner akin to living organisms.
Throughout the interview, Mapes underscored VERSES AI’s role in pioneering the Spatial Web. By developing cutting-edge protocols and embracing Active Inference AI, VERSES AI is at the forefront of creating a more interactive, immersive, and intelligent internet.
Dan Mapes’ insights provide a compelling glimpse into a future where the Spatial Web and Active Inference AI redefine our digital landscape. As these technologies mature, we stand on the brink of a new era in human-computer interaction, one that promises to be more intuitive, immersive, and intelligent than anything we’ve seen before.
Chapters:
04:30 – Historical Perspective on Technological Evolution
08:14 – The Internet of Everything: Connectivity and Accessibility
19:50 – Envisioning a 3D Earth for Everyone
21:30 – Interacting with AI Systems
22:20 – Natural Language as a Programming Tool
26:10 – Exponential Growth of Digital Connectivity
41:49 – Addressing Surveillance Capitalism
42:56 – Introduction to Spatial Web Challenges
44:39 – Enhanced Privacy with Zero Knowledge Proofs
45:27 – Data Ownership and Privacy in Spatial Web
46:32 – Developing Safer AI Systems
49:00 – Community Moderation in the Spatial Web
49:52 – Development of Intelligent Agents and Virtual Environments
51:33 – Progress Towards AI Sentience and AGI
59:10 – GIA: Intelligent Personal Assistants
01:03:00 – Learning from the Flaws of the World Wide Web
01:04:46 – Anonymity and Self-Control on the Spatial Web
01:07:20 – AI as an Augmentation of Human Capabilities
01:14:09 – Evolution and Technology: A Continuous Journey
00:13
Speaker 1
I have a very special guest on our show today, and I’d love to welcome Dan Mapes, president and co founder of VERSES AI, and the founder and director of the Spatial Web Foundation. Dan, thank you so much for being here with us today, and welcome to our show.
00:32
Speaker 2
Great, great to be here. Denise, great to see you again.
00:35
Speaker 1
Nice to see you. So, Dan, you guys are doing some amazing work with VERSES with the Spatial Web Foundation. I’d like to talk a little bit about what that looks like and what that means for the world of artificial intelligence, how the Spatial Web kind of plays into that and what makes it different than what we’re seeing right now.
01:01
Speaker 2
No. Happy to share some of those thoughts. What a lot of people are missing right now and really has been since the smartphone came out, is the fundamental power of the Internet itself. The Internet really began 50 years ago, back in 1960, 919, 70, with just four computers. By 1994, were up to maybe 40 million computers. And that was enough to enable the World Wide Web View, be able to be spread on it, and enough for Jeff Bezos to go there’s 40 million computers today. That’s going to go to 400 million. That’s going to go to 4 billion, and then it’s going to go to 40 billion. I could probably make a business on here. And he started Amazon. And guess where we are today? We’re at 40 billion computers plugged into the Internet, all from those four computers back in 1970.
01:55
Speaker 2
So this is a massive multitrillion dollar engineering project with fiber optic cables under the ocean carrying the Internet and satellites and giant server farms that we call cloud servers. This is a massive, the largest engineering project humanity has ever undertaken. Makes NASA look like a high school science project. I mean, the Internet is massive. It’s connecting all the cities, 5 billion of the humans all over the world. And so then in the early 90s, really flowering in mid 90s was Tim Berners Lee began this idea that, well, if we’ve got an Internet of machines all talking to each other that created email, maybe we’ve got enough speed on the network to create a network of all our documents. And he thought that would increase scientific research by doing that.
02:49
Speaker 2
So we invented Http and HTML, a couple of standards, and ten web pages in 1993, 10 billion web pages today, all by everybody, all over the world. Largest library ever assembled in history, available to everybody on their smart device or on their laptop or whatever, immediately, globally. Wow. These are big, massive undertakings. And then certainly with the rise of the smartphone, this device, it’s not smart and it’s not a phone, it’s not a smartphone. It’s a powerful supercomputer, and it’s been able to miniaturize the chip. So it fits in a very small box now. But that smartphone that I was just holding up. The recent iPhone has more processing power than the most expensive computer in the world in 1995.
03:48
Speaker 2
And so the Cray two supercomputer in 1995 was $35 million and there were only 50 of them in the world, the largest corporation that’s had them. And so if you interviewed a Futurist in 1995 and they said, well, what’s 2020 going to be like? He said, we’re going to give everybody a crate and then we’re going to network all the crates together, put every document that’s ever existed in history all online, so everybody has the full library of human knowledge. And we’re going to put every product that’s ever been made ever, and is being made right now into various kinds of ecommerce things so you can just order them and have them delivered to your house overnight. We’re going to do all that. People would have just looked at you like, yeah, number one, you’re crazy.
04:30
Speaker 2
The Craig computer is 6ft wide and 6ft tall. No way you’re getting that into something like a phone that you could carry around in your pocket. And no way is that going to network to every other device in the world. It just looks crazy. But that’s the way that these exponential technologies emerge, really. We realized that we’ve got 25 years now of they’re web pages, hypertext Transfer Protocol. So Hypertext focuses on text. And then we’ve got 25 years of 3D computer gaming over here where everybody’s playing on Xboxes and PlayStations because the internet can’t handle 3D graphics at that point.
05:18
Speaker 2
But we realized that the speed of the networks were going to rise in such a way that and the chips were going to get faster, that there would be emerging of 3D computer gaming and the internet in the early two thousand and twenty s. And then we would need a 3D internet. Oh wow, need a 3D internet. Then you need a new 3D protocol. Because Hypertext transfer protocol is about web pages. Hypertext Markup Language is about formatting a page and where your picture goes and what this Hyperlink does and all this kind of thing. We would need a formatting language for the world. Where does this house go in this city? Where does this chair go in this house? All these kinds of formatting things. The way you set up a computer game, they’re called a scene graph.
06:06
Speaker 2
So we realized we needed a whole new language just like HTP and HTML for a spatial web, and that people could use all these game tools like Unity and Unreal and 3D modeling programs. And people are going to build Ready Player One, but it won’t be from one company. It’ll be millions of individuals just like websites. So a good way to think about these things is they’re 3D websites. Even a big game is like a Facebook is a huge two D website, just millions of pages. But these will be 3D websites. So we needed a way to model the world that you’re building.
06:51
Speaker 2
It could be an ecommerce world, like, welcome to the world of Gucci, and you’re in an immersive virtual reality or augmented reality experience, or it could be an educational environment or health environment or any kind of thing that people will make. Then they’ll make everything. We look at the World Wide Web, all that’s now going to be done in 3D, right? So we set ourselves then, of course, to develop the protocols. And the problem with a protocol, the reason nobody wants to write them is you’ve got to spend, gosh, four years and $4 million writing the protocol and getting it through the standards committees at the IEEE and turned into a standard like Wi Fi or HTP or TCPIP, any of the other big protocols.
07:38
Speaker 2
And so if you go to an investor and you go, hey, we figured out a way to write the protocols for the Internet of everything. It’s not just documents and machines, but it’s buildings and cars and drones and everything, verses and everything. Wouldn’t that be amazing? Yeah, that would be incredible. The Internet of everything. Yeah. How much do you need? It’s going to take probably $4 million and probably take about four years. Awesome. How much do we make when it’s ready? Oh, no, it’s a public standard. You just give it away to the entire world for free.
08:14
Speaker 1
Investors don’t like that.
08:16
Speaker 2
Yeah, that’s the way TCP is. That’s the way Http is I don’t have to call IBM or Microsoft or Apple and see if I can join the Internet. As long as my phone has an IP address on it, there’s the Internet Protocol address, and I can build a website, put a URL on it, boom. Websites are everywhere. So that’s when you get exponential growth, when there’s no friction. That’s what the Internet is really all about, is decentralizing. So we’re really just the third protocol. So the first protocol connected to machines, second protocol connected to all the documents. Then the third protocol is kind of the frosting on the cake that connects everything. All the cars, drones, ships, all the physical objects in the world, buildings, bridges, whatever. You can even count all the trees.
09:05
Speaker 2
But then it also links into all the fantasy worlds, too. So the planet Pandora from the movie Avatar will be a giant full 3D hologram that you’ll go in there and play games in and have dramas in and be part of it all. You’ll live with the people there locally. So all these IP environments that go from Homer’s Odyssey through to the latest film, they’re all going to be made into virtual worlds. Well, they all have to be connected as well, because some of the virtual worlds will be like Nike might have a virtual world where you go in and you design your shoe, but then the virtual world is connected to a digital copy of the factory.
09:49
Speaker 2
The robotic factory which makes the shoe for you, custom fits it to your foot, and then that’s connected to the supply chain that ultimately Drone delivers it to your house and then even later picks it back up for recycling when you want to recycle it back into the system. So you really get kind of a nervous system for the world where you’ve got this kind of a global kind of connectedness of everything now and then. That’s where the artificial intelligence then comes into play to inhabit that big spatial hologram of the Earth and all the metaverses and then help to make our flows of information and then the equipment and products and ideas and everything flow more smoothly with less friction.
10:39
Speaker 1
So it’s so interesting that you say that because I read something recently where it was talking about investors, and they’re all waking up from their slumber and the bear market, but they’re only interested in AI. Companies, AI Development. And they’re spending a lot of money in that sector right now. But at the same time, they’re also saying web three is dead. And with what I know about the spatial web and the AI within, I’m like, this is the glue that bridges all of these technologies together. But the majority of people don’t realize that’s what’s happening and that’s what’s coming. So maybe you can speak a little bit more on that as far as people who are developing these technologies and how it’s going to bridge them together. Because to me, it brings it all into this augmented reality world.
11:35
Speaker 2
When my wonderful partner Gabriel, Renee and I kind of started to whiteboard this project back in 2017, we realized it was only partially a technological problem. It was really an ontological problem. It was almost the architecture, how you structure this thing. And if you understand how the Lego pieces of a stack of technologies fit together, then that’s the key to kind of a successful software creation. And so here’s what we kind of basically came up with. It’s kind of like a three layer cape. And once you understand that, then you kind of see what’s coming and why it’s so obvious and how all these separate pieces fit together. Because there’s terms like blockchains and web three and augmented in virtual worlds and XR and artificial intelligence and IoT devices and all these things.
12:33
Speaker 2
It just sounds like a bunch of word salad, but actually they all fit together like a little Lego piece, just like a car has tires and an interior and a radio and an engine. But you put it all together, we just go, car.
12:46
Speaker 1
Right? I love that.
12:49
Speaker 2
That’s beautiful. It’s a design. You see what I mean? The car of the spatial web, that’s the pieces that all fit together are these four major pieces with a lot of little stub pieces that tie them together just like a car. So the primary core of the engine of all computing going forward is artificial intelligence because the core engine of software has really been software. So a computer without software is just a box. So it’s the software on the computer that makes it exciting. And that’s what cracked the German cipher code in World War II. And that’s what created all the accounting programs and database programs. And then the programs that we use on our laptops today, they’re all software. And so that software is getting smarter.
13:41
Speaker 2
And it started to enter the age of AI maybe even 20 years ago with kind of very simple forms. And when you get recommendations for YouTube videos to watch or books to buy at Amazon or whatever, that’s artificial intelligence running in the background. So AI is kind of running our big networks for mobile phones and computers and the cloud and everything else. AI is in our cars and helping to run the engines. So AI has been around for quite a while, but what we’ve been dreaming about is the AI that we see in science fiction movies, and that is AI that we can talk to and it talks back to us. So whether it’s the Hal 9000 in Space Odyssey or Jarvis in Iron Man or other really cool examples of AI, that’s the dream that we’ve been really having.
14:36
Speaker 2
The functional AI that’s been running our airplanes and ports and other kinds of things, our cars, everything that’s kind of hidden behind the scenes, you don’t realize it’s there, but it’s creating a foundational understanding of artificial intelligence, which is allowing it to move up the levels to the point where it can now mimic humanity, right? So with Chat GPT and some of the other generative AI programs, you ask a question, the answer is be one of the smartest people in the world answering the question. They’re still early, they still have a lot of flaws, but boy, they’re really exciting. And went from no users of Chat GBT on the 1 December, and within a week when it opened, within five days, there were a million users, and within two months, there were 100 million users telling each other about it.
15:37
Speaker 2
So AI is the core technology, and that’s what we call the logic layer. So in the three layer cake, the logic layer is the middle layer. It’s the core of everything. It’s the code you write. Then on top of the logic layer is the interface layer. It’s how we touch the logic. And so in an Excel program, you got little full squares that you can put numbers in and run a spreadsheet. A Word doc looks like a typewriter with a page of paper in it. You can type on it. There’s no paper there’s no typewriter. But we just use that as an interface because we’re comfortable with it. So interface design is the top layer. It’s how we interface with the logic. And then the bottom layer of the three layer cake is the data layer.
16:27
Speaker 2
So after I do something to the logic, I make a difference that makes a difference. I kind of make some adjustment into either some model or some thought or some number or something, and then I can store that in the data field. And so then later I can pull it back up, modify it further, draw on it for other purposes, whatever. So you got these three layers, the interface, the logic and the data layer. Well, the interface layer is moving toward AI. What is AI? AI is simply software that writes software. Right.
17:01
Speaker 1
So the AI then becomes the master of the logic layer too, then completely.
17:07
Speaker 2
So humans and AIS together will figure out cool advances to the logic layer. But ultimately, AI will probably just take it over and go, cool, I got it. Now then, the interface level is how do we see what the AI is seeing? Well, what does the AI see? Well, it sees a digital copy of the physical world. They’re called digital twins. So we make scanners and other kinds of things, and we can make a digital twin of a hospital or a factory or an entire city, ultimately the entire surface of the globe. And there are probably ten projects right now around the world by various government agencies and things to make a complete, perfect copy. A hologram of Earth, right in three D. AI then can see into that through cameras and satellites and sensors and everything. So it’s seeing into the world.
18:01
Speaker 2
And the way we see what the AI sees is by wearing AR and VR glasses. So they let us see the world the way the AI sees it. So with an AR pair of glasses, I can actually see the building in front of me. But with AR glasses on, I can also see data about the building showing me where I go and what I do or and so that’s what the computer is seeing because it’s seeing a digital version of that world. But you can shrink wrap the world with the digital version so that the AI is inhabiting the physical world, it looks like. And so by having an AR or VR interface, then I can kind of be in the world with the AI. I can see AI seeing. So that’s the interface level for humans.
18:49
Speaker 2
Of course, the AI, though, needs an interface level. How does the AI see the world? Every camera, there’s cameras all over the world and can see through them all through APIs and through every kind of there’s buoys in the ocean, measuring the temperature of the ocean. There’s weather balloons that are capturing information. So all that data is coming into the AI. So it’s seeing a real time, living, breathing planet Earth. Well, that’s going to help us manage our climate and other supply chains and avoidance of weather tragedies and dealing with earthquakes and other kinds of things. It’s an amazing model. And so then we can run simulations on that model to kind of see what the near future might be under different scenarios. And we can build better models of how we might want to do carbon credits to cut down the carbon influence.
19:50
Speaker 2
I mean, just many things once you’ve got that. So every child in probably 2030 or 2035, every human being as their birthright, will just have a beautiful 3D Earth that they could call up anytime. Kind of the way we call up Google Maps now. And it just floats right in front of you. There it is. You can spin it, you can zoom in to any point all the way down to practically a leaf on a tree, all the way back up to the whole thing. And so when the astronauts went to the moon, and even when they were in space floating around the Earth, they had some psychological effect from it. They started to realize that there were no borders. They started to realize, oh my gosh, it’s all one island in this beautiful dark sea of space and they live on the planet.
20:45
Speaker 2
And so it’s called the overview effect. You can look it up. So kind of well studied. But imagine now every child from the age of two on will have a complete real time 3D model that planet Earth has seen from some perspective in space. You can change your perspective. You can run forward back in time, see the Ice Age, you can go forward, you know what I mean? It’s just going to be remarkable. That’ll just be a basic thing that you have and then you can zoom in anywhere and then dive down and next thing you know you’re in Switzerland or something and you’re walking around digital twin of and you’re full in. So I mean, it’s a remarkable kind of natural interface.
21:30
Speaker 2
So I think that brings us then to the final part of the AI, and that is that for the first time now, we can actually talk to the AIS and they understand us. So that opens up the power of the computer because we’ve always wanted to get this processing power and use it either to run our businesses or to educate ourselves about something or whatever. And so initially we had to write code in these arcane languages of Fortran and Cobalt and later Lisp and other cool languages like Java and C. And then in the initial days, we would have to give our card stack to a white code scientist who would run the stack and then give us our program back after they’d run it. And we’d debug it and slowly do that kind of thing.
22:20
Speaker 2
So basically, we’re trying to program that computer, but we’ve got to go through all these interim steps, and then we finally get to applications where we’re programming the computer by just putting in natural things like, hey, add up these row of numbers or typing this letter, do something with it, make it a PDF, and we’re going to email it to somebody. So now I’m programming the computer directly. I don’t have anybody between me. I don’t have to learn a programming language. Just my natural language is the programming language. And then now we’re at the final step now where there’s no application, there’s just the AI. And you just talk to the AI and then it uses all the tools it has in the world to solve your problem for you. And it’s getting smarter every day.
23:12
Speaker 2
So if it can’t solve a problem today, it might be able to do it in a week or a month. So we’ve entered a new age now where we don’t have to go through a specialist or we don’t have to have special arcane knowledge to use this multitrillion dollar global network of 40 billion computers to interact with the world. We can just talk to it, and the AI inhabits the entire network. So the AI I’m talking to is connected to the AI you’re talking to. They’re really all one kind of Meta AI that’s interacting with humanity and helping us manage our planet and solve our problems.
23:54
Speaker 1
So let me ask you then, so let’s talk about a little bit about what’s coming this summer with VERSES, with KOSM and with GIA. And what I’d like to kind of point out is that with Chat GPT Four, they just launched Plugins, right, where people can, where it can plug into different companies that already exist, like I know, like Expedia and Kayak and different things that are for travel. They’re already working with it. They’re already using the plugin, but they have a whitelist that you can join. They’re going to start slowly rolling it out. But to me, what’s really interesting is that’s going to add a lot of functionality for those companies, for their websites, but it’s still taking this AI into applications in a Web 2.0 environment. So all these applications still are siloed.
24:55
Speaker 1
They’re all still these siloed tools that are being built. And I know what Versus has coming is completely different. It’s joining everything together in this interconnected realm. So please talk about that a little bit.
25:14
Speaker 2
Well, I kind of mentioned it earlier. That is the very fundamental vision of the internet itself. Let’s connect all the computers and that way everybody can send free messages to each other all over the world and that’ll foster greater communication. And so gosh, we did it in spades. Started with four computers, we’re at 40 billion. Then we connected, oh, let’s connect some documents online. Well, guess what? Every document in the history of the world is now online. It’s like 1012 billion documents now online, each one with their own unique address, fully indexed by Google. So you can quickly search the library like a Dewey Decimal system and find exactly the page. You know, it’s the very ethos of the Internet. Connect everything and give it to everybody. Pretty much for mean, if you want to build a website with a paywall, you can.
26:10
Speaker 2
But the web itself is open access, right? So we just followed that exact ethos, something that’s obviously a proven architecture from 50 years now and then 25 years with the World Wide Web. So why not continue to grow that thing? You know what I mean? It has nothing to do with silos of AIS answering magical questions. It has to do with networked intelligence all over the world, all interacting and we’re all growing together the way we do culturally now through our normal media by writing articles and reading them and doing PhD theses and contributing to the scientific knowledge. The spatial web is simply that. That’s what I mean. You’ve got to get the architecture right, otherwise you put a limit on it. For instance, right now we don’t have artificial general intelligence. We don’t have like a human scale intelligence.
27:08
Speaker 2
We have what’s called artificial narrow intelligence. So we can train a neural net to play chess, and it can play chess so well that it can be a grandmaster, but it can’t drive a car or it can’t calculate other kinds of it’s built to do that thing. And it’s just in the nature of the way these artificial neuro intelligence are made. They’re just a box of neurons that we’re training on pattern recognition. So we’ll show it a million pictures of cats and then a million pictures of dogs. Then we’ll show it a picture of a dog and hold it up to it and say, what’s this? And it’ll go 90% dog. That’s what it does, right? It’s a train.
27:52
Speaker 2
But if I show it a horse, it doesn’t have any idea what it is and it doesn’t even know what a dog or cat is. It’s just going that kind of group of things looks kind of like pattern A. That kind of group of dots kind of looks like pattern B. There’s no sense of self. There’s no world model. There’s no understanding cats and dogs or animals or anything like that. It’s just a pure pattern. And so it interprets A into the word cat, but it doesn’t know what a cat is. So that’s the AI that we’ve had up until now. The AI we want is more like a human being that’s knowledgeable about the world. So if I tell you I’m going to go to Europe, you already kind of have a sense that’s another continent.
28:29
Speaker 2
I’m probably going to get on a plane and fly there and that sort of thing. So these large language models are wonderful and they’re great tools, but the power of the collective wisdom of humanity in real time all plugged into a network, that’s what we want and that’s what you’re doing. That’s what the spatial web enables. So these protocols allow AI to be networked. And our vision is that rather than build a giant machine with a 175,000,000,000 or a trillion parameters and then be able to ask it, tell me a story about Paris. And all it’s doing is calculating mathematically the words that are most closely related to the word Paris, and constructs a story about a romantic holiday in Paris along the scene and the Eiffel Tower in the background, because that’s all it’s doing.
29:31
Speaker 2
It’s just running mathematical measurements on what words go with Paris, and it’ll construct a story, and therefore it can construct false stories. Like I say, okay, what’s the best medical treatment for type two diabetes? It may construct a very logical story, but it got its facts wrong and actually gives me really bad advice because it’s not good. So what we said is, let’s don’t do that. That’s a great first step, and it’s handy for writing stories and other companies. You can’t really run your world, but it’s too random, it makes too many mistakes to put in charge of running anything like an airport or a city or even your metaverse. And so we said, let’s go the way nature did it. Let’s start from the bottom up.
30:19
Speaker 2
So we’ll make these intelligent agent AI tools, and we’ll give them to individuals almost like website building tools, so then they can build their own AI. Maybe it’s a person who’s a specialist in nutrition, and they build an AI that just helps you plan your meals for the week and look at your diet and help you achieve your dietary goals. Another person might be building something on how you run a garden and do it properly. So everybody’s got these specialized knowledge sets, and they can now build them. But then, because they’re running on an Internet protocol, they can all communicate with each other, so they can learn from each other as well. So then we have, I think, 5 million apps now in the iPhone App Store. And so Apple didn’t make them.
31:12
Speaker 2
Apple just built the tools and said, hey, there’s an App Store. Rock and roll. Build apps, and we’ll help market them. And so we kind of followed that model, and we said, well, let’s just give everybody the tools. So they’re building their own AIS, their own little models of their specialty, and then they’re actually maintaining that knowledge graph. So we’re getting the intelligence of individuals all over the world and groups of companies and whatever all over the world all working together, and then the AI can connect all these things. So it’s called emergence, when collective intelligence arises out of many small cellular intelligences. And so that’s really the way we’re approaching AGI, and we think it’s probably the only way to get there.
32:05
Speaker 2
In fact, there’s kind of a humorous cartoon that goes around in the AI circles, and it’s like, AGI is the Moon. Artificial general intelligence, human intelligence, and artificial narrow intelligence is being here on the ground, looking up at the Moon. And the thing is, how do we get to the Moon? And so artificial narrow intelligence is climbing like the Eiffel Tower or the Empire State Building. They’ve gone up ten floors, and they go, See, we’re closer. Then they go up 20 floors, see, we’re closer. 30 floors, see, we’re closer. And so that’s why a lot of people think, well, you can’t get to the Moon that way because you’re just training neurons, and you’ve got to get bigger and bigger and more expensive and more flaws in it.
32:54
Speaker 2
So that’ll not get you to an Artificial General intelligence, even though most people think it will. But a lot of high end artificial intelligence specialists do not believe that’s a good path. And so a better path, rather than climbing a building to get to the Moon, would be to build a rocket. Well, that means you got to start over that whole entire new approach. You’re not just getting from the 10th floor to the 20th floor in the building. You’re going like, we’re not even using the building. We’re going to go over here and build this new kind of thing, kind of like a building. It’s tall and thin, but it’s got a big engine on the thing on it, and you can get to the Moon with this thing. And so that’s called artificial general intelligence tools.
33:34
Speaker 2
And what we’ve done is we’ve partnered with a brilliant neuroscientist in England who’s been working on a new theory of AI called Active inference. AI based on the free energy principle. And the work that were doing and the work that he and his research staff and PhDs were working on, just merges together so perfectly and beautifully. So then the synergy out of it’s not ten plus ten, it’s ten times ten. Right. So then they got their stuff that far, we got our stuff this far. We put them together, and then so that’s the magic that’s taking place at versus AI. Right now is this extraordinary research that’s been going on at University College of London under the direction of Dr. Karl Friston.
34:29
Speaker 2
And then our teams work in traditional artificial narrow intelligence and using that in this new architectural stack of interfaces, logics and data frameworks. And we brought that all together with these spatial web protocols. Wow. Now we can give anybody the ability to create AI applications anywhere in the world and at any age, and so it’ll gather the collective wisdom of humanity and then link it all together. And then a larger metai then has access to all of that to help us manage the planet as an example.
35:12
Speaker 1
Yeah. So what’s really fascinating to me about that is by allowing people to build their own intelligent agents on top of your operating system within the spatial web that’s powered by this artificial intelligence. To me, we’re looking at a very quickly coming, approaching explosion of development. And just this entire world of anybody can create whatever they want and have it work for them and anybody can do it. You don’t have to have special knowledge, and you’re democratizing access to AI development in a way that is just astounding. So maybe talk a little bit about that because one of the things that’s really interesting and what I’d love for people to understand is if you’re developing a project right now, you want to be developing in this environment because it’s going to empower everything through this network effect.
36:22
Speaker 1
So maybe you can talk a little bit about that.
36:27
Speaker 2
Making a piece of software has been a pretty difficult process for the last 75 years. Since the first programmable, computers kind of appeared on the scene. You have to learn arcane languages. You’ve got to later develop, learn how to make apps on smartphones. It’s always been this kind of journey to making applications that you might want to create for either use on a smart device or on the Internet itself. But over time, it keep getting a little bit simpler. Things like WordPress came in and shopify for building websites, and there’s app development tools and things like that. But as we enter the age of the spatial web with powerful AIS that can help you create anything, you really can talk applications into existence.
37:23
Speaker 2
So an eight year old girl in Kenya might end up being one of the top programmers in the world because she’s conceptualizing an AI application that really helps people in ways that we have never imagined. This is how evolution works. I mean, evolution of ideas gosh, there was hardly any physics 500 years ago. And then we go through Newton and then Einstein, and now we’re into all kinds of new subatomic worlds. So everything just keeps unfolding and you never know what corner it’s going to come from. I mean, even the Internet wasn’t done by a major corporation. It was just done by some PhDs working together. World Wide Web was not done by the Big Apple and Microsoft companies, just done by some young PhDs just kind of messing around with how could they make a web environment?
38:19
Speaker 2
And so a lot of these breakthroughs come in surprising ways and not from where we’re not where you would expect them to come from. It’s just like Mozart. Just genius ideas pop up and then come up. Now what’s great now is because everybody’s connected, if you do come up with something amazing, you can share it with the whole world very inexpensively. So the AIS will really, at least for the next 1520 years, are going to be tremendous. They are going to be our interface to this entire multitrillion dollar computer network system with all of its capabilities and just basically talk to it. And it’ll help you either create things or connect to things or meet the people you need to meet or whatever. So AI becomes the interface. You don’t really have a touch screen. You don’t really have keyboards.
39:14
Speaker 2
Right now, we watch old movies and when somebody runs into a paper and looks for a quarter to put in the thing, it kind of looks kind of humorous. We all chuckle. But I mean, can you imagine even ten years from now, kids will look at people using keyboards like, what are you people doing?
39:28
Speaker 1
Right? Yes. Okay. So it’s funny because to me.
39:35
Speaker 2
In.
39:36
Speaker 1
This type of an environment, because it’s connecting all of the Internet of things and all of the trillions of sensors that are coming on board and everything else. To me, it’s that wealth of all of that data that’s going to bring about innovation that we can’t even imagine right now. And that to me is one of the advantages of being in this environment. But it’s so fascinating to me when you think of this connected world and you think of the interaction within the intelligent ecosystem of it and the assistance there. And then I know one of the advantages is with HSML, the formatting language for it baking in context into it allows the protocol to be a gatekeeper as well.
40:28
Speaker 1
So when you’re talking about programs that can interact with all of the other programs and all of the data that’s out there, I think it’s also important for people to understand that you can also block off certain aspects of your data and the information so that it’s this protected environment as well. It’s not just a free for all. So how do you see that playing forward, especially when you’re talking about enterprise corporations coming and building their own programs within this space, and how do you see that playing out?
41:01
Speaker 2
Yeah, I think if we look back at the founding of the World Wide Web 25 years ago, or almost 30 years ago now, when Tim designed Http and HTML, particularly Http, there was no identity layer in there. So for you to have any identity on Facebook or anything, you have to log in. And so when you log in, you have to agree to the terms and conditions of using the application and there’s terms and conditions we get to use all the data. And so then you end up being kind of without realizing you’re working for Facebook. And so then they start to manipulate you to get you to stay on longer because their business is selling ads. And the more your eyeballs stay on there, the more ads go by and the more money they make.
41:49
Speaker 2
So you get kind of an unvirtuous circle there with that. So you’re really just a data provider to the system that they’re selling. They’re harvesting data from you constantly, google and everybody. So that’s one of the problems of the current World Wide Web is called surveillance capitalism. Another problem with the World Wide Web is hacking. And another problem with the World Wide Web is faking. The deep fakes are getting so good, you can’t tell whether this is Biden saying this or Trump is saying that the mouth movements accurate. So we need some kind of verification process on the as the Late Show would go, the truthiness of anything. Those three things hacking, tracking and faking you really need to address those before you build something like the Spatial Web. Otherwise you’re just really creating more of a problem than the solution.
42:56
Speaker 2
So having 25 years of looking at the flaws in the World Wide Web, were able to kind of attack each of those flaws. A very simple example would be in the Spatial Web, you own your own data. You have a blockchain ID, and you have your own data vaults, and you have your data. If you want to sell your data, you can sell your data. There’s data exchanges on the spatial web. You can sell your data and you can anonymize your data and sell the anonymous data. For instance, let’s say you had a Tesla car. Tesla might want to buy the stream of data off that car. They don’t have to know it’s Denise’s car. They want to see their fleet to understand how their batteries are performing and when the brakes fail and other kinds of things.
43:50
Speaker 2
So they’ll buy the data from you. And probably the data is tokenized. And so you’re getting tokens into your token account and all your actions. Then in the Spatial Web, you can monetize them if you like. And so that becomes a source of income for every you get paid to use the web as you should. That comes with what’s called self sovereign identity. So that enables that. And with self sovereign identity, you get a second order benefit. And that is called zero knowledge proofs. So that just to kind of use a real world example. If you’re standing in line for a nightclub, you have to show your ID at the door. Well, they can see your entire name, address, everything. And all they really need to know is that you’re over 21.
44:39
Speaker 2
But in an augmented reality networked world, the door person might have a pair of augmented reality glasses on, and they can see a green check mark above everybody’s head. Because biometrically, that person is over 21. They don’t know your name. They don’t have to know anything about you. You just have proof that you’re over 21 because it links to your driver’s license online, which is in your data vault. And so you’re green. Green. Oh, here’s, person with a red X, come over here, have a look. What’s the deal? Oh, I’m visiting from Argentina. Let me see your passport because they’re not in the system yet or something. Okay, you’re over 21. Go on in. So zero knowledge proofs mean you only get the knowledge you need about me, not everything. And so even your medical records.
45:27
Speaker 2
When you go to see your doctor may request an MRI scan or something. Well, that goes to your data vault. And then you loan it to the Doctor for a week or a month or whatever the treatment time is, and then it goes away, then it disappears and only exists in your vault. So these kinds of things really help with the tracking. And then, like I said, you get an income stream from it, if you like. And then the faking having AI in the network can really tell where the source of this is in the network. And so probably this isn’t really Joe Biden talking or Donald Trump talking, it’s somebody else. And then I think also with the hacking, new kinds of cybersecurity, not using relational databases, using graph databases, using blockchains, things like this, we can cut down on the hacking.
46:24
Speaker 2
So then we start to get a much safer web to exist in. We’re not being surveilled, we’re not being.
46:32
Speaker 1
Hacked and safer AI.
46:36
Speaker 2
Right?
46:37
Speaker 1
The AI becomes accurate, deliberate, mindful. It’s funny because people are sitting there with the Chat GPTs and stuff and they’re trying to break them. And it seems to me that in the spatial web environment with the AI, there isn’t a way to break it because it’s real time and it’s based on real data. It’s not just this pocket of enormous amount of data that it’s pulling from to try to form correlations and determine accuracies. It is accurate. All of it will be accurate. Is that a good way to look at it?
47:15
Speaker 2
Just think about it logically. Let’s say that the two of us had a company and we’re going to build the largest AI ever in history. Like 10 trillion parameters. It’s going to take us two years and a billion dollars, but then it’s going to be this monster thing. Well, what did you put in there? You got to filter everything that’s going in, because if you put garbage in, you get garbage out. Geigo is the oldest story in computer programming, so you got to clean all the data. So that’s called training data. That costs time and money to clean. The more data you have, the more expensive it is. They’re hiring people in Kenya and India to clean the data. It’s a mess still. Bad data gets in there.
48:01
Speaker 2
That’s why you get in there and get around the filters and cause Chat GPT to even threaten people or do things like so I think there’s a problem fundamentally with the giant AI thing, the silo that somebody built. And so what we’re saying is, let’s do it more like the World Wide Web. You’re responsible for the content on your website. You update it. Google isn’t running around validating all the websites. You’re responsible for maintaining your own website. And if you put on some bad information onto your website, whether it’s false or misleading, or whether it’s something to do with like, terrorism or some other kind of thing, the community sees it and points it out. And then usually you’re taken down or you have to modify the thing.
49:00
Speaker 2
So the world Wide Web kind of has an immune system a little bit against, like, a terrorist website or something. And so by having millions of intelligent agents that are being maintained by individuals and groups for themselves, they’re like little websites when there could be very big websites. In fact, I mean, it could be a big virtual environment that you build out there. It could be like the entire planet of Pandora could be a metaverse, and you go and you hang out there, and it’s a billion dollars to build that thing. And so you really want that safe and well done and well handled, and that becomes a big money maker for James Cameron, who developed the whole they these individual agent worlds. They’re like agent worlds.
49:52
Speaker 2
They can just be knowledge bases, or they can be virtual environments, or they can be augmented reality things in the world. They’re all maintained by the people that built them, just like websites are or just like apps are. Apple. If you got a problem with Uber, you don’t call Apple, you call Uber, hey, my driver was horrible. You call Uber. You’re not calling Apple. So we really are just enabling everybody to do what they’re already doing. They’re already building websites all over the know. They’re already building apps all over the well, there’s a new set of tools now. Build intelligent agents, AIS. All over the world and let them talk to each other and communicate with each other, and each one of them getting a little bit better and learning. And the whole thing grows steadily over time, just like the web did.
50:39
Speaker 2
I mean, we started with ten web pages, and now we’re at 10 billion web pages. I mean, that’s how this thing happens. When you’ve got an open network with no friction, then people can add to it all over the world, and you get this collective intelligence that rises out of it. And so we actually put out a white paper on using intelligent agents to get to collective intelligence. And I think it’s on our website. You can go there and download it. Our website is VERSES AI.
51:12
Speaker 1
I’ve read that white paper, and in it, you guys have a timeline, a preliminary timeline for when you expect the potential for reaching AGI and then even super intelligence. And maybe you can speak a little bit on that.
51:33
Speaker 2
Yeah, I think more than the actual dates, what we’ve done in the white paper is show the steps that one takes to get there. And so they’re very logical, and you can understand them. And we’re already like I say, we architected a rocket rather than a building. So we kind of have a mechanism now that run the math on it. You can get to the moon and back. Cool. So that’s laid out in the white paper. Here’s where we are. Here are the interim steps as we move towards Sentience and then AGI, and then ultimately beyond that, onto a kind of a singularity model where we’ve got a superintelligence that’s so that’s the fun part about it. No one’s actually been able to lay out the thing in the past with the Agis. They mimic intelligence. They do.
52:31
Speaker 2
They can score in the top 10% on the law boards, they can score in the top 1020 percent in the SATS. I mean, they absolutely mimic intelligence, but they themselves are not intelligent. And because there’s no self there’s no world model. Part of our intelligence, the way we measure our intelligence is how big and how accurate is our interior model of reality. An intelligence test checks for or they can be other kinds of intelligences too. We realize now there’s about twelve types of intelligences that they’ve discovered, like emotional intelligence and aesthetic intelligence and these kinds of things as we develop these systems. And they’re growing intelligence. They’re not just mimicking intelligence, they’re actually mimicking the functions that we use to develop our intelligence.
53:27
Speaker 2
Whereas an AGI is trying to stuff all the an AI like large language model is stuffing a box full of just tremendous number of parameters related to language and then you can query that and it’ll put the language together in amazing ways but it actually isn’t learning anything in the conversation. So what you want is an AI that’s actually learning and growing the way a child does, right? So that would be a rocket that would get you to an AGI. And so that requires an entirely different approach. You’re not building a box full of sentences, you’re building a model of reality. And then that model of reality is growing and evolving as the AI engages the actual physical world or the metaverse worlds.
54:18
Speaker 2
And so that model then is growing and it’s the model that’s so if you look at generative AI, why do we call it generative AI? Because it generates the sentences based on the sentence that you put in. So oh, tell me a story about a girl who lives in Paris in the 1960s and whatever, some story and it’ll write it for you really beautifully and it generates everything right through these transformer models. But what we’re generating isn’t that we’re generating a world model. So then you can query the world model. It’s getting bigger and better. I can ask Siri what’s the weather in London? And Surya will tell me. But now it’s going to be really like I said, you’re going to have one in the palm of your hand practically. You can just zoom in and out on.
55:02
Speaker 2
So that world model will be evolving for the rest of our lives, right? And that intelligence then will cause us to increase our intelligence. So a lot of people are thinking, oh my gosh, you build these HEIs and then they get smarter than humans and they just go away. No, if you take somebody here from 500 years ago or 1000 years ago, we seem really smart to them, like, how do you airplane? What is electricity? We’re, like, really smart. So we’ve evolved our consciousness just through universities and talking and writing. Well, imagine now human minds engaging an AGI that can build models and show you things. You’re going to have an eight year old girl, ten year old girl doing almost PhD level work, probably in 2030 years in physics because it’s a conceptual thing. Most of you look at Einstein’s work.
55:56
Speaker 2
I mean, his lab was his mind. He would kind of meditate a little bit. They called Gadakan experiment, and it would just be thought experiments. He would imagine himself as a photon and that kind of thing. Well, now, man, with virtual reality and children growing up with this thing, oh, my God, we’re going to have some really smart kids.
56:18
Speaker 1
Yeah. So it seems to me that you guys have solved the alignment problem for all of these generative models. They’re having to do the reinforcement learning based on human feedback with their data scientists just to train it to be more in line with what a human expects the answer to be to appear more accurate. But that reinforcement is baked into your system for the human connection.
56:50
Speaker 2
Yeah. Again, we’re not building it. You’re building it. You know what I mean? Right?
56:56
Speaker 1
Yeah. Okay.
56:57
Speaker 2
By connecting all the humans, your knowledge of nutrition will go in there. And if it’s wrong, probably another nutritionist will talk to you about it, and you guys will all work it out. So the large language model of the spatial web, actually, the large knowledge we call them knowledge models, knowledge action models, even. That’s why it’s called active inference. They can actually take decisions on your behalf if you give them the ability to do that. Those models are built by you and maintained by you. So we’re not going, hey, here’s our new magic box. No, we’re going, Here are tools. Let’s get a billion humans building artificial intelligence. Exactly. And then they’re all networked together, and then that creates a bigger model than you could build as a large language model.
57:48
Speaker 2
But just like the World Wide Web, no company could build a 12 billion page World Wide Web. But millions of people working together, not even coordinating, just in separate parts of the world while we’re having this conversation, probably 1000 more Web pages were just uploaded to the World Wide Web. So that’s the path forward. That’s the evolutionary path. That’s the alignment not just with humans, but with evolution itself. We’re at the most core issue with evolutionary development. And so the system has also some really nice immune system qualities. Very hard for a terrorist to get control over it because you might get control over one or two agents, but the rest of the agents could seal that off and go, look, there’s a problem there, and get shut down.
58:36
Speaker 2
So I think it does solve the alignment and the immunity problem when you don’t really have rogue AI. Now either being captured by a malevolent group or just going malevolent, you got protection because you got millions of people building these little three D AI website kind of thingies.
58:57
Speaker 1
That’s very important. So, Dan, tell me a little bit about GIA. Tell me why GIA is so different than any AI assistant that we’ve seen so far.
59:10
Speaker 2
Yeah, so GIA is anagram for right. So GIA is just a more intelligent siri. I mean, really, siri does a lot of things, but is not very you can tell me the sports course, the stock market, answer some basic questions, that kind of thing, but what you really want is one way to look back in history is look at what are really wealthy people have that nobody else has. Probably the one thing that really wealthy people have, that average person doesn’t have anymore, because we all have cars and houses and things like that better than kings and queens used to have. But what we don’t have is a staff of assistants that are really working on our behalf, that know our taste, that understand what we’re trying to achieve.
01:00:01
Speaker 2
I may have one person that handles all of my travel and all of my speaking engagements, things like that. Another assistant might be looking at my diet and health, and another one might be handling my fashion or whatever, things like that and so on, right? I mean, financial advisors, that kind of thing. So what GIA is it brings that level of intelligence so that instead of everybody having a smartphone, everyone has a smart assistant. And that smart assistant kind of functions almost like a fairy godmother or angel or something. It’s like wholly dedicated to you, more loyal than your dog. You know what I mean?
01:00:43
Speaker 1
I love it.
01:00:44
Speaker 2
Dedicated to your well being, helping you achieve things, and getting better at doing that every day. So simple things like, hey, I’ve got to go be in Paris to give a talk on the twelveTH of April and it already knows my preferred airlines and my preferred hotel groups and things like that, and just sets it all up and then just gives me the readout, I did that for you. While you’re in Paris, I know you probably want to go to see the Monet’s. It’s one of your favorite things. I got you tickets at that museum on the off time when you’re not speaking and things like that. So it functions like a very customized personal aware assistant to you, dedicated to you.
01:01:27
Speaker 2
So obviously in the beginning it will have some minimum viable capabilities, but then over time, as it communicates with other GIAs around the world, then their capabilities will grow and then you can put your own wrapper on it. So if you want, if you’re English and you want to have it in the face of an English butler or something, you can have it. An English butler, or you could have it as your best friend or whatever you want to model it up as. You could model the name and the face of the general intelligent agent technology so that it’s friendly to you, but basically it’s a powerful capability that’s there to serve you and particularly to be able to interact with the world through IoT devices.
01:02:16
Speaker 2
Turn this on, turn this off, go to this, pick this up, but also through all of the digital worlds of the metaverses and digital twins that we’re going to be dealing with and then all of the companies in terms of ordering things and other kinds of things. So just a capable assistant that begins with some nice powerful capabilities and over the course of a decade becomes mind blowing.
01:02:44
Speaker 1
So we have some questions from the street people who have they’re wondering about the AI that’s coming and they’ve asked specific questions, so we have video of them. I’m going to cut to that and then maybe you can answer their questions for them.
01:03:00
Speaker 2
Sure, yeah. Well, I touched on that before. We learned from the world wide Web that has no identity layer, it has no location layer, it has no security layer, just an open web. So therefore you can be surveilled in anything you do. And when you have to log in anywhere, then you agree to the terms and conditions of everything. So we had the luxury of looking at that 25 years of the kind of flaws in the current world wide web model and were able then to build in capabilities into the spatial web that would protect us against this. And one of those would be self sovereign identity data vaults, other kinds of things. So basically, you own your own data and you have zero knowledge proofs.
01:03:54
Speaker 2
Which mean that if it’s an over 21 website, or you have to be a member of a club to join this website, or you have to be a subscriber to read this information. As long as you’ve achieved that, then you’re just a green checkmark to that system. They don’t know it’s you, Denise or Dan or and so that way you have anonymity on the web kind of the way you do walking around the world. You don’t have perfect anonymity, but in general anonymity, if you do have to log in at some places in the spatial web, you’ll do it because it’s somebody you want to join. You’ve already joined that club or something and they know you and love you and you’re part of it. So you’ll have total control over how much knowledge about yourself is shared with the world.
01:04:46
Speaker 2
Well, it’s hard to see. It’s very hard to see over the horizon. I mean, Wilbur and orville Wright invented the airplane in 1903 and we had jets in the 1940s and were on the moon in 1969. So in 66 years went from never having flown to flying to the. Moon. I mean, it’s remarkable what happened. So it’s very hard to see over the horizon. But I think at the minimum level, I think the AI will help businesses run more efficiently, run more effectively, help in the ordering process, help in the pricing process, being able to run. We do spreadsheets now, but imagine an AI spreadsheet. It could run multiple variations on things, and maybe prices change during the week. It’s higher on the weekends and lower during the week. I don’t know.
01:05:39
Speaker 2
You get all kinds of nuances that we can’t do yet, because we just can’t see well enough into the data yet. But I can read that data more clearly and help you run more efficiently. So, again, it’s like having a car that gets 10 miles per gallon versus having a car that gets 30 miles per gallon. That’s a big breakthrough. So I think the AIS initially will help us run our existing businesses. But probably as you learn how to build your knowledge about crystals or whatever specialty you have that allowed you to create the business, you’ll be able to build that knowledge now into an AI and then share that knowledge with the world. And as the AI is getting pinged and sharing its knowledge, you’re receiving some kind of a token for that. And so that becomes a new income stream as well.
01:06:28
Speaker 2
So now you’ve got access with sharing your knowledge with the entire world. There’s almost 8 billion people, and probably other than babies and people in hospitals and things, pretty much everybody’s online now. So we have five or 6 billion people plugged into the global internet. And that’s a big community to talk to and share your wisdom and knowledge with. I think over time, we learn to use it. If you’re a horse farmer and when the cars came in, you may not sell as many horses, but you can get a car and figure out how to make that work. So, calculators, instead of adding columns up with paper and pen, I think we’ve gone through these shifts every once in a while. So this is really an augmentation of our capabilities as a person.
01:07:20
Speaker 2
If you’re a good business manager now, you’ll be a bit better business manager with AI. Very good.
01:07:26
Speaker 1
And thank you to Hodljake correspondent on the street for gathering those questions for us. And thank you to Roth and Tom for asking them. Dan, I’d like to wrap this up with something that I think is really important, and I think it’s important for people to understand. I’ve watched you over the last handful of years and the evolution of VERSES and the Spatial Web Foundation, spatial web protocol. I’m a member of the IEEE working group, so I see what’s happening there. And one of the things that has struck me is just the level of integrity with the leadership in putting all of this together. And it’s really clear to me that everybody involved understands the gravity of the responsibility behind creating this technology and putting it in place.
01:08:28
Speaker 1
So I’d like people to hear that from your perspective, how you feel about that, what the intention is on the side of the leadership in this effort.
01:08:41
Speaker 2
No. Great, sure. No, it’s obviously core to the whole thing. I mean, when you’re dealing around with protocols that will touch everybody’s lives, you want to really do them as well as possible. And obviously it is technology. You do have minimum viable products, so you get it out there architected in such a way that protects people well and then continue to evolve it. The first cars did not have airbags and seatbelts. We only learned about those when people started going through windshields. So it’s not only the core design that we’re doing now, but the ongoing protection of the technology and the use of it by people that is part of the process. Because, like I say, seatbelts were added later to cars and then finally we figured out how to do airbags. We added those so less people are dying in car wrecks now.
01:09:30
Speaker 2
So I think it’ll be the same with these technologies. But we have had the luxury of looking at 25 years of the world wide web. And so from the very beginning, we’ve really tried to address how do we create the most beautiful thing that’s the safest thing to use and protects us against malevolent use in the future. And we wrote a book about it while were doing it so everybody can see kind of our thinking on it. It’s called the spatial web. It’s available on Amazon, and it’s a nice kind of look at the overall picture and problems we’re facing as a global community and how something like the Spatial web might help with that. And then we also brought in ethics people to surround us as well. So we have the head of ethics and AI at the IEEE.
01:10:21
Speaker 2
John Havens is on our advisory board. Sarah Mansky, who’s a cyber ethicist, is on our advisory board, and they write position papers and talk with us and guide us in these areas. So we’ve got a great board of directors. We have the former vice chairman of Deloitte on our board and he’s chairman of our board, former head of the Toronto Stock exchange is on our board. So we really try to bring that ethics to bear, not only in the code itself, but how we run the company and then how its financials are handled. We are a public company now. We’re listed on the Neo exchange in Toronto and versus AI. So if you want to get any information there’s information on our website as well at versus AI. And yeah, we just try and be openly as open and transparent as we can.
01:11:12
Speaker 2
And we’re pretty clear in our corporate governance models and the way we talk to all of our employees about what our ethics are and what we’re trying to achieve here. And we’re definitely part of a community of technologists. We had the number one neuroscientist in the world, Dr. Karl Friston. Join us. University College London He wouldn’t have done that if he didn’t believe that were doing the right thing for the right purposes. So it really is a lot of care around it, but nobody’s perfect. We’re going to make mistakes, but they won’t be intentional mistakes. Hopefully we’ve got enough conscious people on the team and also around us that won’t be able to ferret out any fundamental errors or correct them quickly if we find them well.
01:11:59
Speaker 1
And it sounds like the whole entire protocol itself has guardrails in place to where the mistakes won’t be astronomical, they won’t be super detrimental. They’ll just be learning things to learn.
01:12:15
Speaker 2
From and evolve 100%. I mean, the funny story, the other day, Google launched their barred AI to compete with OpenAI’s system chat GBT. And somebody asked the Bard system, what do you think about the Justice Department and Google and their discussions on monopolistic behavior and whether Google should be broken up? And the AI comes out and says Google should absolutely be broken up. Google’s own AI. It was hilarious. So, again, the AIS that we’re building now are really just kind of Wild West things that can make we call it hallucinating. They can even make stories up and make perfect sense out of them. But I think having a bottom up approach where individuals are taking responsibility for the data they’re putting into their agents helps to solve a lot of that.
01:13:15
Speaker 2
So I do think there is a bit of an immune system and a bit of protection against really bad big decisions from some giant thing that only a big corporate that’s the other thing. Only a big corporation can build a giant 200 billion parameter AI. I mean, you and I can’t build it. And so having the bottom up allows everybody in the world to participate, and I think that’s a big protection right there. All cultures, all languages, the AIS speak over 100 languages already. So, yeah, I think that’s really what we want. Sometimes people ask me, what do you think about the advancement of technology? I go, well, we’re the product of 4 billion years of evolution. We have the whole fossil record. We see how life has evolved, and it keeps getting smarter and more capable over time.
01:14:09
Speaker 2
Do you think that evolution has stopped with us? No, it’s going right through our fingers, right onto the keyboards and right into the code. So we’re writing code that is evolutionary over other code, but we’ve written code that has the mathematics that it learns to evolve itself. So we’ve baked evolution into the code. So it’s pretty cool.
01:14:32
Speaker 1
Well, Dan, as always, it is such a pleasure to talk with you and thank you so much for taking the time out today to be on our show. And I’m so excited to see what’s coming. You’ve mapped out this incredible future, and a lot of people, they get really kind of intimidated by technology. I’ve always welcomed it. To me, I feel lucky to be alive right now. And people like you really reinforce that for me.
01:15:04
Speaker 2
Well, I mean, obviously, if you’re a human on this planet, there’s no escape from technology. So it’s an effect of life, like air. And so the question is, do you let somebody else do it, or do you step in and really bring your A game and try to do something extraordinary? And so it isn’t even pro or anti technology. It’s just that this stuff is burgeoning up, get the best people you possibly can and do the best thing with it. That becomes then the model of how to go forward. Otherwise, you’re just sitting back and kind of hoping that it all turns out okay.
01:15:38
Speaker 1
Yeah. And I have always loved your ethos in that way, and I really appreciate you and others like you that are in this and doing the right thing, and you have this vision for humanity that is just I think it’s what we all want. So I appreciate you. I appreciate what you’re doing. And thank you so much for being here with us today.
01:16:01
Speaker 2
Thanks for having me on. It was really awesome conversation, and your knowledge of this field is really great. So really very enjoyable to talk with you about it all.
01:16:10
Speaker 1
Thank you so much, Dan.
By understanding and adopting Active Inference AI, enterprises can overcome the limitations of deep learning models, unlocking smarter, more responsive,…
The initiative between VERSES AI and Analog in Abu Dhabi isn't just about technology - it's about reinventing urban life. By incorporating…
"We're optimistic that RGMs are a strong contender for replacing deep learning, reinforcement learning, and generative AI." - Hari Thiruvengada, CPO, VERSES…
VERSES AI is transforming AI, shifting us towards First Principles AI, Embodied AI, and Decentralized AI, distributing AI compute to…
Updated - Over 150 New Entries for 2024 Q2 - Research papers, articles, videos, who to follow, and more! The…
VERSES AI has unveiled a groundbreaking framework that "sets the stage" for ASI, and could have profound implications for the…
In a panel hosted by the Financial Times at the World Economic Forum in Davos, Switzerland, two of the biggest…
The Most Complete Repository of Research Links and Educational Content on the VERSES AI Technology & Active Inference AI This…
Computing is changing on every basic and fundamental level. Generative AI is one piece of that computing puzzle. It’s a…