Subscribe
Share
Share
Embed
In-depth discussions and explorations into the whole new world of web3 and the metaverse with leading thinkers and industry experts. Presented by the NYU Metaverse Collaborative.
The Whole Metaverse podcast - Episode #11: "AI Inside the Metaverse" | with Dele Atanda
===
Pierre: Welcome to the whole Metaverse, the NYU podcast about Metaverse Web three, blockchain and nft. Hosted by Dr. Elizabeth Haas and myself Pierre Gervois. Today we are going to have Dele Atanda, an entrepreneur, a c e o, in the web three and, and blockchain space and, and I would say a visionary, and we're going to talk in particular about the introduction of artificial intelligence.
In the Metaverse, and this is a truly fascinating, podcast, and you're, this is a must watch because our guest is bringing us new dimension of what can be done in the metaverse using artificial intelligence, who is going to, boost and to supercharge all the already existing capabilities of the metaverse.
So we encourage you to, to watch the podcast.
Pierre: Dele Atanda thank you very much for being on the whole Metaverse podcast, the NYU podcast, exploring Metaverse Web three, NFTs Blockchain, Dr., Elizabeth Haas, and myself, Pierre Gervois or Co-host de You are an entrepreneur, CEO of metaMe. You're the creator of the Meta Nice Crypto Comics, so. When did you started to get interested in crypto and the metaverse?
Tell us how it started for you.
Dele: Thank you so much for having me. The Metaverse started a while ago. I was interested in Second Life when that came out and was playing a lot there. and then I, you know, I saw that ironically as Web three. I wrote a book back in 2000 and well published in 2000 and. 12, around web three, and playing with Second Life, really.
Let me see that this was gonna be the next big evolution, this immersive virtual world. It was the natural sort of progression in terms of the digitalization of media and experiences. And I was working for Dijo at the time. And we built like an enterprise. Meeting space within Second Life, where, you know, we had a distributed team across four continents that met regularly, like at least once a month.
And we were using WebExes to meet at the time. So the idea of creating a virtual meeting space where we could have presentations and conference links in was really at the time very progressive. So we built an example of that with ibm. And that really kind of got me into the business understanding of metaverse as a, framework, these immersive 3D spaces.
So that was where my metaverse interest really began. on the crypto side, you know, it was really, so from my interest in data, because as I was writing about Web three and the emergence of web three, I started to realize there were big issues around privacy and data and data sovereignty and data ownership.
And I had been looking at technology solutions to that problem for a long while, which was a very hard problem to solve technically. I first became interested in the crypto space with the arrival. Bitcoin and for me, actually what was really interesting about Bitcoin was this story, this story that, you know, this guy, Satoshi Akimoto had created this thing and this really disruptive currency and that had disappeared.
So I found that really fascinating and, and I kind of observed it sort of from the sidelines for a long while and, you know, dabbled in a bit broader little, but it was really when Ethereum was born with the arrival of smart contracts. That I started to see the potential of the technology to solve real problems around decentralization with smart contracts and immutable programmability.
That really opened the pathway to particularly solving issues around data sovereignty and data ownership that prior to that, we didn't really have a technical solution for.
Elizabeth: How, do you think AI is going to impact what we can do?
Dele: It is very complicated question, to be honest with you, but I mean, the short answer is in multiple ways. I do think we need to unpack AI a little bit because, I think it's a little bit of a misleading term, you know, and, and the intelligence that machines have is debatable. But that said, you know, the ability, machine learning and deep learning as abilities to rapidly accelerate process information dramatically is going to impact everything.
So, In many ways I see the technology world in relatively simplistic terms in the sense that data is essentially the fuel and AI is essentially the engine. So I think the engine of ai, which is, you know, whether you wanna call it computing dis, whatever format it takes, I think is is going to become predominant.
It's sort of like, it's a degree of intelligence, if you will, that I think is going to. Permeate the entire digital landscape. You know, it's kind of like the next evolution of the digital revolution in many regards. But I think that, so, you know, the, the, the machine if you will, cuz it It is helpful to think about the internet as a single machine.
You know, that's why it's actually so resilient. It's kind of the most reliable machine that we've ever created that man has ever created. And I think that machine is gonna become more capable and more, able to process rapidly. is that intelligence, possibly. there's, an element of intelligence to it, but I think we get lost a little bit too much in that term intelligence.
And quite frankly, when you look at the things to be concerned about with machine learning and deep learning, and this rapid computer processing capabilities is the unintelligence of the machines, actually, that's perhaps more frightening. In terms of when you look at the impact of it, than the intelligence of it,
Pierre: One aspect of AI in the Metaverse that I found absolutely fascinating is, is the ability now with some tools and very simple prompts to build 3D buildings and elements and one. Of the issues of the, the first iteration of Metaverses is that very often you got vast empty spaces, and one of the reason that it costs money, you need to hire 3D artists who are going to spend a lot of time.
It's extremely time consuming to build a building with a lot of details or to build. Trees or any elements. And now with, with these new tools in a few minutes, you can build an entire city with an incredible level of detail. So I think that the eruption of AI in the metaverse happen at the right moment, and it's going to help a lot, all the builders of the metaverse.
What, what do you think? Do, do you
Dele: Yes, I do. And I think, you know, I mean, again, this is, and you are absolutely right, this is something we saw in the early Metaverses in places like Second Life. They are spatial, and so you are right if it's empty, if there's nothing there, that's a really terrible experience. But I think the other thing as well that is really critical in that sense is people, if there's no one there, that's really critical as well.
And I think particularly generative AI in the ability to create. You know, you see this in gaming, what they call non-player characters, you know, people like, entities within these experiences that can mimic people. And that, you know, in gaming it's, you know, it's fine because that's, you are, you're playing a game and you have these non-player characters and they become increasingly intelligent and you can, you know, model them on different types of behavior and different kind of outcomes.
And I think those will be, Prevalent across the broader metaverse as well. If you can imagine going to a music festival, For example, if there's only four people around the band, it's gonna be a pretty, you're gonna feel pretty lonely. And it's not, you know, you're not gonna have that sense of, wow, I'm at this really exciting concert.
But you know, if you have a thousand, a hundred thousand and you know, characters there, some of them human, some of them non-human, then the experience becomes a lot more immersive and a lot more compelling. But of course that then throws up more ethical issues around, well, should, you know, when you are engaging with an machine entity, whether that entity's real or whether it's not, and you get all of those issues that come as a result.
Pierre: So, at the beginning of last year, 2022, very few people were, were talking about chatGPT. It was not like completely in the mainstream. And now, Everybody's talking about chatGPT or seeing the, the amazing things that we can do with this, program. I mean, how do you think it it is going to, impact the future?
Dele: Yeah, I mean, so it's funny actually cuz I mean we were working with OpenAI and G P T three back in the pandemic. and interestingly enough, we started talking with them around fictional comic metaverse gameplay experiences, and then we actually started looking at health. Conditions and health data sort of monitoring and how it can work with that.
I think the, the really interest, I think MI Microsoft have done a brilliant move in terms of partnering with chatGPT because I think, you know, integrating chatGPT into all of these work tools like PowerPoint, word, Excel, you know, the, the machinery of modern commerce right, is brilliant and I think it is going to be transformative in terms of.
Our work, you know, our ability to create presentations, create narratives to be creative and to be productive. you know, it's gonna be a game changer. And in the same way that computers, you know, I, I mean it was, remember the typewriters and there were word processes and then there were computers.
It's kind of the next evolution in that regards. You know, I think as a productivity tool, it's going to be transformative. Generative ai particularly because, and the thing about generative AI in that sense is that it deals with all that sort of unstructured space. You know, around storytelling or narrative creation or, you know, brief presentation, create, you know, writing, blogs, writing podcasts, all of that stuff that is so important in this.
Communication space. I think that's perhaps the key thing. It, it really plays such an important part in the communication framework using natural language processing and whatnot. And that's such a critical part of how society operates. it can't be overestimated, the importance and how transformative it's going to be, but it then obviously throws up all the questions and all the concerns that come along with that
as well.
Elizabeth: how does that translate into the metaverse? So if we're, in a metaverse based meeting or, you know, there's a metaverse event, how will the, generative AI or chat G p t influence that?
Dele: So I think the, in a couple of ways I can see it happening. So I think the first thing we, we have to put in perspective is that in the metaverse, there is vastly greater volumes of data than in the analog world. You know, we talk about IOT and you know, the mobile world, trying to connect the physical world with the digital world.
But we're still quite far off for, you know, iot provision and you know, smart cars and thermostats, kind of as far as we've really got at the moment. But in the metaverse, everything is digital, so there's a huge amount of data, and I think it's what we are going to see increasingly is this. Correlation between AI and data co-dependence in many senses and more from an AI perspective.
You know, data's a bit more, it's got intrinsic value. AI's really highly dependent on data in terms of the, it's efficacy. It's even in terms of how the AI itself is created and then how effective and efficient it is. So, you know, you have this garbage in, garbage out rule of computing generally.
but then even in terms of it, it's an interesting correlation between data and AI because data shapes the AI itself as well. you put that now in the metaverse and you know, you can start to see that AI could become much more capable. I some examples, for example, self-driving cars, there is a lot of work that's being done.
In terms of creating, better algorithms, better performing out AI algorithms for self-driving cars by creating simulations in virtual worlds, in, in metaverse worlds effectively, because you can train those algorithms much more rapidly. You can accumulate much more data much more quickly, those metaverse like environments.
So I think there's sort of an inside aspect to it, which I think. You'll have more data and you'll be able to train that data more effectively. in Metaverse, you'll be able to create simulations and train ais on those simulations very effectively in the metaverse. But then I think also what we are going to start to see is the metaverses themselves will start to become intelligent.
So you will start to have an AI that oversees the entire metaverse space and. I mean we're already talking about, Microsoft's AI versus Google's AI versus Facebook's ai. So you can understand, I mean, you don't need to look further than Facebook and it's big moves into the metaverse space.
Facebook's. Horizon complete virtual environment soon will have a ubiquitous ai. Siri for Apple, for example, will become this ubiquitous AI for their metaverse spaces. So I think what one of the issues that this is going to throw up very soon is this whole debate between centralized versus decentralized metaverses and centralized versus decentralized data.
and then centralized versus decentralized ai. So I think this is gonna become like a new trinity effectively, data AI and metaverses, around this debate of centralization versus decentralization.
Pierre: let's talk about education. Now we can have a class in the metaverse with students and the, the human professors as avatars. And now with artificial intelligence, we're going to, multiply all these. Capability in the education space. So we can imagine students as avatars, professors, as avatars, and then an AI tool who can give more, you know, work with the professors and students to do more case studies to bring even more information.
So how do you see. AI and the metaverse using ai, having a role in education from high school to higher education.
Dele: so I think you know, very much like how you outlined it. I think the first, thing to sort of accept is that AI will become central to that educational experience, particularly digital education tools. There will be AI imbued in it, in terms of the environments, in terms of the learning materials, in terms of the.
operating models, the teaching methodologies of teachers and the learning methodologies of students. so I think that's the kind of foundational perspective, and I think that it will optimize, you know, be able to optimize the output of teachers, so that, AI for example, might take a teacher's cave, a teacher's framework and present it to different students in different ways based on.
Students particular learning type, for example. and of course the ability to amplify that across time zones is take a simple model, right? you could have asynchronous, You could have a teacher deliver a class and deliver that class in real time to people within that certain time zone.
But you could then have students in a different time zone participate in that class, and based on now giving the teachers lesson an AI extension, like a halo around the actual core lesson. Students in this different time zone could ask the teacher questions in, delayed time, and get responses in real time.
Based on the teacher and the teacher's AI sort of responding to the student together, right? So you can kind of conquer this time difference by using AI as a bridge effectively between them. Right? So I think that there's a lot of really exciting opportunities that are coming out there. But for me, I think the thing that I think the, where this is going to get really interesting and where we get most effective is when we start to have.
Our own personal ais, you know, our own personal ais that can effectively connect us with, system ais. So imagine a student, having a personal AI that effectively. Understands their learning model, understands their actual course frameworks and their interests and all whatnot, and is able to then interact with a teacher or institution AI effectively to really get a maximum learning capability, between both a maximum transfer of knowledge and information between both.
And that's gonna take. two different types of ais to be able to facilitate that kind of connection. And I think you know, you can write that across any kind of organization and individual, city versus a citizen, an enterprise versus a consumer, an enterprise versus an employee, et cetera.
I think this large, small, macro micro AI framework is where things are going to get really interesting and rich.
Elizabeth: As a city, if you were a mayor of a city, what would you be worried about and where would you be proactive with ai?
Dele: So I think you have to, and I think all the stuff that's really current discussions, right? I think the ethics of AI in terms of discrimination, bias, misinformation, toxicity, all of those things are really critical. But I think. when AI starts to drive city decision making whether that's distribution of resources, whether it's allocation of housing, whether it's.
the criminal justice system, whether it's, you know, fine management, all of these things, then the stakes go up considerably in terms of the performance and the propensity for error or getting things wrong. So, for me, like I say, I think. I'm less worried about the intelligence of the systems in that sense.
I'm more worried about the unintelligence of these systems, which is masqueraded as intelligence, right? If we abdicate all responsibility to the machines to make decisions about issues that, are socially complex, then there's, it's problematic because I think this is a thing that we are seeing already from.
Chat, GPT out in the world and generative AI out in the world. These systems, they're pattern recognition systems and they can replicate patterns that very effectively to the extent that we as humans may not even be able to differentiate between something that's human generated versus machine generated.
The classic curing test, the curing test is no longer, I think, appropriate to determine whether. AI has achieved a degree of, comparability to humanity, because what the problem with is that machines can mimic humans as we're seeing with chatGPT but they can't make judgements.
You know, machines don't know the difference between right and wrong. They don't know whether something is socially appropriate or inappropriate. All they can do is just see patterns around what kind of conversations or what kind of commentary has led to what kind of responses, and replicate that as best as they think possible based on the remit they've been given.
You know, are they trying to sound more human or not? So I think that, you know, you can't put the genie back in the bottle. And, you know, this pause of AI is, is not viable. But I do think that we need to, in a sense, right size it, you know, we are very excited about what the potential is of it and rightfully so, but I think we need to put AI at service of humans, and be cautious where AI is making decisions.
For humans, you have to really make the ultimate decision maker a human, you know? And that's the balance that we have to police most carefully. In the short term, what is the balance between decision making that is abdicated to machines, and where is the human oversight of those decisions and the human ability to intervene or correct or adjust those decisions according to the.
Elizabeth: If If I'm trying to really take a metaverse perspective, not an AI perspective, I've heard a couple things that are kind of interesting to me. One is, metaverse serves AI in that it gets a richer set of data than it would without Metaverse, more than AI serves the metaverse. Is that
Dele: Absolutely,
you know, 100%. I think, you know, like I say, I think AI is the ultimate engine of the 21st century, and, you know, everything else is data to ai. Basically, the AI sees the world as sources of data, sources of information that it can learn from, that it can improve on, and it, you know, ai, will struggle.
To differentiate between the metaverse and the real world. You know, in that sense it's sort of like, in a sense the human mind with dreams versus reality. You know, it's, it kind of, it's only the conscious mind that can make this distinction, right? The subconscious mind doesn't differentiate between dreams versus reality.
So I think absolutely the metaverse will serve ai. And as a richer source of data, and it will, and AI will be highly dependent on the metaverse in that sense, and industry will become highly dependent on the metaverse to train AI in this sense. But I think ultimately the question that we, the thing that we just need to really police very carefully right now at these early stages is to make sure that the AI works for us and that, not that we work for the ai.
Elizabeth: The other thing that I heard you say is that, and I just wanna make sure I'm hearing this properly, that, AI will make the metaverse easier. It'll make it richer. So AI makes the metaverse easier and richer. The Metaverse makes AI more powerful, and we're in this cycle where the two kind of turn in a way that really needs controls. And I just wanna make sure I'm hearing this properly from your perspective. Not, not
Dele: No, I think you, you're absolutely right and I think Pierre hit the nail on the head with, you know, you look at generative AI's ability to create, models, create buildings, create environments, create characters, create people effectively, you know, synthetic people, metahuman. just going to increase.
So. with ai, we'll be able to create very compelling metaverses that much more quickly, much more rapidly. Just from spoken instructions, someone will be able to sit in front of a computer and say, create me a massive sort of pantheon like temple. And I want it surrounded by, flying, Pegasus drones, Pegasus horse, like flying horses, drone kind of characters, and it'll be able to generate those kind of things.
So on a fantasy world and on a, actual world basis, Will be able to create much more compelling environments and then run studies, like I say, with autonomous vehicles, for example, create me a, a mapping of I don't know, the lower East District of Manhattan, and simulate traffic at 5:00 PM and show me what happens if there's a road closure.
At between, you know, 12th and then second and you know, third avenue, what, impact does that have on the flow? We'll be able to simulate those things very quickly, just from those kind of instructions using generative ai. So yes, it's definitely gonna create this much more rich cycle and spiral.
as a tool is fantastic, but that could be used for good and bad, I think is the call out.
Pierre: Listening to you, you are talking a lot of ethics, a lot of question that are going to arise, and until now, I feel that people who are invited to discuss about metaverse and AI were more engineers, scientists, I is it, I think it, it might be time to invite philosophers at the table to invite expert in ethical question.
And ask for their advice because I think we would need their expertise. Right now. What? What? What do you think?
Dele: I could not agree with you more. I think it's urgent. to actually bring this, because, you know, I'm, I'm very much from an engineering background or from a technical background, put it like that as much. I, had the fortune of having sort of traversed a creative path and a technical path.
But, so I understand the technical mindset. I understand the engineering mindset, but the problem with engineers in that sense is engineers are very binary in the way that they see the world. You know, everything's zeros and ones. It's always, it's either black or white, it's on or off. they see the world in this very kind of binary state.
The world is not binary. You know, the world is complex. You know, it's, you know, the multiple shades of gray. That's really where all the issues are. So, and you've seen, look at Twitter with Elon Musk, who is arguably one of the best engineers on the planet. Right. But when it comes to society and when it comes to social issues, it's much more complex than machines.
You know? It's like, you know what's, what's the, what's the running joke? It's harder than rocket science, right? This is not rocket science. It's a lot harder than rocket science in that regards. And I think the dangers, which he calls out, and I, think, you know, I have a complicated relationship with Elon Musk, I have to say, but I, I, I think that the concern that he points to about, you know, super intelligence for machines is less of the problem.
I don't, I do think that there is a very real problem with ai, but it's more about what is the intent? Behind the design of the systems. It's more about what are the people who are designing these systems trying to achieve, and what considerations for unintended consequences are they making when they're going about implementing them.
I'm less worried about AI sort of running and kind of wanting to exterminate all of humanity. Then I am about. bad humans using AI to accelerate bad agendas, much more effectively. That to me is a much bigger concern. So I definitely, and the only way we avoid that is to put these ethical guardrails around these systems right now.
And, if you think of it in that context that it's more the people. Behind the systems that and their intent and what they're trying to do with the ais that need policing than the AI itself, then it makes it a more manageable problem as well.
Pierre: Thank you so much for all these very interesting insight and, and, and your view about, about AI and its interaction with the Metaverse. So on behalf of nyu, Dr. Has and, and myself would like to, to thank you very much for, for your time and, and sharing with the viewers these very important, elements.
So thank you very much. De. And we wish you, a wonderful day and good luck in all your
Dele: Thank you so much. Thank you very much for having me. Very exciting discussion and looking forward to seeing you all again soon.