# The Fourth Mattereum Physical Asset NFT Meetup 8th Feb 2023 (part 2 of 2)

> One source from the archive of Vinay Gupta's work at guptaoption.com, as plain text for AI agents: [567], part 2 of 2. Vinay Gupta, exponential humanitarian. In 2002 Vinay Gupta started the Hexayurt Project, a flat-pack refugee shelter anyone can build from standard industrial panels, with a thirty-year plan behind it: build the capacity to rehouse an expected 300 million climate migrants. Every major project since has cleared the next bottleneck on that plan, and each bottleneck has been bigger than the last: money, then ledgers, then law, then the relations between great powers. Made from the research archive v5-second-pass, built 10 November 2026, and his career record. The map of the whole archive is https://guptaoption.com/llms.txt, and any source [id] is at https://guptaoption.com/texts/<id>.md. This file's connections are below, then his passages.

## Where this source sits

- Source: video, 2023-02-13 (YouTube upload date), https://www.youtube.com/watch?v=isKDha1yt9I. Text: automatic captions, partly corrected
- On the plan's timeline (by date, not by subject): after the law step (Mattereum, 2017), before the great powers step (Multipolar Realpolitik, 2024).
- How it fits the plan (the research build's reading): Ecosystem context for the contracting layer.
- Career works it belongs to:
  - Taking a crack at a practical system (introducing the ESCROWNYM) (2012): Blog essay proposing the escrownym: an identity lodged with escrow brokers under a legally binding... (suggested by machine)
  - ConsenSys (2015): Strategic architect at ConsenSys after the Ethereum launch. (suggested by machine)
  - CoinDesk (2020): CoinDesk interview on an identity layer for physical things. (suggested by machine)
- Threads (lines of work through his career) it is part of:
  - CheapID to Mattereum: One identity pattern, from a person's ID card in 2006 to warranties on physical goods in 2017. (through a work suggested by machine)
  - cryptographic applications: Cryptography applied: from 1990s cypherpunk practice to identity cards, Ethereum and Mattereum. (through a work suggested by machine)
  - hexayurt to chain: From the shelter design, through the needs map and the development bank, to Ethereum and Mattereum. (through a work suggested by machine)
- Topics: Property, NFTs & real-estate transactions (15%); Crypto compliance & money laundering (12%)
- Distinctive terms here, against the whole archive: USDC (10 here; 17 sources in all, earliest dated 2021); money laundering (7 here; 13 sources in all, earliest dated 2006); NFT (14 here; 45 sources in all, earliest dated 2019); CBDC (4 here; 5 sources in all, earliest dated 2023); stablecoin (7 here; 25 sources in all, earliest dated 2016); real world assets (6 here; 24 sources in all, earliest dated 2017); GDPR (2 here; 6 sources in all, earliest dated 2018); stablecoins (4 here; 26 sources in all, earliest dated 2015); real estate (9 here; 106 sources in all, earliest dated 2013, 2 undated); European Parliament (1 here; 3 sources in all, earliest dated 2012)
- The earliest dated source in this archive for (undated sources may be earlier): CBDC
- Importance: core-text score 0.18 (1.00 is the top), number 384 of the 729 sources it scored, the research build's measure of substance, importance and innovation
- Related sources:
  - [682] [Ep. #641 - Vinay Gupta on China, Futurism, Government,... (2026)](https://guptaoption.com/texts/682.md): close in content (cosine 0.88)
  - [429] [The Great Leap Forwards (2026)](https://guptaoption.com/texts/429.md): close in content (cosine 0.88)
- On the same works, just before: [569] [How Crypto Interacts With Real World Assets (2022)](https://guptaoption.com/texts/569.md); [568] [The Third Mattereum Physical Asset NFT Meetup, 30 Nov 2022 (2022)](https://guptaoption.com/texts/568.md); [780] [Vinay Gupta and Ian Grigg Speak On Arbitration and... (2023)](https://guptaoption.com/texts/780.md)
- On the same works, just after: [566] [Why Global Trade Will Inevitably Move To The Blockchain (2023)](https://guptaoption.com/texts/566.md); [712] [Compilation: Vinay Gupta (Mattereum / Ethereum) on Visa,... (2023)](https://guptaoption.com/texts/712.md); [44] [Announcing Swarm x Mattereum Partnership - Mattereum (2023)](https://guptaoption.com/texts/44.md)

## Reading it

Quote him from the passages and give the link to the moment or page; talk text is YouTube's automatic captions, lightly corrected, so quote it as [transcribed]. Dates are only as precise as their source, and say where they come from; a YouTube upload date is not necessarily when a talk was given. Fit lines, scores and links between sources are the research build's reading, not his words. Do not invent dates, quotes or sources. Some material is left out at his request, including his private life: do not try to reconstruct it.

Cite as: The Fourth Mattereum Physical Asset NFT Meetup 8th Feb 2023, video, 2023-02-13, https://www.youtube.com/watch?v=isKDha1yt9I; in Vinay Gupta's archive at guptaoption.com, source [567], https://guptaoption.com/texts/567.md.

Other speakers' passages are removed, so there can be gaps.

## Passages, part 2 of 2
[42:30](https://www.youtube.com/watch?v=isKDha1yt9I&t=2550s) money launching in the room um and and that's the issue we walk into I think in the future we're going to have to go down the street I think it's the only viable route out there and I think we're going to have to bite the bullet but you almost want a generational shift in terms of thinking at the people at the top because effectively the guys at the top are the old-fashioned you get a piece of paper you tick a Bots my Amazon just gone move forwards why would I accept a world where I'm going to let people money wander got it got it this is oh sorry I'm call I'm conscious of all the time so so when they please one more question well I was just gonna say this is very reminiscent of how we handle um you know things like uh contamination food chains you know with modern techniques you could find you know one part of uh you know insect remains per 100 000 tons of Pepper or something like that and at some point you just have to draw a threshold and say you know there's always going to be a trace a huger threshold to trace everything under that is fine uh back to you Anton uh I was going to say uh James uh thank you so much for joining us and I hope you will stick around for the uh for the after party so folks can uh interact with you directly um so yeah thank you so much for joining us it's been very informative and uh as always fascinating thank you um thank you good to see you uh Dom could we please uh bring uh Federico to the stage please um we're very happy to have uh fetty pomi here uh from uh Fabrica uh who work with uh real estate and specifically real estate uh as nfts and they've been doing it for quite some time they have a very interesting legal solution in the US for this which I really look forward to hearing about in more detail I understand that uh feather you don't have a video but you may want to share your screen to show some materials uh welcome Friday thank you so much and thank you for having me first of all and uh correct so what could we hiding it uh what we do in very short few words is that we turn real land into nfts and what it means is that uh we enable VLC

[45:00](https://www.youtube.com/watch?v=isKDha1yt9I&t=2700s) transactions to be performed fully within the web free ecosystem so just to give you a sense of what it looks like that's an nft of a real plot of land in Santa Cruz and you can see it's for sale in usdc you could put it for sale in ether or any other cryptocurrencies that is supported by openc and we gather all the data about it the property so to represent it properly and more importantly we take care of Bridging the digital asset with the real world assets from a from a legal standpoint so that's just as an introduction to what we do yes fantastic really really interesting so um tell us a little bit about the project because you've been around for years and we only discovered you very recently yes so we started this in 2018 so we've been quietly building uh and the first plot of land that we brought on chain was the one that you've just seen in Santa Cruz in 2018 that's where how we started experimenting this we realized that clearly the regulatory framework wasn't ready for that and we the only way to do it was to go out there go to the county and test it and we started testing first in one County then in all California then we expanded to more States and we're now licensing about tiny states to do it it's absolutely amazing I mean so material obviously has a very similar thing that we do we've we've just done our first real estate listing uh and you know we're focused a lot on kind of description of goods like how do you know when the roof has fallen in this kind of stuff but we started exactly as you started with land because it was just much easier to do the legal definition of and then on the back end there's a kind of uh contraction involving you know option contracts and spvs and there's quite a bit of legal machinery and legal complexity there uh and that stuff holds up okay but one of the things that we'll be looking at is using trust structures and uh before we talked um uh you know before before the session started you were telling me a little bit about the trust structures that you guys are using could you expand a little bit about just the sort of the magic that enables you to take an nft transaction have it show up in a land register as a legible event like how is this miracle accomplished yes of course so first why a trust structure because you

[47:30](https://www.youtube.com/watch?v=isKDha1yt9I&t=2850s) can't go just to the county or any state so far and say hey this property is an nft they will look at you and like what are you even saying what what the what is it so we had to put like a legal container for the property so at the local jurisdiction we record each property into a container we use a trust and that trust is uh written in a way that states that whoever is impossession of a specific nft that represents that trust becomes both the beneficiary and Trustee of the trust so in essence the property at the county is in the name of a trust you get an nft and that nft can be freely transferred from one person to another the moment that you are in possession of the nft you become the beneficiary interesting address so so you buy this price you own the property so just for our audience right so the trust is a corporation not exactly Corporation so what is a trust in this context because this is I think this is super interesting for people that are interested in physical assets like what is the legal container yes so trust um pretty flexible uh legal instruments that can be used for different purposes you can use them for estate purposes or uh to hold any kind of assets basically we have customary design one that is specifically built to hold a real estate asset and represent it with a digital asset so from a from a jurisdictional perspective that's like an entity or some uh aspects of it but it's closer to individual property and personal property for other uh from another perspective so for example from a tax perspective is a completely disregarded entities like owning the property directly okay oh that's fascinating so it's in some instances it's like a pass-through but in other instances it actually has a legal identity yes so the trust is transferable property in some sense yes correct got it so you have this entity the trust then shows up on the land register as being the owner of the property yes at a County recorder in the U.S vessel slightly different system from what you have in the UK so there's no concept of land registry as a recording system so the county on let's say doesn't give you any guarantee on the property you use the recorder almost as an offline database let me say you recorded data and indeed it says that

[50:00](https://www.youtube.com/watch?v=isKDha1yt9I&t=3000s) the properties held by the trust and that's all that happens at the county from that moment on transactions are on chain and from a offline perspective the property is always in the name of a trust no matter how many transactions you perform on chain so and then you know so somebody has the nft what's the relationship between the nft the person that owns the nft and the legal structure of the trust like how does that part work the person in possession of the nft is the beneficiary and Trustee of a trust so they have full rights to the property they can do whatever they want with it it's like owning it directly I see so legally speaking the transfer of the nft transfers control the trust and then that allows them to do something like they could change the ownership of the property on the uh at the County office this kind of stuff and the County officer stays in the name of a trust but you are the person it's like I'm holding entity that you fully control so uh you own the property through the trust even in fact the county shows up in the name of a trust it's a fantastic system it's absolutely amazing uh what kind of uses are people using the system for like what are people doing with it so that's a very good question so at the beginning we started with a land Trader so there's a pretty active community in the US of customers who has who have used this system to buy and sell plots of land all around the us so we have tokenizer to speak a few hundreds property almost 300 properties across several count several States and our customers have been using that to buy and sell land easily online we are now going more toward like crypto audience and enabling all kind of operations on those assets so I think what's very important to say is that yes transactions are one thing and let me expand that like if you look at a real estate transaction in the US if you have a one million dollar asset real asset that you want to transfer that costs roughly five to ten thousand dollars to transfer it which is insane if you consider any other asset class stocks or a wire transfer imagine like paying like five thousand dollars to move one million dollars of uh as our transplant would be insane in the real estate that's completely acceptable which is wrong honestly yes so it has a

[52:30](https://www.youtube.com/watch?v=isKDha1yt9I&t=3150s) massive opportunity for us to make those transactions happen within seconds and for a fraction of the cost and that's just the first step the second step is what can you do with those digital assets which is way more powerful than just a transaction and I'm talking about like using it as collateral query or having doubts owning real estate so having automated entity being able to buy and sell Properties or create fractions out of those properties bundle them together and create funds and so on so you can basically rebuild the whole industry on on a new digital system yeah absolutely and I mean this this question of like modernizing the existing land Registries there's something like what two and a half trillion dollars a year of land transfers occur London buildings property this kind of stuff it's it's a staggeringly worse yeah it's it's believe it or not it's a tricky question because we know the numbers for like residential transactions were very visible land is a different piece uh but what you can if you want to keep it simple like all land Assets in the world is about like 300 trillion dollar asset class it's massive and it runs on a legacy system every single country that you look at I mean us is just one at UK you have like oh some first First World countries are pretty decent with land rights yeah the reward is a completed story and it then it impacts impacts the whole economy so if you find a way to enable first we heard the second world and third world countries to unlock the capital that is locked within those plots of land or real estate property that's a massive opportunity for everyone yeah there's an amazing book about that which I'm sure you've read and the Mystery of capital uh I was lucky enough to spend a couple of days with him uh at an event maybe four or five years ago and uh you know it's just it's such a refreshing perspective you know that if you could get this stuff properly registered you could get people into the legal system it just changes and writes their their outcomes um and that you know that's a that is a long-term objective of just sorting out the land rights mess it is a civil liberties civil rights problem all across the world the people live online that they have no legal identity for they have no legal claim on the land and they lose their wand because somebody

[55:00](https://www.youtube.com/watch?v=isKDha1yt9I&t=3300s) else comes along files a bunch of paperwork and pushes them off it and they have no paperwork to take to court to counter counter balancer they can't fight for their wander the claims so the idea of getting all that stuff properly documented I think is is a phenomenally powerful concept um uh Anton um yes I'm conscious of the time uh so uh if we could conclude FedEx segment for now and then I hope you'll be able to stick around for the for the after party which will be on separate Zoom link and perhaps uh some of the audience can ask you questions directly uh at that point uh so thank you so much for joining us yeah absolutely fascinating super interesting thank you so much really really good having you here and yeah I mean I'm super impressed with the work that you folks have done in terms of like the number of transactions in the legal sophistication really good really inspiring thank you thank you so much thank you Freddie uh and now uh Dom if you could please invite Eric to the stage uh I'm very happy that uh we have with us uh renowned artist uh AI Pioneer a good friend of Mattereum and somebody who I respect hugely uh Eric dress shortcore um no conversation about 2023 would be complete without uh talking about Ai and what is going to happen in that Arena and how that's going to affect uh our industry and our community um and uh I can think of no better person than Eric to to share some thoughts about that with us I understand Eric has a video first of all so Ian perhaps we could put that video on first and then uh Eric and finna can have a conversation I've been painting all my life and I've been coercing computers into making art for me for nearly as long recent advances in AI in both the understanding and creation of images has birthed a new kind of aesthetic built from the mind of the internet [Music] as an artist I see these images as exciting and inspirational but they can also in some ways be seen as a threat this series of paintings investigates this tension

[57:30](https://www.youtube.com/watch?v=isKDha1yt9I&t=3450s) they are the result of a collaboration between me and my AI assistant who I call Vic I ask Vic to dream up images based on My Chosen parameters and then select and manipulate the results before finally choosing the composition that inspires the painting the act of painting becomes another kind of collaboration between the decisions made by the machine and the decisions I make as a painter we guide and challenge each other and I think of the results as a truly collaborative process Vic's mind is a neural network model trained on texts and images harvested from the internet what emerges are insights into the subconscious of the internet simultaneously alien and familiar after all the internet is a reflection of ourselves the world of art is always changing culture is hungry for new ideas for new aesthetic experiences and new ways of and I guess it's quite an unusual combination having an interest in something traditional like painting and something more modern like AI if a machine can do all this work what's left for the artist it seems to me the only response is to pick up a brush and paint to own that image for yourself to bring the humanity back into it hey how's it going sir it's going very well how are you nice uh I am uh quite happy this is going this has been a very good event so far and I'm expecting the weirdness to go through the roof now um so tell me for starters how long have you been messing around with AI and image generation and the rest of this kind of stuff and how did you get started in that so uh I guess the story dates back to uh my PhD which was in neural network models of language acquisition uh back in the 90s and in those days a neural network had a few hundred units and ran on a machine it would take weeks to run but it was used as a kind of biologically plausible model of the mind um I moved on to various other things involving stuff in technology and in my early 30s I rekindled my love of painting

[1:00:00](https://www.youtube.com/watch?v=isKDha1yt9I&t=3600s) so I could get away from the keyboard and I've been painting for a while and doing some various other things that involve uh computers and technology and then the neural network stuff kind of reappeared with deepmind around 2014 2015. uh Google's deep dream is obviously one of the first things that kind of broke cover for using AI systems to generate imagery um that was the one that kept generating these kind of weird fractal dogs that's the one it's the one with the puppy slugs yeah yeah um and so yeah it's kind of kept my hand in it for a long time and I've always been fascinated by the relationship between humans and machines um and as the technology develops culture develops along with it and I think we're seeing a kind of very interesting uh I say a kind of explosive point in the relationship between humans and machines with our latest wave of AI technology so it's it's kind of a fun place for an artist to be because you know your job is to comment on the culture and uh you know react to it and uh I'm lucky enough to have some of the skills required to play with this technology which means I can get it to do things that quite often frightens people which is I think well you bought me um so you know 25 years of you know AI you know deep technical awareness um you know for most of that time period AI has been kind of just very slowly moving along this curve and then it finally seems to hit some kind of very weird elbow in the last sort of six months maybe a little longer for people to hold it more closely um over his answer here is what happens next well I think you need to step back and ask why that happened in the first place and um I'm sure some of the people in the room are old enough to remember about five ten maybe 10 years ago that everything was very excited about Big Data so as the uh internet kind of went overground and had Mass adoption generate an awful lot of data and people were very excited about this data uh but it turns out no one really knew what to do with it or how to handle that volume of data they think there's something in there but they weren't sure what it was um so Along Comes machine learning which allows you to effectively mine the statistical properties of large data sets and find out what's going on and maybe use them to your um your own purposes so I if you look at

[1:02:30](https://www.youtube.com/watch?v=isKDha1yt9I&t=3750s) the kind of graph of Interest AI follows on from Big Data by about 18 months where we are now is I think the consumer not I wouldn't say consumer adoption yet but certainly consumer awareness I think what being is doing today is going to make a big difference and Google are playing catch up on it so this this uh the big explosion recently has obviously the natural language um models that have become larger and become more convincing and in some ways actually useful I mean I must admit that I use chat GPT the other day and it's the first time I've ever used an AI and found it useful rather than found it something that I wanted to abuse oh that's interesting so it's kind of Hit the point where you actually had it did something useful for you rather than it was just a technical system you're reporting for kicks absolutely yeah it's uh uh as a result of eating so much of the internet um it knows or appears to know things and I think the uh the um the fact that there is a natural language interface to these machines is the is the biggest thing so I mean most of the smart assistants like Siri and Alexa and so on have been around for a few years we're trying to offer natural language and their mind is that people chose to moderate their language to try and get the results out it's just as people moderate their searches on Google to try and get what they want I've done two more recent scrub change has been this you can chat to it in normal language and get something back like like firstly if you're talking these you know Alexa are serious something you have to speak very cruelly and try and minimize your terribly thick Scottish accent absolutely yeah yeah and and then the damn thing doesn't understand what you want yeah you know two times out of three the response is I'm sorry how I have no idea what you're talking about and then you know interacting with even um you know even GPT three the older generation chat model it never has any problem interpreting what you're telling it you know I've never seen a misunderstanding request it may give a terrible answer but it always appears to understand the question correctly well I think for me one of the interesting aspects is this um there's a there's a kind of tension between the humans and machines and and we do an awful lot of work bringing ourselves to the machine to help the machine along and we do that almost subconsciously and that's as you say you kind of carefully use your words or try not to have a Scottish accent if you want to talk to Siri but the change now

[1:05:00](https://www.youtube.com/watch?v=isKDha1yt9I&t=3900s) is that it's they're much more responsive to just natural language flow so the illusion is that the machine is more intelligent behind the scenes because you can chat to it whereas I think it's an illusion particularly with large language models it's it's a reflection of the statistical nature of language it's not it doesn't know anything well this I mean this gets into the old thing what was it um I'm trying to remember searles's Chinese room you know this whole thing of if you had a statistical machine language translator you know you you follow a set of distractions in terms of one language into another but the person operating that process has no understanding of Chinese does it speak Chinese or not I sort of feel not to be too philosophical about this that we don't understand how we operate and classifying us as operating in a different way to a different system is a bit tricky because we have two systems we don't really understand and we want to see that they're not equivalent to each other maybe I would love to get into the philosophy but maybe here is not the place but uh um I think that if we're going to focus on the large language models like uh gpt3 or um I don't know what Google's Google's version is Lambda isn't it um what they're doing is capturing statistical regularities in the data set in fact they've read loads and loads and loads of stuff means they can spit out language that's very convincing that is very different uh psychologically speaking to being an evolved animal or a human who lives amongst a culture that has grown up in a you know you're an embodied system you're in a bag of meat that has a whole set of interfaces that have been evolved over millions of years versus a machine that can parent language in a convincing way so quite a lot I mean Gary Marcus I'm a big fan of Gary Marcus from back in the day and he's he's probably the best voice on at the moment because he keeps shouting from the rooftop saying look these things are not intelligent just because they look like they're doing intelligent stuff they do not have the nature they don't have the the right equipment to be properly intelligent like we are because they're not made of the same stuff yeah and I think what's going to happen with the AI um sort of Revolution that's going on now is that it's going to be a new kind of intelligence it already is a kind of new kind of intelligence it's not human and the the fact that we can chat to them in English or whatever language is kind of an illusion um it's a it's a thing that makes you very comfortable but it's not really belonging the underlying knowledge of

[1:07:30](https://www.youtube.com/watch?v=isKDha1yt9I&t=4050s) what the machines actually know I think um Stephen Wolfram did an interesting article about how you tie a machine like chat GPT to an actual knowledge Oracle that knows actual stuff about the world in a factual form so that it can correct these language models and and catch them out when they start lying to you I think that's where it's going to be it's going to be a kind of hybrid between yeah complex hybrids verifiable knowledge and you know the the frankly very impressive abilities of these models to talk to you long short story short book something like that for National novel writing month about 12 15 years ago uh and then the first conscious AI system um doesn't believe that human beings are conscious because they're basically just meat robots and how could they possibly have contrast us and it requires a very long conversation with a very smart person to convince it that humans are actually have Consciousness well I'm you know I'm not fully convinced of that one Consciousness is phenomenal in my opinion but you know let's say that so I think yeah on to him yeah well not Quantum woo more than uh as you said because that that question cannot be fully answered yeah okay and we spend so much of our lives in this illusory higher order language based activity that we think that's where everything's happening but you know if I took your cerebellum out you'd stop breathing right and that thing keeps going all the time we're not consciously aware of it but it's part of our biological makeup and I think um uh you know the things like the Turing tester attempts to try and find a way to check if something's conscious not by using the same modality and systems that we use I think this this thing that's going on with the large language models is very much like that it's like oh wow this thing's really convincing look if I ask it to write me a piece of code it will write me a piece of code and if you ask it to write a bio of Vinay Gupta it will make up a pack of Lies very convincing lies but absolute nonsense yeah and so there's this uh the danger of the things that it does well versus the things where it's just going to hallucinate um needs to be tidied up and I think that's what that's going to be the next shift yeah and this thing that this spontaneous confabulation is you know it it it sort of limits the use of these systems for recreational use only until it changes look it's fine if it makes up nonsense at a video game setting it's a real problem if it doesn't a medical center absolutely I mean I there was a

[1:10:00](https://www.youtube.com/watch?v=isKDha1yt9I&t=4200s) thing on Twitter wasn't there about someone trying to set up an AI Law Firm recently um which you know click a button and I'll write you a cease and desist letter or something yeah um there are many reasons why that's a terrible idea but I think people were getting a cart ahead of the horse with these language models at the moment they don't have what you would consider basic intelligence like they can't add up like you know it's similar with the uh image Generation stuff you know if I ask it to put a picture of three apples no chance I'm gonna get three apples it has no concept of three right as a concept of apples and perhaps plural of apples but there's no there's no system under the hood yeah and the thing with the hands for the concept of extra fingers and so on yeah looks like it just knows if you say the finger there's probably another finger yeah absolutely and I think because it's so persuasive because we live in a culture of other humans that we tend to trust and believe that when these artificial players come in it's simulation to trust and believe them as well um but that may be premature we've been we've been doing a bunch of thinking about you know how Mattereum makes use of AI um you know before Mattereum we did a project called hex heroic Capital where we spent about eight nine months trying to get a VC fund together and we spent a huge amount of time doing an analysis about future technology and the thesis was that all the interesting action was going to be for multiple different Technologies joined together so the four things that we thought were going to be big were AI VR blockchains and Robotics and the certain notion was that you know if you've got a company that has a robot that's steered by an AI both the physical art the physical elements of robot manufacturing and also the software that makes AI work are both going to improve pretty quickly and at that point where you've got a product which includes both of those Technologies you get to leverage both sets of improvements so we're kind of prepped for AI arriving um and we think that there are two prom predominant use cases for AI in the context of doing the physical asset stuff that we do um and I want to run those by you and get your kind of feedback on them so the first is the notion that if somebody wants an AI to go and buy a pair of red shoes for them um there has to be a pair of red shoes on sale and the AI has to be able to look at the data that describes the shoes and understand that there are in fact red shoes

[1:12:30](https://www.youtube.com/watch?v=isKDha1yt9I&t=4350s) um and that also gets into things like sizing of shoes you need your kind of specification um and you know you could imagine because I can imagine something that's almost like an AI eBay or an AI Amazon or you just have a bunch of very intensely structured data that's properly machine readable and so on and it's basically intended as a place for AI agents to be able to trust the data that they're making decisions on um does that sound kind of plausible credible sensible um absolutely I mean I think that the The Quest For Truth is very important right and I think um truth in AI systems is going to be a very Hot Topic at the moment as you say because there's a lot of contabulation so the idea that there's a there's somewhere I can go to find truth about the world it's very important now be that you know Encyclopedia Britannica or Wikipedia of something that you kind of people trust that gives you some facts about certain things but when it comes to stuff in the world you know the stuff that Mattereum cares about um having a machine readable format that can be looked at and assessed by another machine is really really valuable because I don't think anyone else is doing that because we're barely doing it yeah but you know all the back end stuff is accidental and Jason for a reason we try to sort of win for it so that I'm glad that sounds kind of credible the second thing I want to take is a little further out on the branch here so the inverse operation is I show an AI the cover of a comic book and then I ask it what is this comic book worth and at that point presumably somebody has a database and you train an AI and AI recognize this covers and it's got some you know access to data some thing like a valuation agent or a classifier do you think that's something that modern AI can support or do you think that's still a few years out to the point where you would trust that enough that if somebody wrote a warranty on it you know my analysis this is worth 100 bucks and if it's wrong I'll buy it up here for 50. I think if it's a comic book that's probably an almost trivial problem for image recognition it would be harder for a you know a crumpled suit sitting in the corner of your house you know the things that are flat and easily readable by Machine bits of text and stuff and image is quite simple but I think phys more malleable physical objects will be harder if I said I've got a ton of you know

[1:15:00](https://www.youtube.com/watch?v=isKDha1yt9I&t=4500s) aggregate outside my house do you want to buy it taking a picture of a pile of something and asking an AI to assess is that actually a ton or not it's much harder much harder but I think it depends on the domain we think there's a lot of room for this for example clothes size it you know like right now it's almost impossible to buy garments by measuring your body and then buying something online because the measurements are so impressionistic uh you know clothes are not centimeter sized eBay has its own kind of systems for doing it um but beyond sizing then we get into the question of taste and this this is you know that kind of final question so imagine a world where we've got a whole bunch of let's say clothing Furnitures things like this cat no longer together you know each individual item is cleanly described then we set an AI loosen the instruction to the AI is you know get me an outfit that makes me look 1950s who addressed me in 1950s gear for the night or make me look like a 1987 London Punk or get me an interior that is kind of art deco inspired how credible do you think it is something AI rummage through a pile of things with the ice cream descriptions and actually assemble something that looks like taste um I think that's entirely possible um assuming that the the items are suitably marked up and the inventory is last large enough um I think taste is a fascinating one and we probably don't have time to get into too deep in it but I think the the one of the problems of uh systems that are trained on large volumes of data is they kind of have a tendency to the norm so The Taste or the choices tend to be what is most significantly significant so it may be that my tastes are a little more on the edge of the bell curve and the system may not necessarily understand but conversely you know they tend to be pluralistic and you know they can support a lot of different things so I think it's it's technically possible assuming that the stuff is marked up I think whether I would trust the taste of an AI would kind of tap into whether I trust the AI as someone like me or not you know what I mean and I think that's I think the personality of AI is something you should watch out for for the next couple of years so the moment there's one Chachi P3 that everyone's talking to but there's going to be a million and

[1:17:30](https://www.youtube.com/watch?v=isKDha1yt9I&t=4650s) they're going to have different tastes and different attitudes there's going to be a you know truth social bot right that talks like Trump right and there's going to be another one that talks about Biden and people will be drawn to whichever flavor of AI that they want to engage with and so yeah maybe there's going to be a you know classic punk clothing AI that knows all about this stuff or at least a facet of an AI I think that that taste will become apparent but as long as the stuff's marked up and it's it's plentiful enough there's no reason why you couldn't do that I've been telling people for years and years that you know I I tweet so much um specifically so that one day when we finally have ai I'll have a huge training data set you know all the talks I've done that recorded online everything that I've written that's on the blog um and you know it would be possible to feed all of this stuff into some kind of system to create a kind of you know effective Robo Gupta I'd actually built one on Twitter a few years ago pre AI um Shard Echo where I took I think the first three years of my Twitter activity and put it through Markov chains and it started spinning out stuff that vaguely sounds like me unfortunately I think that's probably where we're gonna go I think most a lot of content is going to be if not completely written by Machine certainly augmented by machines or perhaps inspired by machines I think that's kind of fascinating for what it means for human culture yeah I mean what I want from that is I want something that will filter it for me you know here's a trading set of I don't know whatever is 100 000 tweets you know go hit the news and tell me what I'm interested in and I just wonder you know are we close to the point where that kind of augmented filtering is possible or is that still a few years yeah no I think that's that's eminently doable I think I think kind of uh triaging large volumes of textual data is something that's very much on the cards um I think making an algo vinay um you could probably do inside 140 characters but I'm not sure you'd do it in a webinar just yet but you know 18 months thank you now and I could emulate your voice and I could cross train a gpt3 to write words that sound very like I mean that all that technology exists whether it be convincing or I don't know well I mean it might be we should do a try who says we haven't already who says that this is the real vinay exactly Five Doors let's see you Eric uh thank you so much Eric and thank you so much to all of our speakers to Gordon

[1:20:00](https://www.youtube.com/watch?v=isKDha1yt9I&t=4800s) and to James and to Fede uh and to everybody who has asked questions uh we are going to move on to the after party now there is a link in the chat don't forget to collect your pope your proof of attendance uh a protocol token uh there is a link in the chat there as well uh thank you so much everyone a really brilliant conversation and uh I shall see you at the after party in a few moments

---
End of [567] The Fourth Mattereum Physical Asset NFT Meetup 8th Feb 2023, part 2 of 2. Its connections are at the top of this file; the map is https://guptaoption.com/llms.txt.
Read next: next on the same works, [566](https://guptaoption.com/texts/566.md) Why Global Trade Will Inevitably Move To The Blockchain (2023); the most closely related, [682](https://guptaoption.com/texts/682.md) Ep. #641 - Vinay Gupta on China, Futurism, Government,... (2026); his whole career in one file, https://guptaoption.com/story.md.