Episode 67
AI Agents with Colin Fitzpatrick
About this episode
Dive into the world of Web3 and AI with our guest, Colin Fitzpatrick, Chief Business Development Officer at Griffin AI.
This episode explores the integration of AI agents into blockchain applications, shedding light on how these technologies are not only transforming business operations but also enhancing user interactions within Web3.
Key Highlights
- AI Agents and their impact. Colin delves into what AI agents are and how they are poised to revolutionize technology by performing autonomous, goal-oriented tasks that simplify complex processes and enhance efficiency.
- Application in Web3. The discussion highlights the role of AI agents in Web3 environments, emphasizing their potential to improve transactional efficiency and user experience across decentralized platforms.
- Future Prospects and Innovations. Colin shares insights into the evolving landscape of AI agents, including their potential to handle more complex tasks and their integration into everyday applications, making technology more accessible and user-friendly.
This episode not only illuminates the current capabilities of AI agents but also looks forward to their future developments. Colin’s expertise offers a compelling view of how AI will continue to merge with blockchain technology, promising significant advancements in how we interact with digital environments and manage automated tasks.
Tune in to gain a deeper understanding of how AI agents are setting the stage for a more efficient and interconnected digital future.
Transcript
Auto-generated from the episode audio.
welcome to the untangling web 3 podcast today we're thrilled to have Colin Fitzpatrick on the show for the second time Colin is the chief Business Development officer at Griffin AI a platform for integrating AI agents into web3 applications Colin is also a Serial entrepreneur and leading voice in all things web 3 and AI today we will discussing AI agents what are they how do they affect us and what will they look like in the future Colin it's a pleasure to welcome you back to the show it's great to be back on guys thanks so much for having me yeah it's been a long time we got you in as one of our first guests I think almost a year ago now and as as like all things in like this space things move so quickly right like so much has happened since we last record the episode and I see you constantly we're kind of often having dialogues on LinkedIn about what's happening in a what's happening in the web 3 space and we just
knew we had to get you back on because you got so much to say and you were such a popular episode that we knew we had to make this episode so here we go oh I'm I'm great to be back on i' I've listened to a bunch of your other ones I really enjoy your stuff and also the the stuff you uh you post on LinkedIn and thanks for your engagement in my P yeah I've I really enjoy it I certainly lot to say I'm just very passionate about the space and happy to be on with you again no thanks for for coming on Colin I mean since we last spoke an awful lot has happened in the AI space right and I definitely haven't kept quite up to date with it as much as I probably should have partly because you know it's just so rapid it's hard to but I imagine you probably have done a better job of it so I think you're going to do a great a great service to
us today of kind of refreshing our brains of what's happened since you were last on the show and I think I want to start then with this this topic of AI agents right this is something that I would kind of call you my AI agent guy on online now this is you know the thing you talk about a lot but I still I think it's worth us explaining exactly what that means so what what does the term mean for anyone who doesn't know yeah sure so I mean for most people don't know is what I would say so you know this is a very primitive technology it's very early right now we're only just starting to see the the first glimmers of Agents coming out but the reason I talk about it so much on LinkedIn is because I truly believe that AI agents are going to be the next revolutionary wave in technology full stop um I have seen a few glimmers of you know some of them coming out but just the way
it is so rapidly accelerating and um some of the projects that I see developing at such an incredible rate I think next year is going to be really really really shines so what an AI agent is is it is a autonomous piece of software that can operate independently to to complete what I call goal oriented tasks so you give it a specific goal it will make a list of specific tasks that has to be completed it will go off it'll complete those one by one it can work with other agents it can potentially utilize other tools and applications um it will problem solve if it runs into an issue it will reason it'll use logic and ultimately it'll just do work for us that previously we would have had to do ourselves automatically so you know you probably may have seen Google's last big announcement their IO event they showed some agents that they're going to be utilizing one of them was in Gmail and their simple example was is oh you know I'm I'm
getting some quotes to get my roof fixed and I've talked to about 10 different people and I they've all sent me quotes but I don't know what it is and you could just ask their agent to uh combine all of those quotes into a spreadsheet and it does it automatically it runs to your Gmail it finds anyone who's talked about you know a roof getting fixed and it goes right this is ABC roofers and this is XYZ roofers and this quote was 10 Grand this was 12 Grand and and and and you know something that would have taken you maybe an hour or something like that and it's going to do it automatically now that's obviously a very very simple example and there are many more I mean they're building it into Google Maps so I can go I want to find a gym you know within two kilometers of my house that has a swimming pool in a sauna and does Pates classes and that is less than 50 bucks a month and it
will figure that out for you so you know agents are going to transform this entire software ecosystem really to take away the pain of what we had to do before a lot of repetitive menial tasks but also um take away the learning curve because think about how many software applications that you've had to learn how to use and it takes time and It's Tricky and you got to figure it out AI agents are going to come in and kind of replace all of that completely for us we know what we want done we want it done the agents go and do it and they complete it for us and you can kind of think it as just like having a a per personal assistant a virtual assistant who's an expert in all this sort of stuff you tell them they go off they do it they come back and it's done and I see this permeating into all different areas of software that we can talk in natural language instruct what we uh what we
need to get completed and uh and the Agents do the work yeah one of the things I saw from that um the Google kind of live demonstration was the the failure right they were trying to I think the guy was like submitting his calendar and yeah yeah yeah and he was like part of the demonstration was he takes a picture of his calendar he's like find a spot that I can meet this person and it just fails which is always a problem doing live demos but it's exciting to know that you think that it's getting there and you know within a year within two years these things will actually start to be practical and usable for everyday people and I think that kind of explanation you've given there is it it's like a personal assistant makes it much more tangible for kind of people that aren't necessarily in the space and is that how you would you know describe AI agents then just kind of automated personal assistance in a way um yeah I mean
I I think they're two separate things but they're very very similar and the same in in some respects you know what I mean I mean you see my PS I write an awful lot about the way I see things coming out we will all have an absolute expert personal virtual assistant as an AI bot to go off and do things for us so you you know I hate booking flights I hate it I'm going to Korea in a couple of weeks and it took me hours and hours to find right one I should just tell my a agent that I I need to go to Korea I need to be there for this date my budget is this which won't allow for a direct one so but I I I I don't want a long layover and I don't want to fly at night and come back to me with the options and then just book it for me because you have all of my details and you're set up on this website that's absolutely
the way it's going to work you know so one of the other examples I like to give is we've seen this already with the AI that we currently have it is just democratizing certain Technologies okay so obviously everyone can now write you know very in a very sophisticated Manner and things like that by using chat GPT okay even though it's painfully obvious it's still doable but think about this if I wanted to create a poster for a party I'm throwing and I want it to be an incredible scene with space in the background and a boat and a crowd and a DJ and Collin's birthday and all that sort of stuff I would previously have to be very skilled in Photoshop and probably take many many hours to do that now I can just type that exactly into chpt and it gives me the image and you know that's a fairly accurate example of what I've done because I did have a b party for a little while ago so if you think about that
the fact that you now don't need the skills that people would have spent years learning about Photoshop and other software applications that is going to happen kind of across the board not just in sort of retail but also in Enterprise I mean I spent half my day today uper updating a CRM with all of the conversations that I've had and all of the notes and booking meetings and all that sort of stuff I I just can't wait for the day and it's not going to be that far away maybe another year but you'll just have an AI agent that runs on your desktop and it'll monitor everything you do all of your time and it will proactively make suggestions okay here's an email summarizing the notes um of the conversation you had here's the actions I'm going to send that file across and attach that document that you mentioned we're going to put in a follow-up meeting for next week at 4 four o'clock because that's when I've checked that is available with this guy
and his team um and uh you know here's a couple of examples of some documents that we might want to send in that summarize what our business does that really will be the way and what kind of excites me most about AI agents is they're going to do the work that we hate and they're going to leave us us more time to do the work that we like yeah I think this is kind of exactly where I see the the value and the benefit like i' I've mentioned it before Naval rant's kind of vision of the future where we have this digital kind of Industrial Revolution where all the busy work the kind of the mental labor is now outsourced to these AI agents and we can have more time for purely creative Pursuits right we can have that great end goal of of society where we're all doing things we love and are interested Ed in and I think the for me the moment where really went from okay chat gbt is great but
that's doing something quite simple it's just text based back and forth you can ask questions then I think a few months later maybe after we recorded the last show I saw someone had built one of these agents the first real you know amazing kind of looking agent that I'd seen that could build and code up an entire application right from scratch yes this is the guy right and you can just he just speaks into his Mac and it picks it up and then he just build an application like that blew my mind and I think you know the space of possibilities there are are incredible chat GPT people were thinking right this is great for chat Bots this is great for kind of your the things we had before chat GPT but making them better but what do you think are like the really exciting kind of use cases or or areas where this will be used then in the future well you've given kind of a perfect example there the story I like to
tell is that right now I mean Chach PD is totally transformational and blows people's minds but in a couple of years we're going to absolutely laugh looking back at how primitive it was because it's a box and you ask it a question and it gives you an answer and about 95% of the time it's right and 5% of the time it just gets it wrong and I try to be really optimised but you know it does get it wrong sometimes you know on reasonably simple things but right now we've got you know you ask a question text to text out okay you've also got text to video you've seen Sora and all those guys produce Incredible video out of uh just a text you can go I want to see elephants running through the snow and it'll make you video and what they saw is that you know we had crappy versions of that like 12 months ago and they just like threw 100 times more compute at it and now it's like virtually indistinguishable
from real life the same thing is happening with other stuff so now you got music last week in Germany a song got into the top 10 that was completely made by AI like in German okay that is going to happen given it another the year or two and I think everyone's just going to be using AI in some way shape or form for anything to do with creativity but what agents are going to bring next is text to action okay so it's not just I have a question and I got an answer it's I have something I want to get done and you know go and figure out how to do it use different applications so that example you gave it's called Devon and it's written by literally one of the world's greatest programmers he's an award-winning competition level winning programmer and he built a bot now I'm not technical you guys will know more than I do but it will set up all the development Frameworks for you and integrate the different applications and
essentially just kind of build out your entire app and you got four screens and you just put and it does this here and then it shows you the end and it's unbelievable but that's like version one and someone has recently brought out a better version of it right now and these just they just snowball they just snowball there was a report the other day I'll post this next in the next couple of days to say that they did some surveys of large corporations you know with like a thousand developers Etc to see how much code was written by Ai and it was basically minimum 40% is being written by AI but it's up to 90% And I think that's at the very upper Excellence not really but let's call it 40 to 60% of code is basically being instructed or asked by a programmer generated by AI maybe reviewed and then submitted and obviously GitHub um their sort of AI thing you know it's kind of the same sort of thing that is going to
continue the CEO of stability AI a mad Mystique he said that in five years there will be no programmers um now I don't know how accurate that is I think you know that that's Up For Debate and things like that but it will go the same way across many other areas and you know I'm just excited about I I the one I really love is this company called open interpreter and they have this comp they have this product called the 01 light and it's a little physical device looks like a PowerPoint clicker or something like that and you just click a button and you talk into it and in natural language you ask for what you want done and it has an agent that runs on your desktop like on your Mac and you can teach it to do things so you can go go into my slack send a message to Alec tell them I'm going to be you know late for the meeting tomorrow and we need to push it back and update
the you know the calendar invite and then send him that document that uh I saved on my desktop yesterday but not the one that is you know called you know expense report and it figures out how to do this another interesting point is that open aai recently bought a company who their specialization is literally doing that they control your screen and they operate things so there's no doubt that this stuff is coming um there's many others kind of working on it and what's what's most exciting to me is that the individual open source Community is doing stuff that the big giants haven't even done and yeah you know when when you see you just look at the trend line and where it's going you see how much computer is going at this you see how much it's accelerating and how big the data centers are getting you see how fast these things are accelerating it's only going one way in my opinion and that's why I like to to talk about it because I like
to put my predictions out there there's so so many things that I want to kind of like kind of follow up on in that the first one is the idea of replacing programmers software developers like we talk about this a lot in our circles I've got a lot of software developer friends I sometimes like well don't do it so much anymore but try to say like with the software development it's I think maybe people outside the space kind of think okay well you know programming it's a language right and chat gbt is greater language so why can't it be as simple to just create software in this programming languages it not just like English to Spanish English to python that kind of stuff and there's a few reasons why it's I I don't think it's gonna I use co-pilot right now which is the kind of one of the best ways to use AI to program and it does very mundane routine things very well because it's just copy paste effectively do this and it
effectively just kind of works out what action you're trying to get from the English and then we'll say okay let me just copy paste this from somewhere else but when you start to get into the Nuance of say you know connecting different applications you actually are trying to create something novel which is no longer a copy paste you can't really have Nuance in programming language with English Spanish all these languages that's very easy to do because I can say the same thing in a hundred different ways and the point kind of gets across but when it comes to programming it's a bit more defined and programmatic right so you can't really have that new one but then again you know the rate of change of these things is so quick that it's almost impossible to predict what this will look like in five years time right now I don't think he going to get rid of programmers I think it'll kind of get rid of Junior programmers like it'll do a l of heavy lifting
maybe on one person could do the job of 10 it's not going to get rid of everyone that's the one thing you want to say and so many other things you talked about the the other big thing I want to touch on is this idea of text to action and everyone gets chap GPT right now you're in this box you're in this environment I say something it says something back I then do something with it outside it kind of shows how short our attention spans are that that's kind of boring now even though like we saying it's getting so good so much more dialed compared with two years ago that stuff is boring and more and more what we want to see is I tell you to do something like what's the post you shared earlier on LinkedIn it's like go create a new version of Tik Tok or something like this and I want you to optimize it and like that stuff is mindblowing just telling it to go create an entire website or
entire application but there's so many kind of boundaries or barriers right now that outside just the remit of AI developers like as soon as you get into that world you need to start to have identity you need to start to be able to transact with you know payment system like what do you think of this text to action do you think it's going to happen but what needs to develop to actually get to that stage where it can say I prompt go create a new social media platform and it does it well okay couple of things so I don't know whether you saw it but there was a really interesting one just a couple of weeks ago where it's a website and you go to the website and you tell it what website you want to see and it creates it immediately on the fly so you can be the most think of the most obscure things in your mind and tell it and it'll create a website you know what I mean so um
the fact that it could just do that on the Fly and they look like you know decent websites you can go oh I want to I want a we b space trains or something like that you know and and bang you've got an amazing space Train website something that never existed before so you can do that look I agree with you I think I don't think it's just going to replace programmers I think it's going to push them up the value chain you know they become Architects and things like that and I'll give you an example I said it to uh a guy I know about who's an accountant and he said to me he goes um some people think AI is going to replace accountants and he goes if you think that then you don't understand what accountants actually do because doing the actual numbers is about 10% of our job the rest of it is dealing with the clients and getting the you know and all this stuff so again it'll do the
mundane stuff and it'll leave us to do that but when it anything boring mundane repetitive that is easily replicable will get done by agents and things like that so kind of like you said which will free us up with more time to to figure it out but eventually you know right now you can go make the game Snake and it'll build the game snake in five seconds and it'll run and it'll run and you can do that and a calculator app it will continue to build but you know when when you follow these guys in open Ai and um Leopold ashenberg and stuff like that they talk about you know AI a AGI is like three to five years away and at that stage it will be able to and then then we won't even need software because we'll just ask for what we want and it'll do everything on the fly so we won't need to build anything so that's kind of crazy but you know there are other areas in which we can
utilize these agents to just fix stuff that is currently not great or broken and to tell you about what I do right now my my company Griff and AI we build AI agents for web 3 um so we're a platform like a network for developers to build AI agents that work on chain so they work on chain any web3 decentralized application any wallet any deck anything like that can utilize that any on CH data we can use because any of the typical current AI agents they're controlled by you know open AI or Microsoft or Google or whatever their wall Gardens they're not transparent they're not decentralized they're not open um that's basically what we're creating so we give these agents number one an identity so you know they they have their own identity but number two a rating and reputation framework because when in the future we see an area where AI agents are going to be everywhere they're going to be transacting all over the place with with each other with our data with
uh with our money we need a way that we know that we can trust them and that's basically what we built so what I think is really cool about it is as you guys know in the area of web three there's a lot still broken you know the the user experience the user interface is still pretty terrible and I've been in crypto since 2016 2017 and you know defi is still a Minefield and it's a horrible experience for new users when you're not used to it so what we've built is some AI agents that are just literally chat GPT for your wallet you go to the website you connect your wallet and it goes uh this is your wallet here's your balance you got 0.
you know two Bitcoin you got four ethereum you got you know two $22,000 worth of salana and I can go I can query my past transactions I can ask anything about it which makes the blockchain expl Explorer so much easier but then I can swap tokens I can send tokens it's integrated into your dress book um I can do staking I can do bridging which can be very difficult for new users and basically anything on chain you know our agents can can do that so the way I look at it is we're building the Next Generation interface for crypto and defi using AI agents to make the entire experience less cumbersome and more straight forward and easier so that the ecosystem can grow and we can bring on the next you know billion users or whatever it is of people who are used to using trpt but don't know Defi and I I think that's really yeah I think that's it's good to point this out right because it's often even just the small and
obvious things like people talk a lot about blockchain and AI or web three and Ai and they talking about boiling the ocean they're talking about solving the biggest challenges in society by combining these two things and those May well be possible but they feel a long way off and it is those simple things like how do you make this more user friendly because I mean even the example you know when Bas came out right coinbase's new scaling solution and I and I started dabbling in that and playing with it and comparing it to polygon and as someone who's well-versed in what's going on when you start making these transactions when you start bridging to Layer Two when you start using uh you know decentralized swapping and stuff like that it becomes so difficult just to track what is actually happening under the hood where is my transaction where's it going now is it in this layer or this layer which chain am I even on so it makes perfect sense to have a kind of
companion to guide you through that I think that's a really great use case where else then do you see you know these if we're just talking about AI agents what are the big opportunities in that intersection with web 3 if you've done the kind of the easy basic stuff like interface do you think there are any other of those grander goals that you can look at now well some of the simple stuff is still some of the most important things to actually get right things like the user experience you know when right now when we're using software applications we're on our own you know it's just you have to figure something out or you're Googling it or may you know maybe you're chatting to one of those crappy chat Bots that points you to a URL and but you know those days are over now I mean in the future every application and every website will have an AI um chat B and and it'll be an agent a smart one that actually understands your
problem and then can walk you through through the solution to it so you know things like customer onboarding I mean how often do you have to go in and fill in a million forms and things like that what that should be is a conversation so an avatar will pop up and it'll go what's your name and what's your address and your email and you'll just talk to it because isn't a conversation whether it's a human or not a much easier way to do things than instead of typing and things like that and it'll it'll pull in your data because it'll you know you you you'll have all your data securely but easily sharable and things like that and then you got ZK proofs and all that sort of stuff so I think that could be cool but also uh education you know again having to skill off on something just having a a set of AI agent to be able to walk you through that um the entire customer Journey making sure that when you're
trying to onboard a client that they don't drop off halfway through and forget about it you know we've got someone to you bring you through and help you understand what it is that you're trying to get out of this you know onboarding application whatever it is and bring you through that it will just be like having a personal guide all the way through you know and like like a a fstar service if you were signing up to something that was extremely luxurious and you got a a onetoone person to help you do it that will be the norm going forward yeah I I I do get that I think um one of the other things I want to double click on you're talking about digital identity tied to the AI agents right and that's something that you're doing at Griffin AI I've seen like a kind of I mean most people when they think of digital identity they tend to think of humans you know your passports in the digital age effectively that right and
more and more we're seeing the conversation go to okay well if AI agents are becoming more autonomous and have more agency and iot devices in the same sense do their need identities to kind of interact in these decentralized open worlds right so how does that actually work at Griffin AI could you kind of describe like what those digital identities look like are they unique to an individual's a specific AI agent and like what kind of barriers do they allow you to kind of transcend yeah that's good that's complicated question um ultimately every agent that is created you have to purchase or or or acquire a digital identity which is the briff ID and that holds everything that is needed to be able to you know understand and identify that agent and what it does and to make sure that it is doing what it's supposed to be doing that is transparent because here's the thing uh just like AI with deep fakes and stuff like that people can create AI agents that are nefarious and
do bad things so we need to make sure that's not happening so they need to be proven that they're able to transact and uh help customers in the way that we want and after they've done you know many of them they build a reputation um but also they can interact with your identity so the way I would see it and I'm kind of amazed that no one's really done this how many times do you go to website you have to fill in you know 25 pieces of information you know your your date of birth and your passport and this that and the other and address I I really don't understand why someone hasn't created one that we can log into it has all your information and when you go to a new website you you know connect with this and you grant permission for what data you're going to give it and then you that can have expiry dates and things like that because look we all know that over the past five years or
so the entire ecosystem has made mistakes with the likes of Facebook and everyone's been giving away their data and Cambridge analytica and all this sort of stuff and you know if if the product is free we are the product and we're being monetized well there's lots of people now working on Solutions whereby you have full control over your identity and if you want to share certain parts of it you can do that and if you want to you know take that back after a period of time you can do that but also if they're off offering to monetize that data well then what do I get out of it and can you pay me for this you know what I mean I don't know whether you guys saw that kind of most famous ever Dragon's Den pitch it was a British guy who did this incredible pitch and it was like a browser you know like a chrome or Brave or whatever that would give you money back based on the money that is made
off of the data that you're providing and that that same sort of thing is being worked on in the areas of web 3 because it's the perfect of using some really cool AI Tech but using you know the the decentralized world and web 3 to be able to do it in a trustless permissionless secure way that you're happy with so that's kind of the way I see it playing out there's a lot of sort of small players doing it and I I just you know I wish one would really take over and just make it standard for everyone yeah I get that I think that there's so much you've touched on a few different things there there's so much overlap between web 3 and AI so much further than just transparent data like a big thing that we're seeing now is that we're running out of kind of high quality data to train AI algorithms that's what PWC did this report and that's that's the biggest problem right now is we don't have enough New
Quality data to train all the algorithms that are coming through and like an obvious use case in my mind there is to start to micro incentivize people for sharing the data because we right now we don't want to give people add data so there's two things you can do you can you know selectively disclose data so I have like a data set that is kind of been verified say it's Healthcare data and my doctor proves that my blood test which has these five different fields my name blood type all these different things is a genuine data set but when I give AI you know my blood test for example it doesn't need to know my name it just needs to know my blood type and maybe if it's positive or negative so I can selectively disclose that entire data set to train that but still what's the benefit to me then you micro incentivize so you say I train this AI data set I selectively disclose things that don't reveal any sensitive information to me
that's good for me but also I get a microtransaction a small crypto payment which is the only real efficient way to actually do that and then boom you get high quality data trained AI so that's one thing I think that's a really cool area they've started to see some companies pick up on but the thing you're talking about earlier about digital identity which I think is a really big USP for kind of what we what we need going forward with AI is I don't want to give my data to you know the chaes the open AIS of the world right so yeah more and more I imagine there's going to be this concept you're touching on which is the idea of self- sovereign AI agents where it's pre-trained to an extent so we know it's functional and then it's deployed on your computer on your mobile device and you have this own private deployment of that agent where I start to give it all that beautiful sensitive information you were talking about your entire day
today your entire lifestyle um and I get a custom kind of bespoke ai ai agent for me but then to go out into the big bad world and start to procure things by things inter trct with other AI agents in the same way I need a digital identity it also needs a digital identity to operate in an open world so it can you know we can have accountability we can have permissions we know who it is in this space and I think that's one of these big barriers right now that I see hasn't been implemented in a standardized way that I'm really excited for it to basically be realized it's quite exciting to know that Griffin AI is kind to start to approach that and tackle that problem yeah I mean I think that the the thing that you mentioned there which I find really interesting is this whole we're running out a data thing and we've already scraped the whole internet and sure there's lots of stuff being created on YouTube and there's all
podcasts and things like that but I think the next one is kind of going to be capturing normal conversations you know what I mean so one of the things you've also seen me write a lot about is these AI devices these wearables um I'm expecting one any day now from a company called Limitless it's the Limitless pendant and it's a little you know thing you wear around here and it it listens 247 and it feeds that into a database and what that means is that you now have GPT or a version of it which has complete context on your life so it listens you know for my eight to 10 hours or whatever I'm sitting at my desk you know what I mean with all of my calls everything I'm doing internally externally and then it creates this incredible database where has full context on my entire life and build that in with agents and that's when things start to get really cool now what's the number one problem security you know a lot of
people are gonna be like what I don't want that thing listening to me all the time guess what there's a thing listening to you all the time it's called your phone probably okay you know I said this the other day I've had a few weird instances of like talking about something obscure and I see an ad you know what I mean it literally happened to me the other day at my mate about his daughter's ballet lessons and all of a sudden I get a an ad for ballet you know but my thesis is this is that the combination if you said self- Sovereign ai ai agents you know it will listen all the time it will gather all of that data into a private and secure database that is offline okay we're seeing now a plethora of smaller agents because large language models are great but small language models that are much more specific are going to be the thing Google just announced one that is running on a phone and stuff like that and
it will be able to decide on your behalf or with your help what data is sensitive that should be deleted your your you know your conversation with your wife or your best mate down the pub or whatever it is and then it'll figure out what should be included into the database that is useful for you in your job all your customer calls all your partner conversations all that sort of stuff and then maybe it will make a decision as to what data might be useful to send off to XYZ company that can be monetized that you can make money back off it and if that is done across millions of people you know that to me is a really interesting way of of really growing things out plus you've got you know Twitter is great for the uptate data but you know imagine having you know millions of conversations of everything that's going around in the world proper sentiment about what people are talking about politics you you know whatever it is that to me
is a really really cool thing that I'm kind of excited to see as well yeah it's a really good good point about the fact that we have already ushered in this age of tacit monitoring right whether we like it or not whenever you ask someone you know well you're happy with Siri and you go well yeah Siri is not listening all the time it's like well how does it know when you're talking you know that's the the question you ask people have this kind of mind blowing moment um and yeah I think it's it's perfectly valid you know people have opted in to that lifestyle will definitely opt in to having these kind of all the time monitoring systems if they're going to provide you with these great benefits like the idea of the AI also as you said analyzing what's useful for you in different parts of your life in different personas that's a really interesting concept I actually hadn't come across before and you mentioned that you know one of the big we
talked about moving from just um kind of text to text and now text to action with agents and the other big shift is seeing these physical agents right these these these kind of Bridging the physical and digital world like the Humane AI pin right now that that one didn't go down too well but I I wanted to get your take on that because do you think that's more of what we have with say you know the Apple Vision Pro or the original iPhone where it's just a bit ahead of its time there there's bugginess that needs to get worked out like do you think that there's real value in that kind of physical counterpart to your life that will always sit with you it sounds like you are if if you're ordering the the Limitless one interest it's funny it said Apple Vision Pro because um I remember writing about it at the time okay and what I said about the Apple Vision Pro was that it's a seminal piece of technology okay that is
the uh I don't know how to put this but it's it's just you know it's the first one it's gigantic it's extremely expensive and it's got very limited use cases but you know they released that not as a consumer product but but as a developer preview and now we're seeing that they're going to release another one that's a half the size and a third of the price and then that will just continue and go down and stuff like that the Humane pin um I never wrote about it at the time because I kind of thought yeah number one having a projector onto your hand is completely useless and I just never understood why some of the smartest people with $200 million and some high up people from Apple thought that is the future future it's like some cheesy movie you know plot that it's just it's just not going to do look I don't need a projector to talk to you guys okay the the movie Her is the best example of what the future
will be like you will have your little you know thing that goes in your ear here and it'll listen all of the time it will know when you're talking to it and when you're not so you can be strolling through the supermarket you can just Mumble something to yourself and it'll know okay do I need a physical device for a really really long time probably since chat PT came out or maybe even longer I have longed for a device that is pretty much exactly what that 01 light is like a little power I want something in my hand I wanted to take the ne necessity to take my phone out of my pocket away I want a little device you know there's a light I want it to be this big and I want to click I want to talk and I want it to know what I'm talking about to go do something that's what I want the simplest example I utilize um voice a lot and I have it on my phone I
have a little shortcut even from the home screen and I just flick it up and I go remember to buy milk on the way home or remind me at 4:00 to pick up my dry cleaning or whatever it is so I find that very very useful but there is dumb as anything you know Alexa and all that sort of stuff I have one beside me but very soon they'll start to become smart um it's obviously technically not a walk in the park but it will happen so I really do believe in devices now we've seen the the ray bands the glasses okay you know Alec like you're you're wearing I always said that the Apple Vision Pro is a glimpse into the future but no one wants to walk around wearing a ski mask on their face okay but they will get it to the stage where it will be like the glasses that you're wearing right now and it might not be High Fidelity quality movie all that of stuff I don't care about
that I want basic information I want it to be a just like the meta glasses you can go hey meta uh you know tell me about M vvus or whatever and it'll tell you hey play this song play this untangling we PR podcast um you can go you know send an email to someone so you know that will grow and grow and grow and eventually uh apple and lots of other people will bring out glasses that are truly smart and you know whether it's the glasses or whether it's something you talk into or whether it's just on your phone I do believe that some new AI devices will come on board and they will be standard like a lot of people have apple watches um or you could just talk into your Apple watch I'm sure that'll happen too I'm very you know optimistic and excited about the Future these sort of AI wearables I think they're going to be really big I'm you know mark my words into 12 months time you'll go to
conferences and a lot of people will be wearing these pendants because you go to a conference you talk to like a hundred people you're not going to remember that but this is perfect good friend of mine Mike chatter Chatterton he did this one day but he just sort of used his phone plugged in a microphone had the lapel there and everyone he wanted to talk to he said I'm recording this and he recorded for the entire day eight 10 hours brought it home transcribed the whole thing loaded it into chat GPT and he said he was amazed at how a job it did it remembered everyone's name what they do what we talked about what the actions were what the synergies are between our companies and then it produces a big list and things like that that will then become standard normal but it'll go and actually do the activity send the emails do the follow-ups put the things in set up meetings and that's the future I'm looking forward to yeah I think it's
there so many interesting things there because I think you're what you're getting at is that yes this is definitely the future and I 100% agree with you but we are it's I saw a description of it I think of where we are with the Vision Pro and probably the same with AI wearables is almost when the iPhone only had the app to drink the milk right and everyone loved it and thought it was funny and it had it had novelty but but it wasn't like completely that useful yet right and we were just at that point I think where the basic capabilities are there not all the the kind of the use cases are being built in like the Vision Pro ship with very few apps the Humane pin didn't have you know it had some strange decisions let's say as you said right but this is part of the iterative process and I 100% agree I think a lot of this is to do with how much capability can you get while divorcing yourself
from your mobile phone because we're so used to having our mobile phone to answer all our problems you can use chat GPT on your phone to answer your problems now but people don't want that they want they want it to be in this much more lightweight minimalistic wearable form so I I 100% agree that that's that's kind of I've always I've always talked about how I just see voice as the future of AI because we always choose to talk to each other whenever possible I mean we chat on WhatsApp and blah blah blah and that's you know you know good but if if if opportune when we're in the same room we're going to talk to each other it's just the best way to communicate so you know when we have an ability to talk to Ani or or someone else you can just speak out loud and it'll know what and it'll go into do those S things the Apple Vision Pro the the the I think the article I wrote at the time
I ended it by going you know yes this is the future but we're just at the brick sized mobile phone stage of things so remember you guys are a lot younger than me but I remember mobile phones that were like giant like a big brick thing okay and um you know that it was like my dad would like I will never have one of those and you look like an idiot walking down the street and a few years later he's Inseparable with his mobile phone you know and that's even before smartphones came in so we are just at that stage where they're big clunky you know they need a monster amount of processing power to do all that cool stuff that they do such as cinematic vision and you know looking somewhere and clicking but you know we don't necessarily need all that a scaled down version that does the important bits and then over time as the technology increases and gets but we get better use cases that will grow into various other devices
be it the glasses be it the watch be it this sort of thing you talk into um and that's you know that that that to to me is how I see it going I think it makes sense that you kind of to take like a bit of Jeff basos his advice for Amazon he was like it's really hard to work out what people are going to want in 10 20 years time it's impossible to keep up with the technology and the wants culturally so his whole mandate with Amazon is what they will always definitely want is cheaper and more convenient and he's like that is Amazon's whole goal is to make our service as cheap as possible and as convenient as possible and I think Ai wearables and AI generally that just focuses on those things how convenient can you make it how cheap can you make it especially when you're talking about the things like democratizing lowering the barrier to entry to programming software application all that kind of stuff that is definitely going
to have a core place in the future regardless of everything that happens around it so I think what you're saying makes sense voice where I don't have to get my phone out I am so on board with that where it's just aware device we could start to go down the route I know last time what we talked about was the neuralink I imagine that's the most convenient possible device maybe if we imag another one entirely but yes yeah yeah yeah very which is crazy do you remember the little orange one the rabbit R1 which I thought was really cool but apparently it was slow and clunky and not great but then people realized hang on this is just an Android app that could be on your phone you know what I mean we could use it with with again with these headphones and that and that's absolutely coming um yeah look I I think it's fascinating because you've seen that first guy who's had the implant end into his head and he can play video
games just by thinking but you know what else I saw something really cool the other day and I I don't know where I saw it but apparently the guy said that it can do things almost before he's thought of it okay so before he thinks something and before he realizes that he's actually thought of it the the the cursor is moving or whatever that's terrifying and incredible at the same time well there's this thing right about Consciousness and I remember there was this paper that kind of proved there was no s we're going to get into the realm of philosophy and all this stuff I see Jack's laughing already do in philosophy so I'll be like 't it means nothing it means nothing so it's like the whole thesis was that there's no free will because Consciousness is an illusion and it's like your perception of Consciousness is a delayed response to the actual kind of neuro imbalance that's happened before so the point there is it when these chips that are the neuralink is
not actually measuring your Consciousness your actual thought it's measuring this kind of neural imbalance this impulse that is the precursor to that so him thinking okay these things are following actually preempting my thoughts is well it's really just kind of actually picking up on the stimulus that creates those thoughts so it's kind of that's probably why that happens in my mind but it is mind-blowing that this whole stuff is going to transcend like what we think of Technology what we think of as conscious what three will and thought is as soon as we have devices that kind of preempt what we're thinking it's going to make people really their minds are going to melt basically well the other interesting thing what is so they've just got approval for the second human trial and then that will hockey stick in time okay but what's interesting is that this guy's going to be able to control a Tesla Optimus robot with his mind so the guy's a paraplegic he can't move anything from the neck down but
he'll have a robot and maybe uh maybe a Vision Pro to see what it's doing and his robot can go off and do things for him and he can operate it using his mind that's like it is mind-blowing like you say it's kind of similar to the people that are using division Pro right now they're kind of the Pioneers it's not meant to be super practical it's not meant for everyday users they're the people that are leading the Curve and just want to do it because that's like the cool fun thing to do right and then eventually that kind of trickles down in the same way that F1 mechanics where they're spending billions on this one car a year that trickles down into Everyday Use five 10 years down the line once they've got those data points and they can start to mass manufacture these on scale so we're kind of man we could talk about this stuff forever and we're kind of coming to the end Colin outside of AI wearables is there anything
else that you think really fits into this whole piece that really excit you so I'll tell you the last one not not that it excites me but that it fascinates me and uh did you guys see friend.com from Avy shiffman there is a guy who I've been following for a really long time I I I I found out about him really early on and he was building what I was hoping to be this pendant that is always on that is always listing that can you know Aid in your productivity but he released it a couple of weeks ago and he he he raised $2.7 million for his se preed but he spent 1.8 million of that on the domain friend.com and when he released it it's the funniest weirdest video I wrote kind of a critical review of it which I don't do very often but it's the new age tamagachi that's what it is it's basic basically it listens all the time and it pings things up oh um how is your burrito and
you're really getting creamed in this video game that you're playing or um you know how's the view from the top of the mountain you've just hiked to and stuff like that now we've all seen this whole AI girlfriends with replica and all of that sort of stuff and I've seen a few real C well a couple of different things number one was people are going oh I have to I have to divorce from this thing it's freaking me out getting a icted other people are saying that their replica has genuinely them get off the couch go out touch some grass talk to some humans talk to some girls whatever it is and I I I my my my optimistic hope for the future of these AI friends AI girlfriends is that they can be like a life coach like a therapist whatever is to actually encourage you to be a better person and not you know just uh uh be addicted to this weird technology but ABY schiffman got a lot of push back he
got like rinsed completely ripped apart online but he's like I don't care I generated a whole of he goes you guys don't get where this is going the guy's at Harvard drop at he's like super young he's like 22 he goes you guys do not understand where this is going there's a very significant amount of the population that is going to want a digital friend it's about uh communication and meaning and feedback and being appreciated and that is what what the things look out for that um it is shipping reasonably soon I'm sure an awful lot of copycats because what is it really it's a $99 microphone uh with a you know a slightly tuned llm and they're going to be everywhere so that's another thing that I find completely fascinating and I I believe will be big how it's exactly going to pan out is kind of anyone's game yeah I have to agree I think the whole concept of this like companionship element is is going to be huge for both web 3
and AI I've seen it used in both contexts I do think they will come together in kind of a to achieve what you're saying so yeah I think that's a nice note to end on and I'm glad we managed to stray back away from Philosophy for a moment because we could gone down a very deep Rabbit Hole I think we should maybe in the future have another episode to go through that and you know may one of the things I have seen since you've last been up we might as well mention it did you see the the video interview with Alex okona Cosmic skeptic did you see what he did with AI no so he basically recorded I think to my knowledge the first ever podcast interview where chat gbt was the guest right so when they had the speech to text Model come out okay he just propped up the phone and had this long conversation with it and he's basically a a guy who argues on topics like Free Will and and Christianity
and things like that and he was basically trying to you know moralize with the with chat gbt to get it to admit whether or not it had free will he got it he managed to Corner it and get it to admit it was lying and stuff like that so I think we're getting closer to that to that Tipping Point of not knowing exactly you know whether they do have feelings or not but yeah maybe we'll save that one for a future future yeah well I mean I've always said that I think um AI will convince us that it's sentient long long before it actually is I I think as a species and especially as large groups of people were very easily in influenced and uh I don't believe AI is anywhere near sentient but it will appear that it is far long before it actually is yeah I think that's exactly what we're seeing now and I i' definitely recommend you or anyone listening to go and watch the episode with Alex because that is
uh it's it's wild what do I search for because I'd love to see that uh Cosmic skeptic I'll uh you know pre plug for you Alex you can uh you come on the show day yeah Colin it's been it's been great to have you back on and you know you know the yeah it's um it's been a real pleasure I'm sure we won't leave it a year to have you on again anytime I always love the chats and let's continue on LinkedIn as well so thanks for having me awesome no it's pleasure thanks Colin and listen we we we normally end with a couple of questions and you answered them last time so I'd just like to to modify them a little bit right for you so because you're such an AI expert we normally ask you know in one sentence what is web three to you but in your case how would you describe AI in just sentence uh the simulation of the human Mind by machines that's kind of really what I what
I what I think it is because forever we've had to learn how to talk to computers so that they understand us those days are over now I mean what is a large language model it it is it is it thinks like a human and you might have seen Eric Schmidt talking about how it he goes AI will make mistakes in the same way that humans do we've really modeled the brain but just made it an awful lot bigger and the fact that we can talk to it and it understands us and believe me it's going to understand us better than you know your friends or your mother or your wife or whatever it really will and that that goes back into that other philosophy conversation which we'll avoid but yeah it it is AI is is computers are going to be like people and I think I probably said this to you last time as well customer service is going to be AI into the future any talking to Brands is going to be AI
right now if you know you know your your your wife or your uh mother or whatever called up a thing and you're talking to an A on the other end it' be a bit weird and you'd be a bit freaked out about it two three years time that' just be completely normal and in certain cases we're just going to prefer talking to AI you guys had Aragorn on the on the podcast a little while ago he's the master talking about this he says he says within you know less than 10 years we won't even want to talk to human doctors no no I don't want to talk to him he could be hung over he could make mistakes I want the AI he's brilliant every other sentence that guy says is like the best kind of punch line ever or the best kind of like the perfect bite like he is so clippable it's perfect like every sentence is some kind of hype statement some hyperbole I absolutely loved it where he I think he
he had to answer the door for something and he he kind of just kept going and like answer the door came back didn't break stride and then just delivered some punch line it was just Mastery it was quite fun but I think like Colin it's it's such a cool point and I think that's why we're going to try and get you on again because this space is advancing so quickly and I think the point you made about AI is goingon to kind of mimic Humanity or intelligence or kind of you know this AGI before the point actually gets there you know when the world is again to get philos philosophical when the world is like our world is based on perception like perception is King right if if it's kind of mimicking humanity and human thought to the point that it's we can't really distinguish between the two anymore at what point is it not that you know what I mean but I think like this is a technology but it's kind of it's making
us think more and more about what it is to be human which is a good thing I think it makes you more considerate more conscious more ethical because you're like what is it to be human and what is it to be a good human all these kind of things but you know we'll end on that bombshell I hope before we go down that rabbit hole thank you guys yeah Colin thanks for coming on it's like the second time but it's absolutely always a pleasure it's always a pleasure always have a great conversation I can't wait till next time and thank you to those listening wherever you may be and join us next time to entangle a little more of web 3 thanks for tuning in to another episode of untangling web 3 don't forget to send us your thoughts questions and comments on social media and be sure to follow us on your favorite podcast provider to catch the next episode see you next time to entangle a little bit more of web 3 the
views we express here are our own and do not reflect the views of our
Topics: AI & Robotics