The Death of the Human Middleman. AI Agents, RWAs & the Automation of Everything

Transcript

Yeah, >> good to go. Okay, >> I'll do it like this. Like this we can actually see the screen that we know what's on the screen. >> Okay, here. >> Testing. One, two, three. >> Yeah. >> Yeah. Perfect. Okay, so upcoming topic is really interesting because are AI agents really removing us? The death of human middlemen? Um, we have two speakers. Uh, Nei, I hope that I'm pronouncing correctly and Frank.

>> Thank you. Thank you for the introduction. Uh, yeah, you go first. >> I'll go first. So, what we're talking about today, obviously AI is trending right now, but this is something that we're going to be talking not a typical talk. So I'm a marketeteer. Frank is a developer. So we have our own different mindsets and own different uh insights on AI. Um a little bit about me. I'm Irish NFT gal on Twitter. Um I work for Super Team Ireland which is a Salana based community. We run hackathons weekly, monthly. Um and we have events uh every single week. And I also my second client is Kudo Compute. It's an AI factory based globally. It's basically um we we uh educate people on data centers um so really large scale

tech companies um and why they should actually have their own data centers instead of spending a lot of money on uh compute elsewhere using other people's AI. Um a little bit about me um I recently join or not recently I've been six plus years in blockchain um working for marketing for mainly protocols. I've worked for uh marketplaces and I joined originally as an artist um as a full-time artist back in 2021. Um so this is how I know Frank.

>> Okay. And then yeah, a little bit about me. So um I'm Paul said Frank. Yeah, I work for Cross which is the largest incubator integrator in Belgium. We're like 10,000 people, 600 companies something like that. And under that I run the Hyperlapse brand which is our web three brand. And indeed I did get to know Neve in 2021 on a Discord where we both were grinding uh to get an invite to to be a foundation. >> Such nerds. We met each other on Discord.

>> Yes. Uh that was fun. Uh foundation doesn't exist anymore. We're still here. >> Um yeah I I started a a little bit older than Ne. So I started a little bit earlier. So I was mining Bitcoin in 2010. Uh left the scene, came back with Ethereum, left the scene, and then finally came back in 21. Still here. Um and I'm mainly work for financial institutions. So no tomatoes, please. Uh but I'm an advisor uh to the large banks in Belgium on how they need to handle crypto. Okay. So Frank is going to talk about the kind of traditional ways and financial ways as well of AI.

Um and I'm going to talk about like the practical solutions like why people are using AI. Um so this screen this is our punks. I'm on the left hand side and I'm talking about how it is replacing us without us even being fully aware of it. Um I think most people nowadays don't understand AI. Frank, you're a developer. Can you say comfortably 100% you understand AI and like the tech behind it and what we're actually doing? >> No, I'm always amazed that it actually just works. >> Exactly. You don't care. And that's the problem.

>> Yeah. Well, we don't care. That's true. We just use it. And that's why why basically Yeah. Ne is saying they're replacing us. She's the skeptic here. I'm saying no, they're not uh replacing us. They're enhancing us. >> I am a curious just go back to the the other slide. I am a very curious person. I like to ask questions. I don't like to just accept things. And I feel like today we are just accepting that. I'm not saying AI is a bad thing, but I'm saying we're not really understanding as to what we're actually doing with it in terms of like our identity, uh what we like things nowadays, how we actually value what we're creating. We're all content creators. Whether you're a

developer, whether you're a marketeteer, um it doesn't matter whatever you're doing in crypto today or block blockchain in general, h I think we just need to like step back a little bit and actually fully understand what we're doing uh doing with it. Um yeah, >> maybe do a quick crowd check. So, who's the developer here? >> About half the roomish. Yeah. >> Okay. Who's a marketeteer maybe? >> Okay. >> Yes, we got one. >> Okay. Good. Uh, yeah. Next slide. That's your slide.

>> So, I like there's been so much news lately in the bad media about AI, like Antropic recently. It was only in the past week. I didn't even get time to update the slides because AI moves so fast. Um, but this was a tweet that I made on the Super Team Ireland's account about Facebook. I actually used to work for Facebook, so this is why I tweeted it that they cut 350 jobs in Ireland recently. I know so many friends from Facebook Assenture where I used to work and they're literally like terrified of AI because they've already lost their jobs. It's happened to me already. Um but I do believe right now because we're able to create and use like vibe coding more and more like jobs are being created but people don't know how to

sell we can make a business in like 24 hours. So I think even marketing is going to become more important. It's not going to become less important today. I have had so many like job offers recently since AI has come up. But honestly, I was terrified of it. Even though I do like to specialize in AI marketing, I just have to like kind of educate people a little bit more because we're not really fully understanding it. Even you as a developer, you can say that it does still terrify you. Frank like doesn't care because he's going to retire soon. for for young developers nowadays like in your company you are you wouldn't like to be in in their shoes >> it is it is a problem it's I in our company we mainly we hire people

straight from school you know college uh that's what we do that's how we did grow to 10,000 people and what we see now is yeah we give all the seniors we we give them a whole set of agents that they work with and they don't need these juniors anymore. They just talk to their agents. But how does somebody who come straight from school then will actually get to become a senior because I in my in my company we say there are basically eight levels of maturity of using AI and I assume here I'm going do a check hand everybody here in the room is level one.

Level one is I ask a question to chat GPT. Everybody did that? You guys don't Oh my god. Yeah. And I'm not going to explain all the levels. Uh but level eight is you just give it an idea and the AI runs away with the idea. It spawns up a whole set of other agents. It builds it for you. The next morning you wake up and the product is there and it is completely deployed. >> You're level eight. >> Now personally I don't personally I don't believe that we can actually achieve level eight at this moment. Uh, I think I think we're at I think we're at level seven for those that are really advanced.

>> Are you still level eight? You still believe he he believes he's level eight or >> Yeah. >> Okay. >> You come up on stage now. >> Yeah. But the level seven is is it keeps the human in the loop. It It is very similar to the level eight, but it's not completely autonomous. So, you're still controlling it. You're still checking it. And personally, I don't think the AIs that we have right now are already smart enough to go to level eight. They still make too many mistakes. And so there's still this human in the loop effect that we need there. And yeah, maybe indeed in three, four years we don't need that anymore and we can go to that level eight, but right now we still need these seniors that actually control

that AI that tell them what to do. And I think with AI, we're losing that kind of value behind it. And because again, anyone can nearly become a developer. And Frank will probably disagree and say, well, they're not a good developer. We're we're not going to have great developers. And that's the problem. We're going to have basic developers that are studying right now. If they're already using AI from even primary school, I have lecturers here um from pri primary secondary school in Ireland and they are teaching everyone AI. I'm not saying it's a bad thing, but I'm saying maybe we shouldn't start it so early for like young children. Um, because we're not going to have that like high level of developers in the future. Like Frank can educate and pass

on your knowledge, but why would they when they have AI? >> Yeah, that that's that's basically the problem. So, how do we get these young people now to become actually a senior developer who can work? Because what we say in our company is the the way A developer in the old days could be a super introvert guy sitting behind his computer and just toggling away and building this application. This will change with AI because now he will become the interface of the AI. He will talk with the client or whoever had the brilliant idea and he will tell the AI how to build it. So the the way of working for developers will actually change. So they

will have to come out of this uh room where they're at their basement where they're working and and actually go talk to people and what do you want how you want to build and they they still have this knowledge on how to build an enterprisegrade application. So and that that is this problem with with the younger people. How are they going to get that knowledge on how to make an enterprise grade? And indeed maybe in three four years we don't even need that anymore. The AI will know it too. right now I'm really convinced that he doesn't >> but it is very true humans are the bottleneck um so like I I do believe that yeah we have like I spoke earlier about like we're losing a lot of jobs

because of AI yes it is creating more jobs like a lot of our um in super team Ireland we have a lot of members that actually are full-time their their job as an auditor they actually like audit the AI code right because AI makes mistakes why because humans make mistakes and it's based off our knowledge. That's the problem, right? So, we can't rely 100% on AI right now. Yeah, in five years, >> we probably can, but right now we can't. So, we have to be mindful and still use our own judgment. And I think a lot of people are not doing that. Just simple like even like little reviews. We're we're letting AI do that for us and we shouldn't. We should be still using our own brains.

>> I fully agree with that. >> Okay, we agree on something. That's good. Um, so this is something actually that Frank took a photo of. He's a bit of an artist. He's a developer, but like SL artist. He's is a unique one. This is why we get along. Life of an of an AI artist prompt tweet sleep repeat. Something I probably like say when I go to sleep and I like I I dream this. So AI doesn't just replace work. It replaces our identity. we do lose a little bit about like our value uh like what we actually like our personal identity what we create I still like to create with my hands as I do like to talk with my hands um it's it's a part of you it's a part of your personality

AI cannot replace your personality and as a marketeteer I do you use AI every single day I love it but it is absolutely terrible for humor it can't replace my personality h in Ireland like it's it's a huge huge part of our culture is to like not take things seriously and to really like be truthful to like who you are, right? AI can't do that right now and I don't believe it will ever, but I could be mistaken. Um, so yeah, this is the identity part AI can't replace. Therefore, I feel like I will always have a job as an AI marketing expert, let's say, in the future.

We'll check about that in four or five years because my thesis more is clients they don't care and they just they have a problem they want to have that problem solved. Um we were talking about that earlier on today. I had a client and they needed an application built. Uh we looked at it and we looked at it the oldfashioned way and we said okay it's going to cost you 25 mandates days to build it. And looking a little bit further, the the project was shelfd a little bit and at a certain moment said, "Oh, now we need to build this." And myself and uh another engineer, we said, "Okay, let's do this and let's do this with clot." And we built it in three days.

So we just got a lot more efficient and yeah, we built the the client for 25 days. uh but he was happy he had his application you know the application was there it was on time it was it worked you know they they didn't care how we did build it that we used AI or not >> but did they care how much you you build them for the thing is we are still billing right our business model is changing because of AI and therefore we're probably charging the same we could we actually probably should be charging them more now with AI because it costs us a lot of money And if it's taken you less amount of time, but it could cost you more in the long run.

Yeah. Less hours humans spent doing the work, but more in terms of like spending for models. >> Well, in this in this example, we were we were lucky because we were still in the drug dealer model. And that is what uh the AIS are doing to us. You know, all all these companies, they gave us AI like for almost free or free, you know, everybody can use chat GPT for free. Everybody can use cloth for free. Still now if you start doing it professionally you had to pay a little bit but not too much and that era it stopped literally like two weeks ago. Yeah. If you now are going to use AI professionally you're going to pay for it. You know you have to pay for the the data centers of the company you work for. You have to pay

for that energy. The the drug deal model is over. And so yes we have to find and with us we haven't found it yet. We we're actually even thinking of giving our AIS a time sheet code so they can actually build the client like that like okay we weren't just two guys building an application. No, we were two guys and 14 agents building an application and that yeah you have to build that somehow to your client. So yes, our business model will change >> uh for sure. If there's anyone here today that is like exac accountant or ex like if you come from like the traditional finance sector, there is definitely like opportunities for you in

AI in terms of like because people don't know how to charge anymore. They really don't. They're lost. And with Frank, he's he charges probably the same. I'm not like 100% sure, right? You probably charge the same as before AI. >> Yes. Yeah. Yeah. >> That shouldn't be the case. >> My rate doesn't really change. And that that is that is a problem that we have right now. So, how are we going to tell our clients, "Yeah, okay, my rate is like €1,000 a day, but yeah, I use these 14 agents, so you have to pay uh a few thousand more." >> Mhm.

>> Uh, okay. >> So, what do you value more, your time or the cost it it it takes for you to actually build it? Right. >> Well, well, basically that was that was this slide. You know, we can we can build a lot more now. Um but we yeah we we have to adjust the way that we that we build our clients, the way that we talk with our clients. Our model is changing there. Yeah, the model is changing without us even realizing. If you go back to the slide, yeah, we're definitely asking the wrong question. Um I tweeted this recently. I don't think we're ready to give agents full control over our decisions just yet. Humans are still best, but for not much longer. So, I

don't know if any of you have seen this. There was a massive AI budget uh spent uh from Claude subscriptions and for like a good while, we didn't know what company it was. They were like afraid to announce it publicly, but it was actually Uber up. Um Uber spent uh 3.4 billion by accident, their agents spent. So, who pays for that at the end of the day? They're not going to get a refund of that money. Oh, that's that's easy. Did you take an Uber here? Did you take a bolt? Uber is more expensive. We're paying >> Uber is more expensive. You we're all paying that because of this mistake. So, blame AI agents.

>> Yeah. But yeah, somebody has to pay for it. You know, somebody have to pay for the electricity, somebody has to pay for for the data centers. It's it's it's as simple as that. As I said, the the era of the drug dealer is is over. You know, the drug dealer model, it's it's over. We're billing people now. the the the clots are billing us now and yeah we have to build that to our clients >> out of everyone here that has an AI subscription or like would you use it for free like raise of hands are you using like paid subscription >> yeah most people so we obviously value that it's given us better responses than the free version and obviously it does but I feel like we're not learning so

much the difference of if you if you're not a developer veloper, you don't learn really the difference of what each model is bringing you. I spent so much time researching it because I wanted to learn myself. Uh, and I I was just asking questions to Frank and I was able to create my own um app for a hackathon that helped for Coliseum recently. If you haven't heard of Coliseum, it's like one of the biggest hackathons in the world. Um, and in Ireland this year we got so many like new registrations because I was able to market it in a way I understood it because I also entered it. And I feel like AI agents is like most marketeers probably wouldn't be that hands-on. Therefore, the AI agent is learning the kind of bad mistakes and

like bad uh they're not learning like the best practices basically. >> Yeah, that was that was a fun project basically. Uh I was quite impressed on how far you you got with just using a I did go in like once a day in the evening to see if there were any architectural mistake. You know >> I basically broke everything and you fixed it. >> No, not really. It was just more like are there any architectural mistakes stuff like that I would fix that in the >> you should be breaking things to learn right you have to fail fast and so many people are afraid to do that especially with AI we're we're we're skipping that stage and it's such a important stage of like your journey and it's what we do in our actual real day-to-day lives so why

aren't we doing that in web 3. Yeah, true. Oh, it's already our last slide. So, the idea here is I at least it's my idea. It's AI is not replacing us and many influencers and so on, they they give us this impression and I I don't believe in it. But I think we will have to adapt our business models and the way we work especially people that work in technology uh because we will work together with AI. And I'll go back in time here because I I I love this example. I was working for ING at that moment and I'm talking about 15 maybe 20 years ago something like that and I read

a study uh about Microsoft and that study said if you give a developer a second screen he will actually be 17% more productive and I think with AI it will be the same the number will probably be be higher but if you give a developer an AI he will be more productive if he will get to his result faster. But to go back to my original story is of course yeah ING it's it's like a huge company uh it's a bank so it's a it's a very rigid thing it doesn't move very fast and yeah I I suddenly needed like 200 extra screens that was nowhere in the budgets um and so I got a lot of complaints from from

the infra guys like why the do you need 200 screens uh but I was okay I was able to manage that and then at a certain moment I got a call and I didn't see that one coming at all. Who did they get a call from? From the unions because they also did read that paper because I had to motivate the 200 screens. And they said to me, Frank, what are you doing? Are you going to fire 17% of the people? And we see this happening right now with AI too. You know, a lot of companies are laying off. And my answer was literally at that no I'm not going to fire these people. They're just going to work harder, you know, more efficient. Not

really harder, but more efficient. They will create more code so the bank can make more profit. So I'm not going to fire them. That was the answer that the unions wanted to hear. But it is something that we have seen now in this AI area that some companies indeed are laying off people. But I think they will regret it soon. >> I think so too. I think there's going to be a jump between saving profits, but again, that's because they're not looking at they're not changing their business model. They're keeping the business model the same, charging the same, and laying off people. They shouldn't be doing that at all. I think more people need to listen to our talk. I think people need to listen to Frank and myself more because it's the larger companies that think that they're like losing profits because

of AI and they're they're actually not. And this is probably like the closing thought that we do agree on that there's so many opportunities in AI, but we're not like taking advantage of the transition that's happening. We're just accepting AI without fully understanding like what's really changing in your dayto-day uh day-to-day life. >> Okay. With these smart words, I'll Any questions? >> We're finished. >> Any questions? No, maybe I have a question. Oh, you have a question.

>> Sure. >> Uh, I just want to ask you because you mentioned that the job market is actually changing based on the AI of course. And you mentioned that you are now AI expert in marketing, right? >> Yeah, I do use it every single day. Um, and most of my clients are like I have AI factory that I work with, Kudo Compute. Um, which is like massive in uh the UK in the United Kingdom. So, I have to adapt to it. But it's something I do use anyway as an artist. It's something I've been using for years. So I >> I like to say I understand it, but I will put my hands up and say I don't.

Frank is a developer and says he doesn't understand yet. So you think that each and every position should be like focusing to became an AI expert in their field or should because like lately I was checking the web three market like what is the open position and etc and I saw many roles like content coms a project product manager all in one which feels that this one person is actually doing like job for five people but thanks to AI Yes, but he he will he will have 10 agents helping him. He's doing the job of five people, but he'll have 10 agents helping.

>> I think is the way actually >> I think it's not an AI problem. I think that's a marketing problem. So all of my client, not all my clients, but when I first joined web 3, I was the graphic designer, the video editor, the content creator. I don't do that anymore because I understand like what my talents are and I I can I can I can work around it. Obviously AI helps, but we're still seeing like humans are best. So I still recommend to say I'm not like I'm not as skilled in in this. That's like not what I was hired for. Uh you could use AI, but honestly humans are best. So hire someone instead, right? Um that that would be my >> Yeah, but I agree. Yeah, I'd say one person can easily control five six

agents and those five six agents they they they will have specific skills and something like video editing that can easily be done >> teach them the skills right you need to like know the skills to actually teach the >> yes of course >> okay thank you thank you any other questions >> um thank you for your talk really good really interesting interesting. Uh you guys were mentioning that um you're also in contact with some primary schools that are teaching kids about AI and you Frank were asking the question about how we uh educate engineers at a university level to be

good skilled senior engineers or senior developers right? I see it at the primary level as just another new level of abstraction that that generation is going to have over whatever they do because the argument or the parallel for me is just go or C++ right so what is your current assessment of what is now happening in universities around teaching how AI AI works in depth right or in detail >> yeah I can go first it's I would be apprehensive about teaching it in like primary schools. College, yes, fine.

Secondary school probably okay, but primary school I'd be so apprehensive because like young people today, like a lot of my friends have kids. Um, they don't even post like their kids' face on social media, but yet we're training the AI on their face. So, there's like a really big gray area there that we're kind of skipping. Um, I like I love when people have like their their own privacy. Um, but anyone can take like if anyone here took a photograph today, right? You can straight away give it to the AI without my permission without Frank's permission.

>> And that's a problem like in the future. That is a huge it's going to be so like um it's going to be like IP related issues everywhere, especially when people like start off in a job and maybe they get promoted and then become the CEO, right? um without your permission that that's that's why I wouldn't be teaching and using like it's more the photography and more the the visual is where I would see the issue with with teaching them the actual tools itself and the education you can still teach them without using it in my opinion. um stop.

>> Yeah. But if I I'll go back to the question if if if I looked at the universities. So the guys that are coming out there indeed they did learn one or two programming languages. In Belgium the standard is two. Um and yeah another thesis there of mine is we don't need programming languages anymore. The programming language will be just be English. By the way tip for all the AI users here. Did you know that if you type a question in English, you will use less tokens than if you type your question in your native uh language? Interesting. Uh but the programming language will be just become English. So that the programming languages are going away. But that is

not what I miss from universities. what I miss uh and yeah you can only learn it by doing it I think is how to make an enterprise great application you know they learn to they create small applications in university but they don't work in a large team they don't have uh yeah >> software system >> yes and you learn that by doing it from the older guys you work with you know you usually have one senior engineer controls the team say oh you do that you do that you do that. But if we now give all that to agents, how are these young guys are still going to learn that? That's a little bit my problem.

>> I hope that answered your question >> a little bit too. >> No problem. >> I think we have room for one more, but >> you can hear. Yeah. So, I think this is um more of a philosophical question. if you have thought about um what this is going to look uh like in the future from a societal level. So now we are not replaced yet but we will have uh more and more time for ourselves and we will also be able to do a lot of different jobs like you said I can learn I could be a developer in 10 years if I put myself into learning how to use AI. So that specialties that each one of us has

is kind of gonna disappear. So what are we actually going to be doing with our time? What's what have you thought about this? >> Well, now we come back, I think, to my answer. Are you going to lay off the 17% of people? We're not going to lay him off. We're going to have you work more efficiently. You will be able to just do more. I I don't believe the Elon Musk theory where he says okay nobody here has to work and we will all get universal uh wealth and and we just get food for free. I don't see that happening and and definitely not in the the transition period that we are now. Yeah, we will have to work but you will just work a

lot more efficient. That's I'm thinking more 10 20 years if we continue at this scale we are >> okay I'm I'm going to pull my umbrella there and what she said I'll be retired by then >> he doesn't care >> I I can give a bit of insight but to be honest I I don't think that far in advance like 20 years is a I would never think that far in advance I think more in the scale of like two years time maybe five years time like we did a talk recently about quant quum and that's in the next 5 years 20 years we have no idea like um >> also if you if you look back the chat GPT it it was launched like what now three and a half years ago something

like that and we developers we were actually laughing with it because it it couldn't even write a decent for loop you know it it couldn't do decent mod look where now I can basically I can have it build parts of an application easy So try to project that up another 3.5 years. Yes, maybe we'll reach that level eight. Trying to project this 20 years. I don't think we can foresee that future. I don't think it's realistic to try to predict that.

>> It's scary for younger. >> It's definitely scary for younger people. >> Definitely. >> Yeah. >> Thank you. >> Thank you. >> Thank you all. >> Sure. Thank you. Well done. >> Well done.