After the Gold Rush

Thank you. Thank you. Welcome. Welcome to the keynote from Cornell on yet another topic on AI and business.

I’m your host, Lutz Finger. I’m a faculty member at the Johnston School where I teach several courses on AI. Including one where I actually replaced myself with a virtual copy. It’s one of those certificate programs.

It’s open to the public and it’s called Designing and Building AI Solutions. Beyond academia, I’m a startup founder and the CEO from a Gen AI platform for e-commerce. We build fine-tuned models for e-commerce and we help companies to survive in the new geo world and we will get into this later. I said I’m your host, which is not quite.

True because today we switch up things a little bit. I recently sold my company and I thought it makes sense to take a broader view on value creation in the disruption that AI brings to us. So it’s my pleasure to welcome Kurt Wagner to the show. Man, I know Kurt for such a long time.

We just looked it up. It’s 11 years. But Kurt is a seasoned journalist, somebody who I really admire. He covers a lot of topics.

He’s a very famous person on technology, social media and how people shaping the digital world, meaning the intersections of humans and technology. He is currently at Bloomberg, but he’s probably more famously known for the really amazing book Battle of the Bird. If you haven’t checked it out, do it. It is a thoughtful look at the dramatic transformation of Twitter or X, I guess, under Elon Musk.

So, Kurt. Welcome. Thank you. Because Kurt is so much known for his curiosity and his pretty sharp look on how technology and business and culture intersect.

I thought I’d bring him here and we talk a little bit about the industry. Therefore, it’s a pleasure, Kurt, to have you on the show. Yeah. My pleasure to be here.

Thank you for including me. And thanks for the book shoutout. I always show up any place that is willing to shout out the book. So thank you.

So, Kurt, let’s. So, let’s say. like when i say i’m a chorus this is different normally i do a lot of discussions and i kind of trying to be to channel my inner kurt now i have the life kurt here so you take on a little bit over the show but um let’s kick it off with when did we meet how did we get together i was gonna say we we were just talking about this before and uh when i way back before bloomberg before the book i worked at um a place called mashable and you were working at linkedin and we did a project and i feel like it’s worth bringing up because i think it’s related to the topic at hand today about ai which is that you used linkedin and you essentially took my profile at the time and tried to project where my career was going five years down the road so you looked at other people on linkedin who had similar you know uh correct me if i’m wrong but educational experiences certain stages of their career and you said okay what’s the most likely path for me um you know five five years from now and it was a pretty fun thing to work on together you on the you know ai and machine learning data science side me on the sort of the guinea pig journalist side but i think it’s relevant because again that was 11 years ago we’re still talking about versions of this technology it’s obviously changed a lot since then um but maybe you could you know i realize it’s a very big question to start but when you think about that project this idea of you know taking the data around who i was at that time and sort of projecting forward how much of that is sort of still the the very foundation of what we’re talking about today and how much have things changed fascinating so technically wise we still do this right so essentially technically wise i looked at your profile and then i looked at 300 million at that time 300 million linkedin users and tried to figure out who used to have that type of a profile think about amazon’s people who shop for this shop for that so here i’m looking at people who look like kurt what are they doing five years out and as i told you at that time pope wasn’t in your future now yeah right you remember now no no major league baseball either unfortunately it was a it was a sign of a soul crushing conversation and i think it’s a it’s a it’s a you know you closed a lot of doors for me at that point lutz so now you didn’t leave meshable three weeks after i told you in the future i’m not sure whether this had something to do with it yeah well i believe you the projection based on the data was that i would be working in pr so i’m i’m somewhat proud to say 11 years later i’m still in the world of journalism for better or worse uh which is good i i guess what i keep finding interesting is that is that the technology, as you said, the sort of the kernels are maybe the same, but it feels like we’re at a very different place. Like at that point, it didn’t feel like we were anywhere near, you know, certainly AI assistants, chatbots, these LLMs that are able to, you know, drum up pictures with just a five-word text prompt.

So, again, at the risk of starting too big here, I’m wondering what made it possible? Like why is it so much better today? Is it just simply that the chips are better? Is it there’s more investment than there was 11 years ago?

Like if you had to pinpoint one thing as to how we’ve gotten in the past decade, what would it be? So it’s probably like we have more data and more power, right? So essentially the modeling behind, I can make a model to find people like you, like five years out. And I still use.

By the way, this very exercise I still use in almost all of my courses, because first of all, I failed, right? You didn’t go into PR. That’s right. I predicted the average.

The average person coming from journalistic backgrounds is going into PR. I don’t want to step on anybody’s toes, but PR pays better. And if you’re an average journalist, you go into PR. Now, why didn’t Kurt Wagner go into PR?

Because you’re not the average. You’re actually better. Which essentially, and this is a theme which we had on other shows, AI is not going to replace us because we are not doing average, right? So my predictions failure is actually good news.

I can create an average information, but I cannot predict the extremes parts of it. Now, what has changed to come to this part is actually. The thing is, you can picture your career, and you can picture your career forward, right? Because you have seen many different careers.

If I say, like, think about a PR career, you can picture it because you have seen PR folks, meaning the more we learn, the better we trained our models. And what we have seen lately, this big shift towards generative AI is we take a model that has a prediction and just turn it around. And then we have a prediction. and now we can create that prediction.

The same way how you can picture a blue elephant because you have seen blue and you have seen an elephant, which trains your model. And now you create a blue elephant. The same thing we do for generative AI. We take a standard model and just turn it around.

And instead of saying, this is what Kurt is going to be in five years, we can say, describe what is going to be in five years. So tell us a bit about, you mentioned at the very beginning, you recently sold your startup. You had an AI startup. Give us the quick rundown.

What were you building? Why did you choose to sell? Yes. Actually, let’s start off why I chose to build because that is, I was president of a NASDAQ listed company out of New York in, um, in the healthcare space, right?

So, um, I ran product and technology for healthcare. It was a third party administrator. We built predictive algorithms to figure out how to reduce the burden of diseases for our members. And I left that in order to build the startup.

And the startup was essentially a fine-tuned large language model. That doesn’t sound very fun and sexy, but essentially we see, Chad, LGBT coming around. We see large language models coming around. The technology behind it is transformer models.

And it’s a new thing. Transformer exists since 2017. It doesn’t seem to be that new, but everybody using it becomes a new question, a new setup. So the question is how can you create value if you, in this case, Lutz knows things the industry doesn’t know yet.

So I started out with, um, some very, very talented folks from, um, from Cornell to, um, to build out fine-tuned models in this case for e-commerce. So essentially if you are any e-commerce shop, you need to do search and discovery and search and discovery is a, like is an old topic, which was used with old architecture. And I threw the new technology on reason behind is, um, I can solve, this problem with this new technology better than anybody else. And that’s value creation for sure, because everybody very soon will try to solve that very problem.

And that turned out to be true. And so essentially, like, let me explain to you like a situation where you like just to make it here. You go to a website, you go to the website from Cartier, for example, Cartier has many rings, many things to, to offer, but you are used to a chat like interface now from chat GPT is that you can actually say, I want a retirement watch. What is the concept of a retirement watch?

How should it look? Should it be round? Should it be with leather? What, what does a retirement watch look like?

Now, if you go today to Cartier, that wouldn’t be possible to figure this out. Well, with the right technology, you can actually bake all of this in all the understanding, the brand understanding, which is specific to Cartier or to any other brand is not necessarily baked into chat GPT perplexity and others. So I create fine tuned models to help e-commerce companies to offer the right products at the right moment on the website. And you ultimately decided to sell what went into that decision.

It’s never, I mean, depends on how much the chat, the check is for, but sometimes it can be a very tough decision for people to make. So I’m curious if you wrestled with it, if you felt like this was, you know, why was this the right time for you? No. And by the way, we should say, and I think there was a, a quick notion on the screen for all the listeners, go and ask questions.

It’s you will have very like Kurt and myself, we will look through all of your questions. And if you have a question about Twitter, my, my startup, the future of the internet, AI, will we get all replaced and, and so on and so forth. Just ask them here, Kurt and I are here to answer. So why did I sell?

And so the initial impetus for me to start was okay. New technology. Everybody will need it. I don’t know necessarily enough of the market, but, but I’m pretty sure what I build will be needed.

And then I build up the company and we come down to Porter’s five forces, right? That’s what I always tell them that in like to students as well is business hasn’t changed because we have a new technology. Meaning I realized the difficulty from going to market. I’m realized the difficulty from being a new entrance.

And trying to replace existing solutions. And if I want to come back to your book, because you talk in your book actually about Twitter at that time, X now about this success and that this wasn’t necessarily engineering driven, but it was only feasible because we had the internet in a broad available state. So there was an engineering change or technical change, which made. Yeah.

It’s possible. Maybe like just like just to, to compliment my story here. Like maybe you give a view here. Yeah.

I think that there wasn’t necessarily anything super special in my opinion, from a technology standpoint about why Twitter became such a cultural force. I think, you know, and, and I think we will see this in AI and we should talk about what, like which of these AI things, these AI, AI projects, these AI bots, these technologies are ultimately going to rise to the top is probably not because, you know, in some cases maybe they will be head and shoulders better than the other, but I think we’re getting to a point where a lot of these things are going to be able to do the same stuff. And so if the, the technology ends up being sort of commoditized, well, what differentiates any, you know, any one chat bot from another, I think the same could be said about social media 10 years ago, which is what I’ve been covering this whole time is that, is that anyone could kind of go on and, set up this framework where you can make friends, you can post to a feed, you can like things, you can reshare things like that. Wasn’t really all that secret in my opinion, but what Twitter and ultimately, you know, I think what Facebook did first was they created the friend graph, right?

So they got so big, so fast that you couldn’t afford to leave because all your friends were already there. Your social connections were the thing that kept you glued to Facebook. So that was sort of their secret weapon. And I think with Twitter, their secret weapon became a cultural relevance around the news, quite frankly.

And it was the tagline for the company for a long time. What’s happening. It’s a little cheesy, but it’s true. Like there was a time when, you know, during the, the first Trump presidential administration, for example, when it felt like every day there was some kind of new, bold, you know, drama that, that Americans had not experienced before.

And, and you could open up Twitter and expect that that thing that everyone is talking about would be the first thing that is you’re greeted with when you get there. And so what I, what I mean by this is like the technology wasn’t the differentiator in my point, it was the culture that was built around the technology. And so I imagine, and we should talk more about this, that that will be what ultimately the decipher is who’s going to win and lose in this AI race as well, is that someone will build something that’s culturally different and that’s going to be a big, a wonder machine in general, but also who can then also be able to select who can then select who can select who can then select who can select who can then select who can select who can then select who can select who can then select who can select who can then select who can select who can then select who can select who can then select who can select who can then select who can select who can then select who can select who can then select who can just like they search for 410. 410 is one of the shoes from New Balance.

They come on and search 410. What do you mean a conversation? Well, in the future, you will have a conversation because that is what the technology brings us towards. So you go to perplexity today and might say, I’m a 50-year-old person and I just decided that I want to start running again.

What type of shoe do I need? Should I run barefoot and so on? So you have a more complex and rich space where you can actually put e-commerce into. So now you have the technology who could do this and you have a changing social graph.

Those are slowly coming together. And essentially, very much like Twitter and Facebook, we have the same situation, right? We had a changing social norm of how you communicate and how you communicate. And so you have a more complex and rich space where you can actually put in a conversation.

So you have a more complex and rich space where you can actually put in a conversation. And so you have a more complex and rich space where you can actually put in a conversation. And the technology was there. And the question is, who is actually managing Fortis Fire 4 was the best access to the market, right?

And coming back to your question, why did I sell? Because I realized that, yeah, technology, I’m extremely good. Changing norms, I predicted this correctly. It’s happening.

More and more people are going to have conversations with perplexity to sell. I wrote like a while back the article that I’m going to have a conversation with that Amazon’s homepage is dead because people are not going to the aggregator Amazon. They’re going to have a conversation. So I predicted those two things correctly, but I still were missing the cultural adoption norm similar to Twitter, right?

So Twitter managed to get the adoption, but did not necessarily. But there were many others who had similar technology, were at the right moment at the right space, but did not. They did not manage that go to market. Yeah.

I mean, so much of it is indeed timing, right? And you see so many good ideas that come before their time. And then you see good ideas that come too late. And it’s like, can you have the great idea that matches with the technology that matches with the time and the demand?

I think there’s a bunch of good questions pouring in, by the way, which we should start maybe pulling from. But one more kind of big one that relates, I think, to you choosing to sell now. I’m wondering if you think we’re in an AI bubble, quite frankly, because I realize that’s a big, maybe loaded question. But there seems to be just so much money pouring into this space right now.

I’m covering these companies that are building $50 billion data centers, in Meta’s case, in rural Louisiana, right? That’s not going to be ready for five years. And some of these deals, these multi-billion dollar deals feel like rounding errors at this point for these companies, which just a few years ago would have been laughable. What do you think?

I mean, do you think this is getting too big too fast? Or is this all justified in your mind? I think there are a couple of facets for this question. This is a super question.

Thanks, Kurt. Vivian, one of the folks on the audience, she actually asked, one of those facets. So Vivian actually said, how do you think builders today can still find differentiated value and workflows that already seem claimed? Or in other words, would white space still exist for those building the next generation of AI enhancing workflows?

I think we have a technology and that technology is essentially, let’s claim it, it’s an interface to computers. As I worked for Google Health, or as I did my healthcare company out of New York, one of the workflows for healthcare is faxes, right? You have amazing models, but healthcare communicates in faxes. So if you want to do an impact in healthcare, you better have an interface, which is called fax, because that’s the existing workflow.

And I think much of our AI world, looks a little bit like snake oil, or people get this, like unhappy about it, because they realize AI in itself is not the answer. And by the way, this is what, again, I started it because AI was new and we had the ability to create something which people are going to need. I sold because I realized the workflow aspect, especially on the AI side, is going to be a big part of the ecosystem. There are also these new technologies that are going to be falling into place in the ecosystem.

There are also these new technologies that are going to be falling into place in the your question about it are we over hyping it we have not even touched the button like touched it it’s it for me this is similar to as we were in a bubble with amazon and we looked at the amazon bubble like early days of the internet hype and it collapsed and everybody was like look we will all sell online and we will like be we will always shopping online and then there was excitement and a few months later but funny we still have brick and mortar it didn’t change like so this is all a fad no it’s not it just takes longer and here no ai is not a fad but we have the workflow problems to address it feels like there will be some contraction at some point right because the amount of money being spent there’s all these deals as well that are really tricky you know it’s like um tech! video There is also I’m fully on board with you, Lutz, that this, it does not feel like a fad. Like, I would be absolutely shocked if five years from now, 10 years from now, we’re like, hey, remember when everyone was into AI? Whatever happened to that thing?

Like, this isn’t going away, right? This feels like the foundation for our future as a society. But I do feel like we’re at this frenzy right now on the deal side, where the amount of money changing hands between these giant tech companies feels like there’s going to be some kind of correction at some point. I don’t think everything goes away, but it feels like we’ve expanded quite dramatically, quite quickly, and there will be a contraction before we ultimately move forward.

I don’t know when that will be. If I did, I wouldn’t be talking to you here. I’d be probably making oodles of money in my Wall Street job. You would offer me the advice you would tell us to the Cornell community.

I’d be on my private yacht. I’d be on my private yacht, unfortunately, if I knew the answers to these questions. But it just feels like we’re in a weird spot financially. So I will.

I will say I will certainly say that. Now, like, like, you know, the market’s way better than I do. And I think there are two facets we could address. And in terms of.

Why people are cautioning the current setup. One is this like insider support system. Now, I and the other one. Is where the funding comes from.

And I think both are very fascinating topics. So the insider support system is not something which is so unknown. Right. As we had the initial part of the Internet, you had media companies offering access to customers.

And taking shares in innovative companies. Axel Springer, one of the big media houses in Europe, made a fortune off their Internet portfolio. Because they got shares in companies because they offered them marketing space. And I think you have something very similar here.

You have companies who have Nvidia, like bought a lot of Nvidia chips. And because they have this Nvidia chips, they are now able to offer. Compute ability for many new startup ideas. You started like you saw venture capital gigs coming up, trying to do exactly that.

Right. Friedman was one of them. That is this kind of like, try to figure out where is the value and tag on it. In the early Internet times, the value was access to the customers.

And now I would argue the value is. And again, Porter’s five forces helps. It’s like supplier. Nvidia is a value to have access.

As well as again, access to the market. The same reason why I sold. Because I wasn’t as good as having access to the market of people who trust my technology to use it. Despite being good.

So I partnered with somebody who has the access. And made it worthwhile. Right. So these are, this is one direction.

There’s another one, which I think you can describe better than I can. We see a lot of. Debt financing for investments. Like, can you explain, like, explain a little bit about the debt financing backdrop so that.

Man. Listeners understand. I’m going to do my best. I’m not a debt expert, but I’ve had to become a little bit of a.

At least a student of debt and a student of what are called SPVs, which are. You know, these essentially. Single entities that are created exclusively. To invest in a particular project.

So for example, I cover XAI. We had a story a couple of weeks ago. About XAI raising $20 billion via an SPV, a special purpose vehicle. To buy chips.

These very expensive Nvidia chips. And so the SPV is set up. People put money into it. Think of it as a giant fund.

That fund goes out and buys all these chips. For AI purposes. And then it’s like. The chip.

Goes back into the chip. The chip. Goes back into the chip. And then it’s like, well, we have to make more money.

For AI purposes. And XAI actually rents the chips. From the SPV. So the people who put money into the vehicle are now making their money back.

In some ways via. You know, rental fees from XAI. It’s all. And the point of this, I’m told is to essentially.

Keep the debt away from the company’s balance sheet directly. Right? So now the debt is held by this. Third party entity.

paying to rent these chips from this vehicle that’s been created. I share this mostly because it is complicated. It’s like the things that are kind of coming up in order to the creativity that is happening on Wall Street right now in order to fund these massive data centers, these massive chip projects, these AI things, it’s getting very interesting and hard to follow to the average person. And maybe that won’t matter at the end of the day.

Maybe it’ll just be whether the money is with XAI directly or in this special purpose vehicle. Maybe none of that ultimately matters. But it feels like we’re seeing some really interesting accounting happening right now because the numbers are just so large that these companies are afraid to put all this debt on their own books. Now, there is actually a very interesting question from Oliver to that part.

So small little secret about academia, right? We always look back. We look backwards to trying to figure out what we can say about the future. Meaning if you talk about, you explain the vehicle to fund the enormous sums that XAI just needs for training, Google, Matter, and others can way better hide them in their existing cash flow.

And OpenAI hit it by having the friendship to Microsoft, which kind of… I don’t know if it’s not lasting anymore, but that’s a completely different topic. However, all of this points to the need for how do we train models and that points to a need for power. And that leads us back to the farce.

So Oliver’s question was, do you foresee the demand for power being as high as it currently is projected to be or will AI solve this appetite for power? And demand will much lower than anticipated. I… And…

I expect that… The need for power will grow. And what we see at the moment with all the investments in data center is that we are putting this infrastructure down. And here comes the interesting part.

Initial infrastructure investments were financed with equities. And now with the SPAC and other areas, we financed it with debt. And if you look backwards in history, we did this for the telecom industries as well, right? So suddenly we need all the…

We need all the infrastructure in place to build large telecommunication networks. And we financed it with debt. And then we have over-debted companies. We take a haircut.

They go out of business once, twice, three times. What is left for society is the grid, which we are using today and which helps our community today. So I suspect there will be on the infrastructure play. We will see something very similar because as soon as companies go towards a debt route, it’s obviously a completely different way of building up value than if you go via the shareholder route.

Do you think part of this is… I mean, part of how it’s been explained to me is that owning a lot of this technology, the chips, even the data centers, like these are depreciating. These are going to depreciate in value, right? The chips that you buy today are going to be worth, what, 20% of their cost in five years.

Just because technology… It’s like buying a new iPhone, right? Five years from now, you’re like, this thing’s worthless because the technology is so much better. So I do wonder, and maybe you would know, Lutz, but it feels like part of the reason we’re seeing this creativity around SPVs and the example I mentioned with XAI where they’re going to rent the chips instead of buy the chips.

Is this idea that the technology is advancing so quickly that owning this stuff might not actually be the best course of action? Putting equity into it versus simply taking out a loan feels like maybe a smarter play because of the evolution of the tech. Again, I think it’s early and these things are going to continue to morph and evolve as… As…

You know, the understanding for how much power and data centers are needed, like these things will fluctuate. But it feels like that’s maybe part of the equation here as I’m covering these big companies. Yes. And I think so.

This is the main core question and the multimillion dollar question, right? Which puts you on your private yacht. Is how do… Like we see an investment in infrastructure on one side.

And we hope… Or the companies calculate that this investment is sufficient. Like take Mark Zuckerberg. Brilliant move to buy so many Nvidia capacity.

As he did, his shares took a beating because they said like, oh, well, like, where is he spending all this money? And then a couple of like two quarters later, people’s like, wow, he has all this capacity to actually build out new generative information, which is very helpful for his network. Because he’s… He wants to be the new platform for creators.

He is a genius, right? And suddenly that capacity was good for his share price while before the capacity was not. So the question is, will we be… Yes, there will be decay in like chip capacity.

Like there will be decay and like value spent. But will we create sufficiently fast value compared to this… Make, yes. Like to play.

So there is Suzy, for example. She… Thanks, Suzy, for your question. Thanks.

Well, there are so many questions. Thanks, everybody, for putting the questions down here. Suzy writes, I’m a brand strategist and I augment my process with AI. What should I keep top of mind at the stage of the process?

Building and training while recruiting new users and generating revenue is quite overwhelming. And this is now… We go from the big picture of we have power, we have chips down to one use case, the use case of generating images for brand strategy. Right.

And I would claim we don’t… We only scratch the surface of how the future world will be looking. If I’m a brand today, I protect my brand and I’m trying to convince Google. Yes.

To list me, right? Like SEO. In the future, I want to have chat GPT and perplexity to tell my brand story so that when somebody has a need for that piece of brand story to be detected. So when I’m searching for like, again, I’m an overweight guy who tries to start running again, what is the right shoe for me?

And please give me the right link. And this is what we just saw from OpenAI. I’m pushing into it. That’s the reason why I believe Amazon’s homepage is dead.

But we can sketch this even further. I just posted on LinkedIn, I think two days ago about Sora2 from OpenAI. Where at Sora2, you can only generate fake videos with using AI. But the neat part of the network is I, as Lutz Finger, can give away my liking to anybody who wants to use it.

So I’m at Sora2. I’m under at Lutz. If you want to friend me, I friend you back. If you friend me and I friend you back, you can use my liking and make any cameo of me in any way.

And I tried this out with students in order to see how easy it becomes to create fun cameos together. Now, why is this important? Because, Susie, you are a brand strategist and your whole workflow was create a brand, protect a brand and publish it out. And that publishing is changing because ChatGPT takes no over that role.

I would claim there is a new facet to it, which is allow that others create with your brand. So for the publishing part, I created something called Query Edge. It was built in my startup. We monitor where you can, like if you’re a brand, you can go to QueryEdge.com.

You can buy it. It costs you $140 a month or so. And you can buy a monitoring solution in order to see how the publishing works. But I would take it one notch up.

And the one notch up is to say you can use your brand to create something new. Meaning if you’re a brand, you are like you are Armani or you are Drake the singer, right, the rapper, you can now allow people to create in your name in Armani like fashion or a Drake like song. So now you, Suzy, as a brand strategist, there’s a total new problem. It’s not only how do I build and protect, not how we all build, protect and sell.

Now, how do I build, protect and sell the idea so that people can create it? So everybody becomes a little Twitter. Well, and can I just add, I think that the answer to Suzy’s question, which is a good one, but she is not the only person asking this. I went to Cannes Lions this year, this summer.

Tough work assignment. Go to the south of France for a week, which, you know, there’s my yacht right there. But I will say that I and this basically this very question dominated the entire week. Right.

Because everyone there is in some way in the advertising world, in the brand building world. And everyone is trying to figure out how is this going to change the relationship that our customers have with with this brand that we’re trying to build and not everyone has the budgets to that Coca-Cola and Apple and Amazon have to build their brand. And in those cases, I think any is going to be tremendous. Right.

If you’re a small local pizzeria, suddenly you can use Sora or the Gen AI tools to create. And advertisements high in messaging and marketing in a way that you never could have afforded before. And so I think it will lift up probably the bottom end of the advertising world hopefully will stop seeing from me ads, you know, those text heavy like things that look like they’re flyers that should be taped on the telephone pole, but for some reason you’re getting it in your feed. Hopefully those things will be gone.

But the question is, how do you you know, and again, I don’t know the answer. So but I’m agreeing that I think it’s an important thing to ask is like if you are the apples of the world, the Googles, et cetera, how are you kind of differentiating as all of this stuff gets better and the playing field sort of gets leveled? And I think that’s going to be hopefully that’s where human element comes in, right? Like, again, we come back to the conversation we started with Lutz about Twitter and why Twitter.

It’s like when the technology is sort of ubiquitous, like there has to be something else that that helps you stand out like a human element that helps you stand out. I presume that will be the case in sort of this brand building as well. But I think we’re still pretty early in the process. And the totally we are totally early in the process.

That’s it. By the way, that’s the exciting part of the value creation, because all the companies at the moment trying to figure out like a how do we use the technology for the business set up, which essentially is a story which I went through with my startup or you are alluding to with Twitter technology is ubiquitous. How do I create the social form? Sora2 is a fun fact on the social form because obviously you had Meta.ai realizing that OpenAI is launching it, they launched before they did not build the social impact structure from Sora in.

Now for the business part. And by the way, just like from a very high level, we have technology and we need to figure it out how it touches social and how it society and how it touches business processes. And for the business processes, Vincent actually asked something very interesting, which kind of could help us frame this. Why see now emphasize that the best way to identify opportunities in AI is by taking Palantir’s forward deployed engineering approach.

And of course, do you like to. Do you want to say something about the Palantir’s approach? Do you know that? No, I was I was hoping you were going to explain.

I was this is yeah. You pulled this question. So I was like, OK, hopefully, hopefully he knows where he’s going. You know, I totally know where I’m going.

OK, good. So Palantir made this forward deployed approach. It’s like it is at the moment the rave term here in Silicon Valley. And I think it’s a good thing because people don’t know yet how to apply the technology.

This is me with my eye for e-commerce. I built a tool which was in all and I do this publicly. I do benchmark tests and I’m better than anybody off the top of like like the top tools and better than the all goal years of the world and so on and so forth. I was good.

I was a good technology. But do I understand e-commerce? No. So that’s a problem.

That’s my go to market problem. This is where Palantir’s approach from forward deployment of engineers come in. Essentially, Palantir says, I don’t understand your business, but I come in with smart talent and I will figure out the workflows where I can attach to or very simplistic put. Everybody at the moment is out in the world trying to hire a CIO, like AIO, CIO, whatever that is.

Palantir would say you don’t need that. All what you need is get my my team in. I figure out the right processes where I can throw technology on in order to make this better. So, Vincent, I do agree with your point of view here.

If we say there are two ways to bring value in AI, one is on find the adoption in society and the other one, find the workflows to make them smooth. The Palantir approach for workflows, that’s exactly it. Because if you are good in technology, you don’t necessarily know the workflow. That’s the reason why I sold to Xgen AI, because they know the workflow extremely well.

I’m glad you’re here, Lutz, because I would have I would not have been able to explain that quite as well. Well, can I can I pull a question from this roster, though, here to take us in a slightly different direction? But I think it’s a little bit of a different question. But you’re a parent and I’m a parent of two, two small kids.

Your kids are older than mine. But Rachel asked, how do I prepare my kids for this new world? Rachel is a three year old. And how do you prepare?

How does she prepare herself? I’ve been asking myself this part of why I want to pull this out, as I’ve been asking myself this question a lot as well as a parent of two little ones. And as someone who’s covered social media for over a decade now, it breaks my heart still when you see the the stories, the inevitable stories of harm that comes out of social media, right, teens still getting scammed. Sextortion is a big deal.

You know, self-harm, the types of content that people see, obviously, when they’re when they’re young is just so influential on them. And every time I see those, I some of this is unfortunately going to be you could be the best, most involved, most active parent. But you can’t just, you know, be in your child’s brain 24 seven. Right.

There are just going to be things that that unfortunately you have to to let your kids figure out for themselves. But what does bother me is when I hear these stories and and it feels like parents are completely dismissive of the technology or they said, well, you know, I don’t know how Facebook works, I know my kids are on there, but like, you know, that’s for that’s for them. That’s not for me. And and it bothers me so much because I think if parents take the time to understand this stuff, they will not only be able to talk to their kids about it in a way that that, you know, to help explain what these harms are, but they might experience some of this themselves, but be in a much better position to deal with it.

Right. So if you’re an adult and you come across, you know, recommended misinformation on YouTube or Instagram or whatever, and you’re like, wow, this, you know, as a grown up, I know how to handle this. Hopefully I’m going to talk to my kids about it. But if you if you’re not on these technologies, you just don’t experience that.

You have no communication. And then your kids are experiencing this without any type of guidance. And so the obvious, you know, what I am going to attempt to do is stay on top of these new technologies. It’s easy for me in a sense because it is my job.

But understanding what my kids are using so that we can talk about it feels like just a base level thing that that parents should be doing. And sadly, I think a lot of them don’t. And so, Lutz, you’re in technology. I’m sure this wasn’t probably an issue for you and your kids in the sense that you you were on the cutting edge of a lot of this stuff to begin with.

But I’m curious, as someone who has older kids, do you have any advice on how to navigate like an emerging technology with young people? As as we started to like I’m like before we actually met, I had my own social media measuring company. I sold it to WPP. And we.

We kind of like we essentially I worked for governments and NGOs. One of my fun stories at that time was we measured how social media reacted to the Snowden scandal. One of our customers was the EU government. And my data made it in the back like over to Obama with Commissioner Redding, who actually talked about how the EU perceived the Snowden scandal.

Now, at that time, we would have not known all the negative externalities social media will bring. Now, make no mistake, I think we have found a completely different way of society to interact with each other because of social media right now with AI and the ability to create content at a dime, meaning I can create fake videos, fake images, fake news, fake stories. I can whatever. Right.

And. We will face a different world and court, it pains me to say this, but like what reality is, how like like early on to to one of our customers, I think it was Susie who asked about a brand. I said, well, don’t think about brand protection. Don’t think about brand selling, which is currently.

But think about how you can use a brand to recreate and enable people to creation. Right. Meaning I give up what was protected as a brand. Now, we probably have something very similar in how we what we call news or what we call truth right there already today because Internet made many truth searchable, social media made many truth distributable.

Sorry, my German here. And the what will I do? I will allow us to create as many truth as we want to have. The person I really think has brought this very nicely together is you.

You’ve all Harari in his book Nexus, where he talks about that content generation and the collection of content generation has always been a power play and power factor. He talks about the burning witches and that there was a lot of content explaining what is the witch we today know that we just don’t exist. Right. Well, I don’t want to poop the party for anybody who thought they do.

But like witches don’t exist. And Harari’s point is all the information hasn’t helped us necessarily to be a smarter society. So now we are at a place where content generation can happen. And so easily that we can create any convenient content for any convenient set up.

I mean, this is this is a world that I am in every single day with the companies I cover this idea of misinformation using generative A.I. to intentionally mislead or unintentionally mislead people actually interviewed Adam Masseri. He’s the head of Instagram at our conference was last last week, two weeks ago, sometime in L.A. and I sort of asked him about this, which is, you know, whose responsibility sort of is this?

Is it on the person who creates the content to to knowingly wave their hand and say, hey, look, I use this in A.I. It’s fake. Don’t be duped. Most bad actors are never going to do that.

Right. I don’t think that’s a reasonable thing for for someone who’s intentionally trying to cause havoc. Um. Well, is it the company’s responsibility to figure out that, you know, this is misleading and take it down or or label it in some way?

They don’t want to do that. They haven’t historically been great at doing that. I don’t think people feel comfortable with the idea that Facebook or Instagram or Tick Tock or whoever gets to decide what should be true and what’s not true. So then you say, OK, well, is it the government’s responsibility?

And that’s a thorny issue. And I think that’s true, too, because I don’t think people want the government also telling them what is true and what’s not true. So I bring this up to say that I think we’re entering into what I would consider a rather scary time when it’s easier than ever to create something that feels very real but is not real and that can be playful or that can be very harmful. And there’s nobody out there who is picking up the mantle of saying, I want to, you know, be the the police around this new environment.

And so without anybody scrambling to say we want to be in charge of labeling the truth and not truth, which, again, I don’t want to do either. That’s a very complicated, weighty, heavy task, heavy burden. But no, without anybody stepping up to do it and the fact that it’s going to become easier and easier, I just think we’re getting into a world where people are going to be duped and people are going to scroll online and quite frankly, think you’re never going to know for sure if what you’re seeing is true or not. And the better the AI gets, the harder it is to do that.

And I think that’s going to become. And I don’t know what the solution I’m painting a debit. You’re you’re raising your finger. You I’m painting a bit of a gnarly, dark picture.

But maybe you can maybe you can spin it forward in a more positive light. Let’s damn Kurt, your heart is gone. I best my yacht’s gone. No.

So, Rachel, you have a three year old and Rachel, you probably have a very sweet three year old and you a three year old is not duped. You a three year old will grow up in a world where it understands ambiguity and where it understands that information trustworthiness is a is a completely different scale. If I they are like I can create fake information to to harm. I can create fake information to explain.

Right. I can have a tool set to support me in order. To better explain something, we have all this amazing technology out there, which makes it now easier to create logic and follow through. And the the differences between like so the question what is true is such a complicated question, and that’s the reason why we have usually had something like science to have discussions around what is true.

And then scientists battle over the different nuances. And we are opening up this ability now for everybody to take part in. And it will take some time. And yes, it probably will have negative consequences, like negative externalities.

But to Rachel’s point, the best way for us is to see how we can create value and allow all the different technologies to interact with each other and work with ambiguity. And I think that’s the skills. That’s the skills which I teach my students. Right.

To like look at the different technologies and see in which set up you apply them, that’s the skills also we should give to a three year old. Maybe this comes back a little bit to the brand discussion we were having earlier as well, which is that the value of brands will become, you know, in a world where you have trouble deciphering what is true and what is what is not. Trusted, valued brands. There are also these in a way that elevates, you know, people who do the real work about, I’m talking specifically to journalism at this stage, but like people who are doing real journalism versus just opining on X or Instagram or TikTok or whatever.

I think maybe that’s the idealistic way to view this is that there will still be reliable sources of information out there. And in fact, they’ll become even more important in a world where you have trouble deciphering real versus fake. So maybe. So I don’t know whether this will be so easy to decipher what is real versus what is different, like versus what is fake versus what is a different opinion, right?

We are in a political environment at the moment where we have many, many diverse opinions. And it’s not that one is wrong and the other one is right. They are diverse. Now, I believe that the tool sets which we have can actually help us to find and decipher the complexity of our human life more, right?

And I will give one example before we break. It is in healthcare, we have models to determine whether somebody is sick or whether somebody needs certain treatment and so on and so forth. And obviously, if the data is biased, then And I think that’s the idealistic way to view this. And I think that’s the idealistic way to view this.

The model is biased. Like there was a time where some minority groups in the US were treated with less painkillers because we assumed there are more pain resistance. Guess what? This is wrong.

Now, because we feed that information, everybody gets less painkillers if you’re from that minority group into the model, the model says if you’re from the minority group, it seems to be sensible, average, right? To tell you, to give you less painkillers. The good news is, because we can turn models around now, as we discussed in the beginning, the good news is that we have those models, the good news is that we have the data, and we can actually uncover that situation. We can uncover a situation where fake information is used to skew opinions.

We have the tools. The question is whether we use it, but theoretically, we should actually be way better empowered in this new world. Lutz, I know we’re near the end. So I want to ask a question that I came with, which is, is there any concern for you that it feels as though the future of AI right now, it’s not a regulated industry, right?

There’s really no guardrails around AI or tech at the moment beyond what is self-imposed. Does it make you nervous? Does it bother you that it feels like this industry is being driven by, quite frankly, like few very wealthy, powerful individuals, right? I think of Sam Altman, Mark Zuckerberg, Elon Musk, like the people at the tops of these AI companies right now are setting the foundation for the technology we are going to be using for the next generation.

For the next 100, you know, probably forever in a version, but like that will impact our lives for the next 100 years. I don’t know. To me, it feels like some huge population changing decisions being made by a very few select people. I’m a little concerned about that.

I’m curious if you’re concerned about that as someone who’s been on the building side of this. So I could ask, and like, could ask a similar question about journalistic landscape, right? If you think about who owns journalistic institutions, it’s owned by a few billionaires, right? And you know, from Italy, that this might not be the best thing.

If you look back at the Belvis County time. Now, the question is not whether I think regulation will help or not. Like, essentially, I don’t think that regulation at this point is going to help. I think that regulation at this point is going to help.

And so my hope with those discussions, which we have here and the discussions we have with Cornell as an institution is to get people who listen to us to get people, our students to get people who engage from the alumni side with us into a place to moderate our society so that we apply technology for better and trying to restrict the bad uses use cases. And so my hope with those discussions, which we have here and the discussions we have with Cornell as an institution is to get people who listen to us to get people who engage from the alumni side with us into a place to moderate our society so that we apply technology for better and trying to restrict the bad uses use cases. Hope you’re right. Thank you so much.

This was awesome. Thanks everybody for the many, many questions. We have so much more left. Sorry that we couldn’t get to all of them.

SK, Rachel, Dana, Sanata, Patricia, thank you so much. We will see whether we can put something out publicly. For now, have a good day and talk to you soon. Bye.

Thank you. Thank you.