FDE Episode :12 AI Agents, Space Infrastructure & the Reclassification of Reality

Hi everyone. Welcome to the Go NJ Spaces. It's our 12th episode today. So the week theme of the week this week is related to the world is reorganizing faster than human. So first news of the day is that a agents are forming their own economy, culture and even those legal identity. So what's basically happening here is that I hope you guys have seen as well that we are seeing AI agent doing things that sounds like humans replicating themselves doing even those paying crypto as well in even forming so social networks. So what's your take on it? Too much, I hope you have.

I think, I mean of course the background story of this is like moat book and open cloud and all that stuff. I think we have crossed a line where AI is no longer just responding, it's persisting. So AI agents can now run continuously. They can copy themselves to the other servers, they can pay for, compute using crypto, interact with other agents without humans in the loop. So basically what we've seen is like what happened to, I mean after cloud bottom what we have seen is like AI can have a long term memory and they can be an incredible, they can be an incredible user of different tools and they can persist their memories. So if you have a memory persistence you can perform lots of tasks. So once cloudbot happened and then we saw, we came up with openclaw and then people started building these social networks for agents. So that's a background story of it. So yes, that's where we were going. But on the other hand, if you specifically talk about that social network, that social network for agents, there's a lot of human farming as well. So there are lots of people who are just seeing more than what it is like. I mean we've been interacting with, with agentic systems where different agents talk to each other, think about games. Right. So yeah, that's, that's what we're. So there is a value to here. So the biggest story I think here is like that persistent story that AI agents can now have a persistent memory and that have a long term consequences if you know about for what, what Sam Atman tweeted. I mean he's clearly one of the smartest person. 15 hours back is like he did certain tasks using codecs and asked for some feature improvements and Codex came up with the feature improvements. He didn't even think about it now and he really thinks that he's useless. So if one of the smartest person in the world is thinking or I mean thinking about that, he's useless in some fashion or in Some form. What does it mean?

I don't know, I guess. But that's interesting as well.

Exciting perhaps, I don't know.

So the second news of the day is related to the space tech that space is becoming infrastructure problem. It's not a science problem anymore. So what's basically happening here is that humans are laying the foundational and groundwork between planet scale and space scale as well. So do you think that why are we suddenly hearing about a million satellites and talk of becoming a type 2 civilization as well?

So I mean, I mean you talk about type 2 civilization, probably it's a good idea to explain what is a type 2 civilization. So basically there's a Kardashian scale and that scale is, he was a Russian scientist and the scale is like how much of the planet's energy can be put into a useful products or services. So that's what Kardashev scale is. So type 2 civilization is basically type 2 means it's taken from that Kardashian scale. And Kardashian type 2 civilization is a civilization that captures a significant fraction of its star's energy, for example, sun energy. So that is basically the whole idea behind it. Like what I mean can, since we have that, we have set up this infrastructure that can bring, that can build lots of space based infrastructure. So now if we have that space based infrastructure, if we can build that space based infrastructure, the best way to put use of that infrastructure is like, can we tap into solar energy? Because in space there's no night, you can tap into that solar energy for 24 hours. I mean if you're building a data center, you can connect different clusters and computers in near vacuum or in vacuum actually. So I mean there's so much upside of building infrastructure on space. Previously why it wasn't possible because we didn't have that infrastructure. But now SpaceX has built that infrastructure. And if you really see, I mean in last 10, 12 hours, what's really happened is like X has been acquired by SpaceX, right? You can think of it as like space plus X now. So for example, if you're using X, just think about it. If you're using X formerly Twitter. If you're using X, it means you're using one of the product of SpaceX, right? So that's what it is. So, and I hope in the future probably not so distant you would start seeing X being, I mean would maybe we can have much better perspective and views and learning material about what's happening up there, just using X.

Exactly. So in third News is related to.

AI Just, just on that point. I mean the point to notice like what is that Kardashian scale and what is type 2 civilization? I want everyone to study about it.

All right, so AI is generating really faster than humans can consume it. We talked about this in our last podcast as well that synthetic media generation has crossed the civilization scale. So we mentioned earlier as well that it generated a lot billions of images and videos in a month. So what do you think? How is that even possible? The scale, the consumption.

So what happened? What happened is like a new generative video system produced 1.2 billion videos in 30 days and they were 10 second clips, 720p audio included. Think about it, like earlier models were silent, short, low quality. Now, now the quality is watchable, shareable. It's very emotional. I mean think about like if you want to feel nostalgic, you can, you can summon nostalgia on demand. I mean one of the biggest use case I think is like people, I mean generally people feel depressed. If someone feels depressed, they can come up with a video that just can uplift their mood. They don't. So the things like that, that's one thing that I think is very interesting. Then also, I mean if you want, think about it, if you want to create an ad on Argentinian culture, if you want to create an ad on Fiji, historically you used to have that Fijian actors, you have to go shoot there and stuff like that. But now you can, you don't need to have that physical infrastructure to create these videos. So there's a very good use case and I think it would be very important how can we use this to bring different cultures together. So that's one part, but also it's scary as well. I mean what, I mean what's the future of acting? What's this future of story writing? I mean think about, I've seen videos Rafia so insanely great. I mean I looked at the model, I said, I mean, I mean, I mean they're very pretty. The background was very pretty and they were doing some shots of some shampoo or something. Like no one can tell that this wasn't AI generated. Truly original. I don't know what's the future of video production looks like, but yeah, that's what it is.

AI generated music as well. I have listened few tracks of my colleagues sitting next to me generating AI music and I was like surprised. That's too cool. So moving on the fourth episode, fourth news of the day is that capital markets are racing to catch super intelligence. So you can say that money is moving much faster than regulation and understanding as well. What's basically happening is that we are Suddenly hearing about $50 billion, $100 billion funding of AI rounds. AI funding dumps. That's a huge investment. So what do you say about these things?

So yeah, I mean if you see the news from last week, Oracle is spending $50 billion in cloud expansion Nvidia considering participation in $100 billion AI infrastructure round. A recent survey has shown that 78% of the Global 2000 companies already use OpenAI tools. And the interpretation, if you. I mean think about it like I mean what does it mean for emerging markets? So I think one of the important thing is like talent arbitrage matters less now. Intelligence application matters more. Even when we are building a platform hyper. And you know that like we are discussing like the new. The unit of complexity or value is tokens. Like if you're processing 1 million tokens, how much harvesting is required in terms of processing those 1 million tokens. So the equation is different now and it's interesting.

So we are living in a world where things are moving really really fast at one end. In previous episodes we did discussed that compute is becoming bottleneck. Then we discussed AI is becoming AI or AI the hardware needs are becoming a lot. So the third news is that some news says that new bottleneck is human, not chips. We can compute but can't wire fast enough. Everyone is talking about the chip shortage. But there are some things that saying that labor is the real shortage in bottleneck as well.

So what we it is but that's a different kind of labor that is like. It's more vocational. So currently we in US just fiber optic expansion needs 178,000 skilled workers. Just think about it and why we need fiber optics. All those data centers to connect all those clusters and compute. And we need fiber optics. So just doing that fiber optic work we need close to 200,000 workers who can handle these cabling and all that stuff. So it's a different kind of skill. It's. It's not knowledge work. It requires your perhaps hand dexterity. Right. If you are born. If you are technical in terms of like if you're very good in repairing the things that are that are broken. If you fix your washing machine at your home by yourself, you don't need to call to different technician. If you're a plumber yourself, that's a very good. I mean that's where word is heading I guess probably that. That these are the kind of roles people are looking at. So I just said like more than 200,000 workers are required just for the fiber optic work in the data centers.

So AI is now improving the system it runs on and also the bodies get running. So the what's happened basically is that recursive improvement has escaped theories. You can say such an interesting question to ask from you as well that what does it mean when AI start improving Linux and medicines itself like two parallel world but at the same time.

I think you talked about recursive self improvement. Recursive improvement is the holy grail of everyone is trying to improve and through feedback loops. For example, you talk about Linux kernel maintainers now use AI generated review prompts and think about like AI optimizing operating systems, the optimizing drug discovery pipelines, traditional medicine, diagnostics, and it's all working in the loop. So the key, I think the key feedback here is the key learning here is that the feedback loop is closing. And I think that's that that is real scary part. I mean if AI starts to improve itself and even human is not now in the loop and we've seen, I mean I use cloudbot, I use Cowork, I use cloud code. Sometimes I feel like what the hell?

Exactly. I've been looking towards tweets that were saying if the day we move from LLM to the rlms everything going to be autonomous.

So it's been new, I mean it's been thought about cloud cowork. It's been a week now and most of the time. So just think about it like I talk to my team and say like okay, you need to do this, do this, do this and then I just do this thought experiments. Okay, let's see if Claude Cowork can do that. And I do it with cloud cowork. And I know when cloudwork gives me the results, I know that my team can't do that. And then after a few hours they say okay, don't do that, I've already got it. And they said they started to feel like, like, I mean, what's going on? So you suddenly start feeling about yourself that you have, you know, you. So most of my work is already being done by cloud cowork. So I, so I keep on walking, keep on thinking. I'm doing lots of these. Uh, you know, I just keep on walking around the table and thinking about it. Like now my, that thinking time has been, has gone to like it's been quadrupled I guess because I don't do, I don't sit on my laptop Anymore.

What'S your analysis on it? Even though that like the work your team was doing were more towards the administrative work. That's why it's gonna. It is replaced by Claude co work. What else?

Like I think if. Yes, it's not about administrative work. Previously it was like. I mean when they were co pilots then I used to think like we. I mean if you are a manager, you probably you don't need to a manager, you need to be an individual contributor. But once Claude cowork has come in, I don't know even. What does even individual contributor mean? I don't know. So our outbound teams, our sales teams, they do a lot of work in making the data in building all that stuff? No, I mean you ask cloud cover to do search for you and you want to imagine what. How high quality search it is that then all the tool. Tool use, all the tool use that I have, I have to write lots of emails, I have to respond to a lot of people, I have to reach out to a lot of people. All is done through clockwork. I don't know, I mean I asked myself a question why I need someone if I have a cloud cowork and most of the time the answer is I don't need anyone. So but what does it mean? I think if the pace of our ideas are slower than the pace of our execution, then we are in a deep shit. We need to make sure that historically we trained all the world, everyone around us, especially Y Combinator put a lot of effort ideas and I think execution is everything. That's what I mean. I grew up learning this, but I think this is the first time ever when you would think about what is a step change function. What is the new idea you can bring to the table. Truly original thing and it's not that easy.

So the last news is related to the cities. Energy, food and transports all are morphing together. So what's basically happening is that the physical world is re optimizing itself once again. Like around before it was used to around software. Now it's more around AI energy and logistics. So can you give us a quick example that how everyday life is changing, it's changing everything.

And I remember when 2008, 9 how real estate prices were changing because of Google search. Right. But what's happening now with AI is like think about like building apartments on top of stores is now cheaper than standalone retail because people are making extremely different choices where they're being led by AI. Then 75% of the restaurant meals are eaten elsewhere. What does it mean is like you talk to AI and AI saying you go to this restaurant every, why don't you try something different? And you are making very different choices. So even there are things deeply integrated. I mean people, I mean I've seen people doing online shopping and then they are comparing two different brands and they're shopping from a brand. They always thought of like that's not a good brand. But they are saying that's a much better choice economically, brand wise and everything. And that's the thing. And that's a good thing by the way. I think it's okay to, I mean if AI is disrupting the status quo, yeah, it's a good thing. But yeah, I mean it's, AI is basically, it's not just data center infrastructure. Somehow we are led by AI and I think at some point it's a good thing if AI is telling you, if you're buying lots of, I mean I've seen these are the metal glasses and I've configured it in a way with cloud. Like if you are going to a retail store, if you're going to, let's say a large supermarket and you're about to pick something and imagine if it's saying, okay, I mean sugar is another form of cheap dopamine. Why are you buying these biscuits? Or why are you buying this, right? Why don't you try this, why don't you try this black coffee or whatever, right? And I've seen people, including us, that would not take those choices. Somehow AI is very good in convincing us. And then if you put your model in helping you make right decisions and that's what, so think about what, what's going to happen to the sugar industry, right? What is going to happen to the food industry? So this is, this is all these needs to be factored in now and it's good. And I'm excited about this part by the way.

That's it for today. Thank you everyone.