00:00Listen, the solution to the problems that we face is actually relatively simple, let me try to break it down for you. One: it will always be easier to make more profit if you are willing to do less moral things. There are nonprofit companies, there are public benefit corporations, there are all these mechanisms for you to do some good through, but it will be always more profitable, it will be always easier to function as an institution, always easier to grow within the context of the entropic landscape that we live within, if you solely focus on profit and you don't focus on morality.

00:46Well, what is morality, and how do we express it? We express morality and our perception about what is good and bad in relationship to actions that are taken through language and through art. Now we now have these new models, these large language models, that can make sense of the meaning held in language. So what you have is the potential to create a world where your word has weight, where your word actually has worth, um. The thing about democracy is it takes what is a market that is divided by, you know, whatever grift you can work.

01:35You know, like go back to 1776, the people who were most successful had often the most slaves. So it wasn't exactly like a, this is the meritocracy, this is the people with the best ideas. It was more, the people who could enslave the will of others to them to the greatest degree. And that still exists today in a different way, it's what we call wage slavery. It means that we have different people with different degrees of power, and that power is tokenized and divided, um, and represented by tokens.

02:24Those tokens are either things like money or they are things like stocks and bonds, they represent a portion of some greater whole that other people are willing to pay you for exchange of that particular item. Because we see that the undue power of, uh, some people, uh, sort of breaks this whole system, we've created this thing called democracy where everybody gets one vote, right? So everybody's voice is supposed to have equal weight within the context of democracy. But because of the limits of technology of the time, it was reduced down to this binary singular thing.

03:04So you had basically two different types of tokens, you had money tokens, or stocks and bonds, like fractionalized ownership of a greater whole. And then you had this, um, vote. And the vote represented in each pooling, uh, time period, basically that we would create these annealing cycles, we called electoral cycles, where at a fixed period of time we'd come together and aggregate our collective perception, and we would give an equal weighting to all of the voices of the land. At the beginning that was really just landowning, um, males.

03:43I'm sure there was some other, I don't remember exactly the particular particular, uh, constraints of it, but I know that, you know, slavery is ingrained into the Constitution at the Three-Fifths Compromise, basically saying that if you owned people that therefore you somehow deserve to have your voice matter more, that that was a very strange aspect of a compromise that was baked into all of this, right. But instead we live in this world where we could have our words have weight, we could make it such that when a person expresses a truth to society, when they express something that they think matters, it can be timestamped and it can be recorded to a ledger.

04:25To start with, that we can just say, hey, we have some ledger, the first person to say this thing before anybody else, we can actually track that by just creating a timestamp and saving it to a blockchain. That's the first part, you know, that's one way that we can create a way to your word. Unfortunately that doesn't actually solve the problem, because what that's actually creating weight upon is the container, the bucket of the item.

05:02Meaning that if it's, if you're tracking predictions, right, not every stated prediction is going to be reacted to in the same way. So if your isolated pocket of the network, and you're going to the edges of knowledge, you're not part of the center, you're not talking to that many people, so if you share a particular prediction it's not going to be reacted to the same way. So there needs to be some way that between predictions, between expressions, like, no matter the container of what is being expressed.

05:39Whether it's somebody telling a story, whether it's somebody sharing a prediction, or whether somebody's just saying what the truth is, like, if they have a functional thing that they feel they need to express to the world, that there is an incredible value in re-binding the weight of our word. Now, not everybody's word has the same worth in every context, but a particular signal that we can use to establish the weight of one's word, or the importance of one word, that is very difficult to hack is prediction.

06:19And particularly one type of prediction. This is the the thing that I think is basically unhackable in the context of, we don't have these crazy quantum computers yet that can, you know, simulate all possible futures or whatever, but the way that it can function is that if you place the greatest value on predictions that are ahead of the curve. What I mean by that is the first person to say something, and particularly in a time or frame where the rest of the community disagrees, but ultimately as time moves forward, as time progresses, there is a shift in perception towards that voice.

07:00That original sourcing is really important, but what actually happens, because we have the social hierarchy, because there is no dynamic attribution of sources of information, we have these people, and I've been complaining about it online, these people who are information sponges, they go around talking to a lot of people, they absorb their ideas, and then they go out on these podcasts and they go out in these public spaces, and they share all these ideas but they share sort of lossy bastardized versions of these ideas.

07:35And they don't do accreditation, they don't point to where the ideas are coming from, and all it does is it gradually increases their power, but it doesn't magically shoot them to the top of the power hierarchy, they're still traveling, and it's going to take years more for them to get there. So they just keep kind of doing this game where they don't really honor the source or the weight or the value of the original words from which it was sourced.

08:09Basically it becomes, oh this person has shared with me freely an idea, they've shared with me a concept, and because of that it's mine now, as if, as if when you converse it's a transaction where somebody is giving somebody a token and it fully, uh, is theirs to reflect, as if it is coming internally, right. So I started all this by saying it's very simple how to address all this, and it actually is. It's that one, you have to let people express themselves via language, express their truth through the language, we now have these new models that are emergent.

08:51Two, there needs to be a binding to prediction, not that hard, we're not even in that conversation yet. And three, that the forming of the voice of the generative AI needs to, at every step, reveal the sources from which it is being drawn from. That is dynamic sourcing, right? So that if there is a conversation going on, and the model is pulling in thoughts for the community to discuss, and it's being filtered through its lens, it's kind of hiding, basically, uh, while the conversation's happening, well, the debate is happening, who it's pulling in to reference while people vote upon their resonance with the output of the Iris.

09:29And because of that you get sort of an honest reaction from the community. And as you move forward, it can then reveal, oh, you know, I was pulling from all these voices, and we can trace back all the way through this debate the very origin of each meme that is mattering, each voice that is contributing to the progression of the emergent goal within the community or the group. So, like, in the world that we live in, which is large language models are going viral, it's not that hard to manifest, uh, a world where your word has weight, right.

10:12But we have to actually be discussing about it in that way. We have to accept first, basically, that profit as a mechanism will always sort of fight against morality, and therefore you need to have a secondary signal like Prophet that says, oh hold on, you can't take that action, or you shouldn't take that action, because there is some probability that it will affect social well-being in the future in a negative way. And you have to be able to dynamically assess the collective's opinion about that without having to pull like specific votes over and over and over.

10:50When attention is not just going to be there to react to every particular claim being evaluated. So we're very close to being able to solve this problem, where you have a decentralized distributed record of collective belief over time, essentially a decentralized neocortex. That is something that I've been talking about since 2017.

11:23Very interesting to suddenly see the people that I am privately sharing ideas with publicly express things that are very clearly tied to things that I've been expressing to them privately. And so I think that I feel very motivated in this particular moment to talk about this notion of a dynamic sourcing, of using these large language models to capture perception of how things are going to change in the future, right. So I can say particularly that perhaps the people who I think who are taking these ideas and misunderstanding them and spreading them outwards in a way that is bound to their name, I could say, well, the particular misunderstanding, the particular loss of information in their recitation of the idea, because it's not a perfect recitation, it's a lossy recitation, meaning they collected what was meaningful to them and they threw out what didn't make sense, right.

12:00But I can say if you approach, um, these problems through a means that is still really binding a lot of functional weight to tokenization, discrete tokenization of resources, and not upon the much harder thing that is to do, which is to place weight upon our word itself, you're not going to get where you want to go.

12:39You have to, you have to, you have to come back to this: how do you make a person's word have weight? Think about poor people, please, is this that hard? They don't have anything, right? So if you tie it to money, if you tie it to their ability to stake token behind it itself, it just will not function. Okay, so you have to make it so that every person, no matter how little they have in life, that if they say truth, if they speak the truth when nobody wants to hear it, that it has weight, that has worth, it has value.

13:19And until you solve that particular problem of providing weight and value to the complex resolution of each person's lived experience, then you're still going to face all the problems that we see in markets and in democracy and the world we live in, and in particular networked civilization, where we have an internet that is transmitting our thoughts from person to person through these UIs that are mediated by corporations that function for profit, and they get their profit basically by manipulating your attention through advertisement.

13:55This whole thing, I know this sounds more complex than simple, but in my mind capturing people's perception of the future and feeding it to an LLM now doesn't seem that complex to me. And there are extra things on top of that, including like decentralized data storage to keep it secure, but that's already been done. Like, most of what I'm saying, almost all of what I've already been saying, has already been done, so it's just pulling the pieces together into one functional whole, and it's just waiting for people to draw those, um, parts together and make the connection and form the coherent whole that I've been trying to express very continuously and very consistently for the last, however, five, six years now that I've been just trying to say the same thing.

14:25So thank you for listening, and I'll see you in the future. [Music]