Wizard Wednesday Episode 87 Transcription

Wizard Wednesday Episode 87 - May 17th

=

Elf: [00:00:00] It's Wednesday, May 17th. Welcome to our weekly cult meeting. Our magical midweek. Our Hermetic Hump day. Our Odin's Day, our Wednesday Adams, our Wizard Wednesday. Channeling in with quantum style through the quantum downs. This is elf.

Dotta: End, Doda,

Elf: and bear. What's up? How we doing? Good. Uh, I just wanna say thank you, slim for that musical request.

Uh, apparently the request line on Wizard Wednesday is now open. There's no, uh, system that guides it. You just gotta be lucky to get your request in. What,

Dotta: what

Elf: song was that? I loved it. It was called, it was called I Can't Party. Um, Dota is, is that something about Slim that we should know? Can he not party?

Dotta: It resonated with me actually. I was like, oh, I love the song. I can get, just cuz of the title. Yeah, exactly. No Slim. [00:01:00] It seems to me slim can.

Elf: Um, all right. We have a huge show ahead of us today. Um, I have a feeling we might go over time, which is great, uh, because I want to give, um, well, I mostly just wanna talk about story and the large lore machine and the local, and I also want to give acknowledgements and, and shout outs to cult content.

Um, and I wanna answer some cult questions, so let's see if we have time for all of that. Um, but first I just wanted, uh, announce something that I meant to announce last week, uh, and that is the winners of the Forgotten World's Minecraft Tower Building Contest by Merlin. Um, we saw a lot of great entries in that, but, uh, at the end of the day there were only three winners.

So, um, in third place. And actually third place was a [00:02:00] tie. Uh, it was mobilizes at Stone Tower and, uh, wander, wander wonders, wacky tree house. Uh, that was the tie for third place. Uh, second place was, uh, Magnus IARs Mushroom Tower, and in first place was Leo the Wizards. Great pyramid. Um, there were a lot of other great entries in this contest, but those are the winners.

Um, and, and, and again, I just, you know, I don't think we give enough love to, to Merlin's Minecraft. Uh, Forgotten worlds, uh, project, but damn, it's amazing. It's incredible. Um, no, we've actually been me jitsu, been talking with Merlin about what we can do and I think we're gonna, we're gonna kick up some fun soon to be announced.

Sort of weekly, um, weekly streams, uh, in Discord. So more, more on that soon. Great. Um, and I wanna give a Black [00:03:00] Sands update. Uh, so currently there is a race going on. It is called Tokas Mine Rush. Um, the winner of this race, we'll receive a one of one mount, and, uh, second and third place we'll receive a Black Sands mount.

Um, Furthermore, I need to announce that there are only 633 mounts left to be minted. So you can still mint one and there is a burning ritual right around the corner. Um, and I think that it has been confirmed as. Indisputable established fact that whoever holds the most mounts, uh, when they mint out, get to give Leus a haircut at the next Wizard, wizard gathering.

Um, so yeah, there, there's your incentive right there to, uh, to mince him. I wanna shave his head. I wanna his head. Yeah. [00:04:00] Well, Barry, you got, you gotta go min, you gotta go mint the most amounts. All right. There we go. Easy. I'm. My, uh, three boxes is, we're gonna roll, we're gonna play forgotten rooms, tcg, get some more points, uh, exclamation event.

Yeah. And, and I just wanna reiterate it is definitely confirmed that you get to give like a, just a haircut if you win. Um, okay. Um, Let's move on to a few cult questions. Um, let's see. Uh, okay, this is from Bridge. Um, it's a Dragon Ball Z question. I'm not, I don't know much about Dragon Ball Z. Do either of you guys know Dragon Ball Z?

Doda Bear? Yes. I, I think Doda had to run from the mic cause uh, there was just a quick little emergency that he is dealing with at home. Um, dragon Ball z I was never a massive Dragon Ball Z fan, like Pokemon [00:05:00] More. Um, but no, I don't know what that was. Okay. Well just let me pose this question to you anyway.

If Wizards, warriors and Souls had a a Dragon Ball Z power level range, what would it be? Well, I don't even know the SL Scale Power range. Gimme some numbers. Here's, here's Bridge. Bridge is the one asking this. I'm just gonna say over 9,000, obviously Bridge. Do you, I just, you have some extra context just for the record.

That was for a friend, like I'm, I don't know much about Dragon Ball Z either. Okay. Yeah, no, I was asking for a friend. Uh oh. Okay. Okay, okay. Yeah. Tell tell your friend over 9,000 at least. Cool. Cool, cool. Yeah, that makes sense. That makes sense. Like warrior warriors under, under 10, uh, wizards above 10 and, and souls are, you know, basically, you know, freeze a level mabu, like, you know.

No, no, I don't know anything about this. Sounds like you know more about D V Z than, than I, than me and Bear, but I'm just, uh, I'm asking chat g b T right now. Okay. [00:06:00] Okay. Um, okay. And then, uh, Snoop Doug was asking for a little bit of lore behind the X, Y, Z. Um, this is a, uh, uh, a location that is mentioned in the Souls collection, and I actually wrote a little piece in the Wikipedia about it.

Um, and so I will just say that the X, Y, Z is, uh, an immaterial dimension beyond the b boundaries of our physical reality. Um, but what makes it special as opposed to all the other Imma immaterial dimensions is that. In the X, Y, Z there's this, there's a specific type of magic which imbue the power of the spoken word.

Um, and so many wizards and alchemists and wordsmiths eagerly tap into the X, Y, Z to invigorate the effect of their verbal abilities. Um, abracadabra is one of the most common incantations used to invoke the power of the X, Y, [00:07:00] Z. Um, and that's pretty much all I wrote. Um, And, and we might get into more of this later, but I've actually been writing a lot recently in the Wikipedia, uh, about locations, about familiars, about, um, magical items.

Just, I, I've been having a lot of fun with it lately. Uh, but we can, we can get more into that later in the show. Um, Let's see. Uh, Tanya is asking, uh, what is your understanding of a wizard who doesn't have aro? Um, is there advantages, disadvantages compared to wizards with ROEs? Uh, I don't think there's any advantage or disadvantage.

Um, you know, I think to answer that question, uh, I, I think first we have to answer, well, what does it mean to actually have aroon? Um, and, you know, if your wizard has aroon, then, then that, I think that can be seen as either, uh, an additive, uh, factor or a limiting [00:08:00] factor to your wizard. Um, you know, if my wizard has a fire spell and I have a, a Saturn Roon, I imagine like, That Saturn ruin either limits or exerts great power to your fire spell.

Um, and, and then, so, so what does it mean if you don't have that? Well, you don't have either that advantage or disadvantage. Um, I would say it's really as simple as that from a lower perspective. There may be other like gaming mechanisms or, uh, on chain mechanisms that come into play, uh, with or without aro.

Uh, but that would be my, my, my brief answer. Um,

Zarris is asking about the forgotten ruins, R u I n S, um, and if they exist somewhere hidden in the universe. Um, there was a funny back and forth as Zarris and I had, uh, a lot of people, um, [00:09:00] Interchange the words ruins and ruins, um, often mistakenly. So, um, are there any forgotten ruins in the universe? I mean, I would say yeah, they're probably everywhere in the universe.

Um, which kind of speaks to the entire ethos of what forgotten is about. Um, When we see ruins in our world today, these are, these are places, these are memories of cultures that have long, uh, been swallowed by the sands of time. Um, and I would say these ruins are, Sort of their room that they put on the door so that we did not forget them.

Um, these are places of, of great cultural significance. We, uh, they're, they're part of our heritage. Um, and so yeah, I, I think the universe probably has, has places like this all over. Yeah.

Dotta: And honestly,

Elf: you get to see that in the univer game. There's a lot [00:10:00] of ruins scattered across the entire world that have been, you know, partially covered up and stuff.

So I think a lot of, a lot of that lore. Um, we'll come through when the game launches, which is cool. Some of my favorite set pieces are those really, for sure. Bridge, you got your hand up? Nah, that's cool. I can't wait to see that. Um, I was wondering if we're the forgotten ruins, you know, in relation to the timeline that the, you know, that you threw out quite a, a little while ago.

Aren't, aren't we technically the forgotten ruins of the universe? You, you could definitely look at it that way. Um, there, there is, there is a little bit of distortion in between the past and the future and the present. Uh, thanks to Ora Boros. Um, but I think these questions, uh, will definitely become clear as Magic Machine, um, puts out more and more storylines, which is actually a good pivot to the next thing that I wanna [00:11:00] talk about.

Um, so yeah, there, there was a lot of chatter in the, in the secret tower. I think it was last Friday, maybe. Um, just, just about the, the cult wanting magic machine to put out more story content. Um, really lively discussion. Uh, lot of great things came out of that. Um, I, I've got a lot to say about this. Uh, but Bert, do you wanna say anything?

Start off with anything or No, no, go ahead, please. Okay. So yeah. Yeah. I think like, like my first question was like, are they asking for just ency, like more encyclopedic information about the universe sort of akin to what you defined in the Wikipedia? Or do you actually want like, I. A more traditional storyline with a plot and some characters like to follow.

Um, and, and, and, and based on what I, what I saw in the Secret Tower, it looks like, it looks like you guys want the latter. You want like a plot based storyline with, with [00:12:00] characters that we can get invested in and we can follow. Um, and that's great. Um, you know, so far magic machine's answer to this has been, uh, we build the base IP.

Um, this base IP is based on the collective unconscious, uh, which is like sort of the aesthetic and narrative expression of the mono myth. And then you, since you own these characters, you create those storylines on top of that. Um, And then Magic Machine will take that creative output and put it into these larger media expressions, like the comic and the show.

Um, but the comic and the show take a while. I get that. And so you all want some stories right now? Dammit. Um, it, it's not enough for me to stand up here on Wizard Wednesday and talk about sort of the, uh, the ultimate narrative driver of the universe, which is, uh, this inevitable moment. Inevitable moment when the singularity radically transforms our existence.

Uh, and the phrase, I am my wizard. And [00:13:00] my wizard is me, is becomes real. Um, you know, I

Dotta: want some, I want some heroes

Elf: and villains, buddy. Yep. Yeah, exactly. Exactly. So, so, yeah. So, so yeah. So what we need to, what I think you guys want is just more of a more traditional storyline. Perhaps digestible in a weekly episodic format.

Um, so yeah. Great. Let's do it. The next hangup though is, uh, what is, uh, the format of this storytelling? Um, is it, is it video? Is it tweet threads? Is it post on Instagram? Is it something else? Um, you all sort of expressed an interest in video content. Um, obviously that's the format that takes the longest, uh, and the most labor intensive.

Um, so, you know, I, I think we'll definitely do some of that, but I think in, in the, in the immediate where we will, uh, go back to these, uh, story threads on Twitter. We used to do those, those were super fun. I loved [00:14:00] them. Uh, so, so yeah, I, I think we definitely wanna start getting back into that. Um, oh shit.

Did we lose bear? Yeah, he'll be back. Okay. Um, and so. You know, we, we, we've, I, I, I've, I have a lot of stories that I think I can tell in this format. Um, you know, we're, we're not lacking on content for storylines. It's real. It's really just about the format. Um, lately, uh, we have internally been talking about how to use AI as an assist in production here.

Um, that alone is a huge conversation that I think we're gonna get into later in the show. Um, I did see about a dozen comments emphasizing, uh, we don't want AI art, we want human made art. Um, and I promise you, I get [00:15:00] that. I totally understand. Um, And, and we can get into that in a little bit, but just to finish this story idea off, I, I think the next hangup is characters.

What characters do we put in these stories? Uh, do we use ones that Magic Machine owns? Uh, or do we put cult cont uh, cult owned characters, uh, in these storylines? And, and I think, like, I think now we've settled on, um, since Magic Machine is telling these stories, it'll be ma. Characters that Magic Machine owns, but they will interact with cult owned characters.

The model for this is the comic book. Um, and, and I think it's a pretty good model. Um, so, so, yeah. So all this is to say is, yeah, I think we're definitely gonna do more of this, more of these magic machine driven stories, uh, told on a weekly basis. Um, Okay, let me, let me, uh, speak to the AI component of this, [00:16:00] which will then lead into conversations about, uh, the Oracle and the l o m.

Um, so yeah, so about a dozen of you mentioned, uh, we don't want AI art, we want human, human made art. Um, I feel that, uh, the trend that I'm seeing that, that we're all seeing on Twitter is that, There's, there's a new AI model that comes out. Like for example, uh, the Wes Anderson model. You know, it comes out and it's awesome for a few days and, and we get to see fun things like, like the Lord of the Rings through the lens of Wes Anderson, and it's awesome.

But then after a few days, the shelf life wears off. And it a all you see is this oblivious robot making rote copies of a model that it was trained on. You can see the I artifacts and, and we all become Blade Runners looking at replicants. We all know ar ai art when we see it. [00:17:00] Um, and I, I think like, you know, we want to avoid that.

I think for me there, there's a lot of ways to, to do this and, and still maintain the magic of human created art. Um, First, I would say over the past two years, we have cultivated a community that has created original human artwork in the book of lore, on Twitter, on discord. Um, and, and I always want to optimize for that.

I don't wanna lose that. The AI tools are only as good as the data they are trained on. And that data, that lore, that artwork is made by humans. Now is this me saying don't use AI art to make things? No, absolutely not. Please don't interpret what I'm saying as that. I'm actually just giving the same message that we've always given from day one, which [00:18:00] is that we simply encourage you to make art with intent.

In earnest and yes, that can actually be done with AI tools just like it can be done with a camera, just like it can be done with a paintbrush. There are two shining examples on how to do this in our community. Right now, Matsui a k a Magis, Devin of the Quantum Downs and Magis was here. The one who rings are doing that.

How are they doing it? They are training their own models. And they are staying in control of the AI and not the other way around. They are using it as a powerful assistant to help in the iteration of creating that artwork. And there is actually a ton of skill and knowledge that goes into this. It's not easy.

I have respect for it. Um, you know, I, I myself have been. Going deep on stable diffusion and mid journey and like Woz has been holding my [00:19:00] hand in, in how to do this. Um, and it's, it's a really involved process. Um, and so, you know, as it relates to this like magic machine generated story content, um, I, I just, I wanna assure you that magic machine is not just gonna like type in a prompt.

Push the button done. There's the image to illustrate our story. We're not gonna do that. We are gonna use AI in the right way. We're gonna stay in control of it. We're gonna train our own models. Um, and we're just gonna use it in a very artistically intentional way. Um, And so, uh, yeah, I mean, you know, thi this kind of leads into the Oracle, which is just sort of taking Yeah,

Dotta: go ahead though.

I think on this is issue, I think on this issue, there's, uh, there, we, we've talked about this before, right? Which is there's always, when a new technology emerges, there's always this new question, which is like, where is the art? Right? In the same [00:20:00] way that, that you had. Uh, you know, we, we talk about this all the time, right?

Photographers and painters or whatever. And, and it's the same thing with ai, which is one, it's still very difficult to use. There's still a lot of art in it. Um, and it's, I, I. You hear a lot of people talk about, um, on the art side, a lot about AI created art versus human generated art. I don't, I don't think that's like the right dichotomy because like, not yet like Right.

Maybe the AI will become sentient and sort of create its own art and then we can have the debate for real. But right now I think that what you have is this AI assistance means like you take. You know, we did this as artists too, right? You go and you look and you're like, oh, I'm looking for, like, I'm gonna read all these books.

I'm gonna look for a plot line that I like. I'm gonna watch all these movies and try to find like the compositions that I like, or I'm gonna look at all these paintings and try to figure out the style that I like. And like when you use the ai, it's, it's, Doing the same thing, just like a lot [00:21:00] faster, right?

Yep. Instead of doing it this like organic by hand way, it's presenting to you 50 options just at the click of the button. And so, but at that point you still have to have your own taste. You still have to have your own discretion. And I think that like, That's really where you kind of get left with where the art is.

The art is in the discretion in terms of like, someone looked at this and filtered through the possibilities and they like navigated sort of the collective hallucination and they picked out this one thing and said it was good. And so I think that if I had kind of one limit of terms of like what's ever gonna end up in, let's say the book of lower, um, which is like the wellspring from which the world building comes from it.

It's this idea that like, A human has to say like, I looked at this and I think it's good, and. They approve cryptographically, uh, that which we, which human said that, right? So even in the case that if you have somebody who's sort of like, uh, you know, spams it or whatever, creates a [00:22:00] lot of garbage, we at least actually ha we can filter those out.

Because I think like the, the net incentives for the project are that people won't spam it. Um, and if someone does, we'll know who it was. And so for me, ultimately the art is in your discretion, right? That, that, that you have taste, that you chose something over something else, and then you present it to someone else.

As good.

Elf: Yeah, there's definitely an element of that. There's an element of curation as a part of creation. I, I don't deny that at all. I, I do though, as we navigate these AI waters here, here, here's what I don't want to happen. I, I, I've, I've been, I've been calling it like push button art, where you just type in a prompt, push the button done, and.

You can say like, yeah, but I chose the piece that I want to put on social media or append to the book of lore. Okay, sure. But we've got these AI models and [00:23:00] the mid journey model. Like I can tell any piece of art that comes out of the mid journey NJI model, and if. I, I don't wanna get into a scenario where 90% of our cult, even though they're choosing the picture they like from the NJI model, it's still all coming from the NJI model.

And now all of our content looks very homogenous and simply curating is not the same as the community we've cultivated for the past two years, where it's all like totally original human art. There is a difference. And, you know, I'm not saying we'll get there, but I'm just saying it's just something we should all be aware of and, and something we should try to navigate around.

Um, I

Dotta: think that like a lot of the problem,

Elf: let's say

Dotta: is, um, I is around like how we surface what was created and how we present it, and then like how we celebrate it, [00:24:00] right? Because. Um, imagine sort of a perfectly, uh, like a book of lore that has sort of the like per like a curation features, right? Where we can show the ones that we, that the community loves, that we can kind of promote.

And the, the, the art that isn't like as, as good sort of gets pushed to the bottom, almost like a feed, right? In that case it almost doesn't. Matter so much if someone is like prolific, but they're using sort of generic models. It's the same way of just if like someone is sort of like a poor writer. I think if we have like the right degree of filtering in terms of like what we ingest, then uh, it's not that big of a problem.

I don't know if that makes sense. Like, like I would agree with you. Like, I would say let's not encourage everyone just posting, you know, JI models, but at the same time, I think like, I would rather have someone do that than do nothing.

Elf: Okay. I between that and nothing. Okay, [00:25:00] let, let me just, Tabitha of the marsh says, well elf, not everyone in the cult are artists and still want to create something visually nice.

Here's my reply to that. Tabitha, you have been making amazing, beautiful, unique pixel art the entire time you've been in the cult. That those pieces are uniquely you. I can see, I can look at an image and I know that you made it. If you suddenly start substituting all of your visual output with something from a Nige model, I, I've lost your voice.

I don't know who made that anymore. I. And so, so my point is, is like putting your ruin on the door means putting your ruin on the door. Creating something visually nice, quote, quote unquote, is not better than creating something unique, even if that unique thing has a lack of skill.

Dotta: Yeah, I think that like, I'm very, very wary about drawing any sort of lines because of this [00:26:00] like belief that like art can be created in any medium.

Like you can figure out prompts in Nigy that are just like absolutely stunning and like, okay, sure, maybe I can recognize the model, but it like still brings this like beautiful worth and value to me. And so I, I, I think that, I actually think that the system is like, Sort of self-correcting, right? Because like you, you, you find something novel, you find the works that define it, and then it becomes a bit tired with the copycats and everyone sort of moves to something new.

And that's like almost the history of art. And so, um, I agree with your, I agree with your

Elf: broader point, but I don't wanna call, I mean, I actually think we're saying the same thing, like Yeah, people get tired of, of a model that gets repeated over and over. Tabitha, why would you use a nii model that's already been done literally millions of times when you have been speaking with your own original voice this entire time?

I am trying to optimize against [00:27:00] homogenization and wrote copying. Um, and that that's what you're gonna get if you use the exact same AI model as literally millions of people all over, all over the world at war. Yes. Um, but you know, all again, all this is to say is like, there's ways to do this tr start training your own models.

Be more iterative in your AI art process. You know, again, like I said, AI can, you can say very unique things with it, but just go beyond prompt, push button done. Um, anyway, um, So this, you know, this, this sort of is, is is like, speaks to this larger project that we've been working on. Uh, well it's, it's kind of hard to stay when it's, when it started.

I actually, I kind of think it might be interesting to talk about the origins of the Oracle, the Lolm. Um, I love that because, you know, we, we sort [00:28:00] of, I think we started, started this with the Gigis Chad Santa thing. Yeah. Um, That was like the first glimmer. And I remember when we did that, I, I think I even said like at that, that wizard Wednesday during Christmas, I was like, yeah, we wanna train this on the Book of Lord and sort of make an oracle, um, that you can ask any question about the universe, uh, too.

And um, and then I remember we, we did some internal tests and it wasn't like, Fully comprehensive, and it was speaking like a drunken cobal. And I think that's even why you have a COBOL speaking for it now. But like, I'm, I'm just gonna ask you Doto like, like what happened between now and then? Yeah, that was sort of a breakthrough.

Um, I think there's a couple of things.

Dotta: So. One is that like we kind of learned a bit better around how these, these tools actually, uh, work both as a team and maybe even as just sort of like a species. Um, [00:29:00] what I mean by that is that like exploring language models is something that like people, even the people that create them, uh, like uncovering what they're possible of or what they're capable of is something that's happening every day.

Um, so I think that's part of it. If I can maybe geek out on a little bit on the details, which would be, At Christmas time, uh, one of the things that we did is we took, I think it was G P T 3.0, um, at, so that would've been December. I'm not exactly sure the dates or what the version number is. Someone can catch me up if they know.

Um, and what we did is we tried to fine tune it. Uh, we tried to fine tune that model. I don't even think it was GPT three. Uh, but uh, we tried to fine tune, like text DaVinci on the, um, On the book of lore and the way that we did this was we generated the synthetic questions and then we had it sort of like complete the questions.

Um, Since kind of learned that like, fine tuning is okay for certain tasks, but what it's not good is for like teaching it to memorize, [00:30:00] uh, different facts. And so, um, one G P T got a lot better, right? There's G P T 3.5, and then we have now G P T four. So that was probably maybe the, the biggest jump. Um, I was of course very inspired by what, by what's been colloquially been called, like the Westworld paper.

Um, and also, um, the work of like, uh, baby agi, um, which are like some tweets that, that I can dig up. Um, and, and basically what they've been able to do is to turn chat g b T into this idea of an agent. And so you, uh, whatever may, I don't know how understanding this is, but essentially what we basically did is we, we figured out better how to learn.

How to, how to operate with G P T to have it asked questions and answer them, and then sort of do its own thinking on its own. So that was another big leap. Um, and then the third big leap was just understanding better the, the way that you provide knowledge to GT four. So, uh, we, and then we basically created these data sets.

Um, maybe [00:31:00] I, I don't remember what the time was a month ago. Um, those of which we shared, I actually super updated them yesterday and I need to push the most really, uh, recent versions to hugging face. Uh, but all that basically combined to just this like net output such that the, well, the responses that we're getting are, are quite good.

Very, very good.

Elf: Yep. So, yeah. Oh, go ahead. Well, it, it was, it was funny cuz like, You've been working on that MAP tool and Oh yeah. I, I think you, like, you must have trained it on the Wikipedia in the book of lore, right? Right.

Dotta: Yes. Okay. So, uh, yeah, I, I, I basically have this vision. Oh, so one of the things in the Westworld paper that was very interesting, so, uh, if you're not familiar, the Westworld paper was they basically trained 25 G B T agents.

They put them in a little R p G and they said, you know, you are, uh, you know this 30 year old woman in this community and you wanna plan a Valentine's party and, you know, it's, [00:32:00] it's 9:45 AM what do you do? Right? And, and so then these Asians run and they interact. One of the things I was like, very excited, uh, about that.

Was they actually structured the, the town as a tree, right? So you basically have like the, the town as a top level, and then you have the different areas, right. You know, it's Bob's house, Jane's house, the town hall, the, the grocery store. And then from there you might have, oh, in the grocery store you have like the veggie section, the produce, like the produce section, the box goods, the checkout, right?

And then they basically had a tree of all these locations and I realized, oh, we could actually use that to assist us in help building out the universe. So I spent, uh, some time building out these tools, uh, which I've been calling, uh, I, I don't actually remember what I've been calling the Universe Explorer.

Um, I've shared some early versions of that. It's not quite ready. One of the key things that I wanna be able to do there is actually edit the map. Um, but the idea is that you'll be able to sort of zoom in to an infinite level of resolution and be able to be given suggestions that are consistent and believable with [00:33:00] the lore, um, for certain locations in the universe to a ridiculous degree, right?

The, the, the universe is, is on par with the scale of our current world. And so it's, that's just too much for one person to create. So, yeah, so, so I've been working on, uh, some tools to generate those locations.

Elf: Okay. So, well, yeah, and, and, and so what was funny is like I've been watching you build that and then like Friday morning, like I saw all the chatter about people wanting magic machines to tell stories and I just casually dmd you and I was like, oh dude.

By the way, can you just give me a little dashboard to that I can ask any question about the universe with, and, and it'll tell me like, all the Book of Lord Data, it's trained on. And you did it in like 30 minutes. Yes, that's right. So yes, it

Dotta: was. Uh, three months of work to then build that UI in 30 minutes.

Not really

Elf: like three months full time, but yes. Yeah. So

Dotta: very quickly, we were once, as soon as you asked that, I was like, oh, that's [00:34:00] really fantastic. I'd already built question answering into this lore location generator. Why don't we just throw it into like a G B T bot? Uh, so yeah, so that afternoon, so I guess was just last Friday, I, um, Uh, wrote the local and I got it deployed and then passed it around to a few different people.

So it's, it's very, very good. One of the things that's super fun is it's, I've actually been able to like learn, uh, Uh, about a lot of stories about the book of lore, um, from its searches, right? So, you know, you can ask it about things that swim and it understands like the depths of the book of lore in a way that I don't think anyone else, uh, really does.

Maybe, maybe bam bam. But so, and then it also will provide citations. Um, so it'll actually link you straight to the book of law and you can kind of see, uh, what's there. So,

Elf: Uh, yeah, so that's, this has massive, massive implications for so [00:35:00] many things. For it has creative implications, financial income implications, social implic.

I can't even say that word anymore. Um, implications. Thanks Barry. Um, but yeah, I, yeah. So like, uh, creatively and you know, this is one way people always ask about how do we ensure continuity in the universe When you're building a world in a decentralized way, how does everybody know that they're building for the same world?

How do they keep their, uh, creativity? Uh, on a, uh, a, uh, a line of continuity. Um, how do we all know about, like cobalts for example, when we're all like sort of creating our own stories separately? This is one implication for the book, for the large, uh, lore machine, the local, you can go to it and you can just say, tell me about Cobalts, and now you've got a baseline upon which to write your COBAL stories.

It's [00:36:00] giving you sort of the amalgamation of everything that the entire community has written about cobalt. Now you've got a line of continuity upon which to work. Um,

Dotta: It's

Elf: created like it's, it's, it's, it's synthesized everything that's been created and put it into a very simple way to extract all the information.

It creates this like yeah, level playing field for everyone and, and, and, and that the result of that, I think EL is what you're also talking about moving forward. All the stories end up getting woven tighter and tighter because people start drawing from the same brain as opposed to just like finding a random piece of information, like shoving it in.

It's like perfectly

Dotta: categorized.

Elf: Yeah, like one of the things we always talk

Dotta: about is this idea of consistency, uh, in the, in the universe. Like how do we keep consistency? And, you know, I've even had people tell me, you know, even my family members, they'll be like, I wanna write my lore, but I don't really know what's been written about in this cer certain [00:37:00] area.

And, and, and is this like automatic synthesis tool where we can be like, okay, what do you know about how fireworks. Uh, work in the universe, right? And it sort of reads the, reads, the lore and kind of explores system there. Like we can basically have this better understanding of like what the magic system is, how we think about the characters, how they relate to one another, um, in these really

Elf: deep ways.

Yep. And then what else is super, super amazing about this? Like, this is one thing I've been obsessed with for the past, like five days now, which is. All of this lore can be tracked. It can be attributed to the original creator. This has massive economic implications. Okay, now imagine, let, let me just spell out like a loop for you.

You create a lore, let's call it a lower unit. Um, that lower [00:38:00] unit is then, uh, Put into the large lore machine. Now the large lore machine knows about this lore unit and who created it. Now content is generated from that lore unit, either the comic or a a fun Twitter thread or script on the TV show. All of this content can be created by humans.

It can be created by ai. It doesn't matter. It's being created from this lower unit. That you made that got put through the L L M and now content is being made with it. Okay? Now that content, either a video or a comic book or whatever, can generate some kind of sales revenue. It could be an N F T, it could be anything.

And now because that unit has tracked who made it, we can properly identify where the. A share of that revenue should [00:39:00] go to, which then incentivizes the creator to make more lore and the economic cycle continues. And so this, to me is just like a very crystallized version of binary star that we've been talking about.

But the large LO machine, uh, enables us to track this in a much more efficient way.

Dotta: Yeah. Yes, exactly. You'll be able to sort of like track, uh, well, like I said, people who contribute things. Uh, when you contribute to the book of lore, now we're able, one of the most important things about chat BT is that it's able to, um, both like combine and divide ideas in these really interesting ways.

It's able to sort of like, take the personalities that it knows, um, about Mag Devon and Mag Waer and put them in a conversation, uh, with each other and create something that's like somewhat [00:40:00] believable, right? And so, Um, yeah, I think that there'll be ways that we can kind of use, um, use this to create effectively merch, right?

I think that like, that's one of the like outputs of where money will come from, right? From being able to create stories that are, again, AI assisted, not AI generated, being able to create, um, I. Effectively merchandise that generates revenue based on characters lore. So I'm, I'm very, very excited to, to be able to,

to

Elf: do this.

Yeah. It's just, you know, be between new methods for maintaining continuity and ways to track. Uh, the creation of the entire community that, you know, stuff like this is what's gonna make the universe go exponential. Um, you know, I, I, I will never stop saying the most valuable thing that we are sitting on is our lore, our data, our creative output.

Um, this is something that [00:41:00] no other N f T project has none. Um, and so, you know, they can, they can try to build their large lore machine or something equivalent, but what are they gonna train it on? They don't have two years of book of lore to train it. Um, this is, this is what has made our cult so fricking valuable is all of your contributions.

I also think this is truly the collaborative, legendary, and with the large lore machine, it, we, it's just massively leveled it up.

Dotta: I ba I basically think that, like for me, um, one of the things I'm most excited about this is to start building tools with a bit more like affordance. What do I mean by that?

Affordance is this idea the, like that the UI. Is designed to help you do the thing you're trying to do, that there's more structure, um, the, the, the local. So I'm gonna try to get the Oracle out to everybody in the cult, um, as soon as possible. You know, I'd love to [00:42:00] do it this week if I can figure out a few, few of the things I'm working on.

Um, and, uh, it'll be beta, it'll be rough. It'll say wrong things, right? It's not gonna be perfect. So I'll, I, I, I, you know, be gentle with kind of your expectations in it. But, um, the, the, the local as it stands now is basically just a chat bot, right? It looks a lot like chat, g b t. You ask it questions, it comes back.

The first problem that I see with using the local is that people tend to think of it as like a search engine, right? That's the first thing people do. Like, yeah. Uh, you know, what do you know about the great old one Alini? Or like, you know, tell me about, you know, elf or something. Right? And it works for that.

It's pretty good for that. But where you start to get like really interesting, um, It, it could do so much more, right? Like it can act, um, maybe a little bit like a dungeon master. It can help give you outlines, it can give you insights into character. It can like suggest poems for you. It can like ex explain ideas for, um, [00:43:00] you know, items in the universe.

So there's like so much by way of creative content that it can produce. Um, but more than that, I think for me, I, I really wanna be able to build tools that are like, uh, that help me write stories. I have a really hard time with a blank sheet of paper. Like, you know, El I don't, you, you are able to kind of like, Sit down with a blank sheet of paper and just like make magic, right?

Both like visually and from story, right? You're, you're like top 1% of 1% of sort of creativity in that way and being able to like create believable, likable characters. For me, I have to work for it a lot more. For me, I like to kind of like have a bit of. Like a system or a template or a framework, which helps me kind of understand like the different pieces that I need to choose, like put the story together.

And so I see a lot of opportunity in the large floor machine for helping sort of like average or below average storytellers. Um, sort of find [00:44:00] their voice and be able to like, be more creative than they would've been be otherwise. And I think, I think that's what I'm looking for here, is not that I wanna like give them the ai our voices, but rather that you feel like it's not so daunting, right?

It's not so intimidating. You have, like someone in your court you can ask questions to. You have an editor, a coach, um, you know, a screenwriter, and you sort of act as the director. But you have this like staff on your side that helps you, you know, know how to fill all the details.

Elf: Yeah. Yeah, I get that. I mean, again, you know, I just, I just, you know, I just want people to like not get stuck in this sort of, uh, this, this process of just letting the AI create for them.

Like the, again, the last thing I would want is someday we wake up and 90% of the content in the book of lore is created by robots. Um, You know, maybe that's inevitable. I don't know. But all, all I know for [00:45:00] now is that, again, forgotten ruins, unlike any other community, has built this like beautiful sanctuary of pure creation.

And, you know, you know, I, I I, I, I gave the example the other day of like, like imagine a, a chair designer. This, this like indie chair designer, and he makes this really cool chair. And then IKEA sees it in a gallery and says, I love that chair. I wanna mass produce it, gimme the design. We will pump it through our factory.

And now 2 million people in, in, in suburbia can have that same chair. And it's like, you know, on the one hand, yeah, everybody gets to enjoy the, the chair, but on the on on the other, it's like, now just this. Homogenous gray ambience across suburbia in all America. And the chair has lost its value cuz the original creation is, is gone.

You know, we don't want to be the mass produced factory that repeats these [00:46:00] models over and over. We want to be a community of original chair designers. That is valuable, you know? Um, and I just, I just wanna make sure that. We can maintain that as, as much as possible because there's always gonna be places in the world where those original creators go to, and that's where the value is created.

Not copied, but created. Not curated, but created. And I, I just, you know, we need to maintain that as much as possible.

Dotta: I agree. I agree.

Elf: I, I also think it's not, it's, and I don't think you're saying this is like, it doesn't feel as binary to me. I think there's so many, there's such a gray area in between.

Like for someone like myself who can't draw, being able to tease out art or tease out a story is so helpful to me. Like, you know how you were talking about a Dota, so I, I agree. I think the human creation, um, is super important, but there's degrees to it. [00:47:00] Absolutely. I mean, if I was to run with that, that chair analogy, you know, there's gonna be people who take that IKEA chair and then they, they hack it and they make their own original creation.

And like, now it's suddenly like, oh, did you, you know, where, where's, where's the line between creation, cureation? Original design. Um, you know, of course there's, there's gonna be a gray area and it's impossible to talk about this in a black and white way. I, I, I recognize that, but you know, if, forget, well, and I,

Dotta: but I do think that it's right, which is like, you want to, we want to create a ground.

That's fertile for artists and a place where like artists can make a, a living creating artwork about forgotten roots. Like I think that's always been sort of one of the things that we've wanted, you know, we started this project before really like stable diffusion was good and our community, the economy of sort of like our community commissioning artists, um, was, was quite large.

Right. Okay. Maybe it was during a bull market or [00:48:00] whatever, but like, Forgotten Ruins is a place where if you were any artist, any artist at all, you could come and there would be someone who would, if you were willing to like draw Wizard Art, there was someone who's willing to pay you to make Wizard Art.

Um, and I like never wanna lose that. Right. Especially you like want people to do that hand creation. Yeah. Um, on the flip side, flip side, I don't know, another idea, which is like, one of the things that I saw. Uh, yesterday when I was doing that tweet storm or la over last week with, with, with, with, with, um, Zuki and Mag Ze, how they have their own custom models is I basically looked at some of my wizards that I haven't been commissioning art for, and I was like, oh.

Wow. I really, really want to commission art for these now cuz I wanna be able to train the model. It was like, I, I wish I had a list of like 20 artists that I could commission to kind of like do this artwork so that way I could travel the model and it's not lost on me that really what I was asking for is I needed 20 different pieces of artwork from human artists so that I [00:49:00] would like never have to ask a human artist again to, to to, uh, to create an image from this character and so on.

One, on the one hand, that's actually a little bit sad. Um,

Elf: well, you're like trying to train them out of a job, but what, what I would say is this speaks to the economic model that I mentioned earlier, which is if we can start attributing the, the data to the original creators through a large lore machine, now it's tracking all of that.

That creator attribution, and we can appropriately distribute any sort of revenue to an original creator that that's what this economic model, I think can offer. Um, as opposed to these sort of blind box models like Mid Journey and, and Open ai, you know, there, who knows what they're training on and they're not attributing their original creators, but if, if we at Magic Machine can, can actually build that into our L L M, that's game changing.

Dotta: Yep. [00:50:00] That's an important point that we will actually know the initial artists. I mean, I think it's really up to the, um, the wizard owner, right? Like our, our, let's say the business of the, let's talk about the business side. Like the business side of Forgotten Ruins is oriented towards, The token owner, meaning like when you own the token, you own the IP rights to that character.

Um, and so then if you kind of commission art, typically the art that you commission is work for hire, which means you sort of own the rights to that artwork because you own the, the rights to the likeness of the character through like a grant of magic machine. And so like I don't think that we actually have it structured such that we would pay commission artists, uh, back.

From like a merchandise, like let's say we make a doll of, uh, mags Devin of the quantum downs, right? Like it, you're right, that it would be feasible for us to track the original artist, but we actually haven't structured the business such that that's what would actually happen, right? Or [00:51:00] what actually happened is that the owner of the Wizard gets that money, which like, whatever, good for us.

Uh, but it's not lost on me. That kind of like, maybe there's a better way, uh, for like long-term incentivizing long-term art

Elf: creation. Yeah. I just, I, I wanna touch on something you mentioned last week about the sort of negative connotation around quote data mining. And I just wanna show you that is not what we're doing here.

The, the reason that that term has a negative connotation is because for the past 20 years, we've had these web two corporations who have. Data, mind you, a k a stolen your data and sold it back to you in the form of ad revenue and, and, and ad spends. Um, that is, in my opinion, pure exploitation and thievery.

And that is not what we're doing because. We're actually doing the exact opposite of that. Okay. This data, first of all, is totally decentralized. [00:52:00] You are free to use it in any way that you want, and you can monetize it in any way that you want. That alone is like, is like the opposite of what the web two companies are doing.

Uh, secondly, yeah. Oh, good. Well, just secondly, with us putting this data into the L L M. We are essentially making the mimetic components of your character that you own more accessible to the world, thus increasing that character's cultural reach, right? Yes. So, so in short, the more you, the more law you write about your character, the more the l l LLM knows about it and the more presence they have in the entire franchise.

That I

Dotta: think too is that like if you commission artists to make like a large lore machine, uh, which I'm including like stable diffusion models or whatever, like, like you have Mag Devin, where you have this unique character that's actually not going to cause. Less art to be created about [00:53:00] mags, Devon, it's actually gonna cause more art to be created about MAGA Devon, because it's so unique, so breakout like you just ha like you see this with existing franchises, right?

It's the breakout characters that have more work. Like how many pictures of Mickey Mouse are there out in the world? Um, because it's just so distinctive and so, Yeah. I, and, and I also think like, uh, I think about my own profession too, this idea that we're like putting the artists outta the work. I think about my own profession of just like as programming, like when you create open source software and you put it up on GitHub, I mean, effectively you're training the l the large language models, um, on how to write code that much better, right?

They get marginally better because they've now seen your human code. Well, that's the idea anyway, depending on your code and, uh, Well, so I, you know, as an eater says something here, she said, you know, it is a little sad that we're training artists out of a job, but it's possibly inevitable. And I think, I think that it [00:54:00] is inevitable that sort of the machines will assist in creating artwork that's already here.

Um, and I think that, like, hopefully the idea is that we can sort of harness that machine, uh, for like a more productive, decentralized.

Elf: Uh, like distribution of the economics. Exactly. Yeah. Yeah,

Dotta: yeah. That's a great way of putting it, like a more decentralized distribution of the economics rather than just harnessing it all in one corporation.

Elf: Yeah. Which, like, let's be real. This has been the mission from day one. It's just the l l m again for me, just offers a real, uh, sort of, um, Efficient web three way to do this that ha wasn't possible before. Um, that's right. Y you know, to me this like, here's what's exciting is the l l m is synthesizing the two most powerful, most age defining technologies [00:55:00] in the world right now, blockchain and ai.

Who else is synthesizing these two technologies like the L lamp? I, I don't know. Maybe there is, but I, I don't know of any. Um, and, and again, yeah, I mean, you know, I don't see this as, as a sad, like a, a depressing way to steal the creative efforts of artists. I see this as a way to more, uh, Equitably distribute the sort of economic potential

Dotta: of, of a community.

I also feel that like the more you are hands-on with these tools, A lot of the like anxiety goes away. I mean, there's still like the like ultimate anxiety that we might accidentally invent the singularity that kills us. All right? So like that doesn't go away. But these like intermediate anxieties around like, am I putting artists out of a job?

Am I stealing someone else's work? Am I, you know, and I think that, When you actually dig into the [00:56:00] tools, you realize it's actually very hard to use. Um, it's, it takes a long time. It takes a lot of like filtering, curation, creativity to get results that are good. Um, you have to put a lot of work into using, a lot of human work goes into making AI art.

Um, and I think that like even when you go into train your own models, How you capture the concept of MAGA Devin, um, is, is quite involved and a little unintuitive. And so like if I commission 20 artists to, to create a model of my wizard, now I can actually use that for like animations or for like comics or for like stories, right?

I can actually use, yes, maybe I'm like using a style inspired by a certain artist, but I can actually use that for my own creations down the line.

Elf: Yeah, I mean, Ima, we, you know, we've, we've been talking about a scenario where now that we've got the 3D models of every character, we can potentially train a model on every single cult member's [00:57:00] character because we have a full like rotation of that character.

Right? And, you know, imagine a world where you've got a highly specialized AI model built on your character's data. Combine that with the book of law entry. You wrote a full 3D turnaround. It's just limitless on, on what you can create. With that, I wanna ask, uh, Matt Zu, could you wanna come on

Dotta: stage? You're like the, you, you know, you and Woz are the experts in our community on this, and Woz doesn't talk.

So do you want to come up and let answer a couple questions? I'm gonna invite you up. I don't know if you're, if you're in a place where you can. Um, but like, one of the things I'd be curious about is just how many images does it take to like create the current model, um, of Mag

Elf: Devin? You know, it, it's funny, I actually asked Woz about that this morning and so far he only used about 20.

Um, right, right. [00:58:00] Yeah. Hi Meadow. Hi. My

Dotta: internet is

Elf: kind of spotty right now, so I hope I'm not gonna start breaking through. Um, but um, yeah, I think, um, I think I've been talking a little bit in secret Tower about this too, cuz um, I know, um, like leaders had a lot of, um, uh, leaders has concern about this too.

And I think it is something to be concerned about in terms of like thinking I agree, right? I think that, but I also don't think that it's not quite as black and white either, where it feels like, you know, like every, like every job is going to be cut or like, you know, um, artists are gonna be kind of left outta the loop or anything.

Um, I think that it is a very highly involved process, I think. It's very, for me, I feel almost as if it's like, like Photoshop, where it's like, it's a different tool, case use, and like a lot of artists can actually, would actually have like a more upper hand on using it. But also it provides tools for people who aren't particularly artists to, you know, like be able to like, like show off their creativity or like create some modifications and such that they otherwise wouldn't have been able to do.

Um, I don't think [00:59:00] it's gonna entirely take away any. Um, the ability for artists to be able to, you know, thrive either, I think that is just an entirely different medium. Um, and, you know, uh, the process of, I think for, for me, creating like the model that I have with Maggie Step, I think, um, my intent for it was never all, was like never any point, you know, like try to like grab from a specific artist.

I think I've done, um, I've been.