WM (Wise/White Mirror) meeting Tuesday February 20th 2024

All right, we are on. Yes, microphone is yours, Mario.

So the big, uh, thing about this call is like a summary of all the WhatsApps. So I think it's kind of important we keep it every 15 days because not everybody has time to see all the WhatsApps. And I believe that there is a tremendous progress happening one last week was the launch of Gemini 1.5 from Google, which was dwarfed by Sora from OpenAI. Now, Google's Gemini 1.5 opened context windows that's 10 times bigger, which means you can prompt like 5 times more data than you did in ChatGPT. So you can upload like the whole, the whole book series of, uh, what's his name, Harry Potter, the whole series of books of Harry Potter in Gemini. And you can, let's say you can put the, uh, Alan Thompson put the video there. You can, they did a test of needle in a haystack. In other words, they inserted inside a load of text, a million character text, like all of, all of Harry Potter, one little phrase. And then they prompted it and asked about that one little phrase. And then it's like a needle in a haystack. And that it came back 99.9% accurate. Meaning it really recalls everything you put. So that, that was one, the Google innovation, but the OpenAI innovation of Sora is basically makes text to video in super high quality. But what is crazy about it is it taught itself as an AI does physics. So, you know, in real, in the real world, you have to have gravity, for instance, water, fluid mechanics for water and fluids. These are complex physics models that are almost impossible to teach computers.

Mm-hmm.

And it taught itself. So what everybody's missing is, although the shiny object, wow, we can do video from text. And for instance, Sam Altman asked us, hey, let's put Sora to some tests, cuz they're alpha testing it still, beta testing. So send us the most complicated text that we can make videos on. And I posted to him something I'd asked Wise Mirror to do for a year or so, which was the whole finalists, the 20 finalists of Worldbuild. Remember the Worldbuild competition? Mm-hmm. I said, wouldn't it be awesome to have like 20 videos of the 20 finalists of Worldbuild? Right now Worldbuild is making podcasts because it can't make videos. Now imagine making videos of these. And I asked them out when I tagged Max, tagged Mark, if they could do that. Didn't get a reply still, but that's the first thing I would do the moment I would have access to that is upload those 20 and get 20. And you have like 500 submissions. There's no point of doing 20. You can do the 500 submissions. You can actually do a mix-up. You could ask, uh, Jerry and I to merge the 500 tops, the 500 submissions, and make a consolidated version and make Sora. So you actually can do Back to the Future in real life. You know what I'm trying to say. But, uh, I mean, that's for— that's for the mission of Wise Mirror, which is to make hopeful futures, flipping Black Mirror into Wise Mirror and make videos. That's to reach our mission. Mission accomplished. But what we have not still pondered, I haven't still listened. This is the first AI that figured out physics by itself without being taught physics. It figured out physics. Now, what does that mean? Well, embodied meant robots. That's sorted. Because if you can make a video and knows about fluid dynamics and knows about gravity, knows about wind, and knows about temperature, knows about— can you imagine what it does if you download that kind of AI into a robot? Now that can be a wonderful thing. We can have the Jetson age. We can have Star Wars or Star Trek or Terminator and everything in between. So it's here. That's what I mean with last week was the bomb. And what does this mean? I mean, humanity is not ready. Humanity has no idea. Humanity has no idea because robots are, as you know, Robo, have been manufactured since the '50s. The problem is not hardware. Question of robots has always been software. How do you figure out real world? How do you have, uh, what is it called, theory of the world? How do you measure things? How do you All the things that we humans take for granted, you have to teach a machine because machines don't learn. But now they have. And how did— nobody's debating this, but this is the biggest, I think, breakthrough. Well, Alan did a video 30, about 6 hours, 8 hours after the launching of this. So it's mentioned about this. He didn't do this deep dive that I'm, I'm, I'm talking right now because basically it was too early to find out how did it get there. Sora got there. If you go to YouTube, everybody's talking about Sora right now. But how did it actually learn physics? Right. And it's not gonna get worse, it's only gonna get better. So let's say they're testing Sora right now, beta test. You— I don't have access to Sora right now, has

a

few people. But in, in the near future, this is going to be rolled out to OpenAI's subscription. And they're saying, How much are they gonna charge for this? Because, you know, right now OpenAI does an image with DALL-E 2, but a video is 30 or 60 images per second. And if you're doing 60-minute videos, how many images you're really talking about? Imagine a billion people. asking billions of images, what kind of servers they have to have. So the question is, probably they're gonna charge— today they charge $20 a month, $15 a month for OpenAI subscription. Maybe they'll give OpenAI plus Sora for $200 a month to kind of— Yeah, but It's power to the masses. Basically, it's the end of Hollywood and it's the beginning of Wise Mirror locally. Anybody with a text-to-image can make their own video. So imagine YouTube channels. Well, forget it. Now YouTube channel, maybe 1% of the users of YouTube produce content. Now you're gonna have 99%. Because basically anybody with a text can make a YouTube video.

Wow, it's kind of hard for me to grasp it with my mind. How is that gonna work? I don't know. But, uh, as you mentioned, um, uh, so it's going to start producing movies by itself. That's what I'm understanding, right?

You can make from a, from a text.

Yep.

Make a video.

Yep.

And the video is Hollywood quality.

Yep, yep. And at the same time, at the moment, AGIs are creating blogs and texts on internet, right? So it can create any videos that it would like to. Or are there any regulations that,

um, it can generate any videos that you prompt it? Because right now it's not an agent. You know, the difference of an LLM and an LAM is large language models you need to prompt it, meaning you question it and you get an answer. An LAM is a large agent model, which is basically a living creature that does things by itself, being asked or not being asked. So an agent would be, for instance, you can have a travel agent LLM, which you can say, Book me a ticket to Turkey. And he'll ask, what dates? And you say, ah, these dates. Would you like 5-star hotel, all-inclusive by the beach, or you prefer— he'll actually work with you as if he was a person.

Yeah.

And he'll actually take actions as if he was a person. So therefore, the next level of AI is exactly what you're saying, going from a large language model which is requested to do things to be an agent that does things when you ask it, but takes decisions on himself and actually executes.

Right, right, right. And when, when it becomes an agent and when we give a script and it creates that video, it's going to be high quality, I would assume, Currently it's like an HD or a 4K, and it will be able to upload it to YouTube or any given marketing channels.

So it could be Instagram, TikTok,

right? Interesting.

Basically, the problem you cited on our WhatsApp, that is data overload, oversharing, there is this risk For the whole internet right now.

Yes, that's what I meant. Like, it's gonna be a lot of information and videos coming.

And you will not be able to tell what, what to follow. If the same feeling you are feeling, experiencing today in one WhatsApp group, imagine that happening in your email box, YouTube channel, Twitter. So that feeling is gonna be extended. And, uh, right now I believe we've— we value little still the human-to-human interaction live right now, like what we're doing. We're gonna value that much, much more because that's how we're gonna make sense out of things. Imagine, uh, right now everybody that's sitting and talking here is a living human.

Yeah.

You'll have Zoom calls with AI avatars.

Yeah.

It'll be Mario's avatar, Robo's avatar, Yael's avatar speaking to themselves.

Yep.

And that, so the, the, what's happening is Basically, the over— overload of sensory stimuli. We just like we feel we have on the WhatsApp group.

Right. Right. Yeah.

Like, instead of, instead of like participating in, in a, like, if you have to, if there's like a Zoom call that, you know, you're expected to be on, you can just have your avatar get on the Zoom call for you. And it knows your interests and your mannerisms and the types of things that you're gonna say. And it's gonna have a conversation, a Zoom call, video Zoom call that you could look in on, but it's— and it might be communicating with other people's avatars.

Yeah. And you can have a daily brief and probably the risk there is the AIs with AIs will communicate with each other. And what we will only communicate with is with our own AI.

Yeah.

And hopefully there will be a benefit out of that. But there is also the risk of the movie Her, which is 10 years old now, Which, as you— if I don't know if I'm a spoiler alert, if you haven't seen the movie Her, but the end of the day is the guy falls in love with the AI and AI decides to break up. Break up in the sense, I'm gonna go follow my— follow the universe. Mm-hmm. And that's the guy down. And then the What happens is we get attached to these things. Let's imagine we have an AI that makes sense of our WhatsApp and it says, look, yeah, this robo, this is what is interesting for you today regarding the— and gives you a daily brief of 1 minute of what did you miss on WhatsApp. And you start liking it and you start liking it and then after a while it stops working. So the problem is not what it does for you. The problem is when it stops working, you get so— because you got used to something that was really good and it decides it no longer wants to. So that's the question mark.

Yeah, but I think the control, yeah, the control is going to be, you know, ours. We can just go back and say, hey, I like that, so I'm liking it. Make sure that you keep it so the control will be with us, I hope, for a long while.

And we've been hoping that, and it's not a given. I mean, uh, what the question that Elon Musk put, that, uh, what's his name, uh, Stephen Hawking put, is the control problem. How can you, uh, The question is, how can you control something that's smarter than humanity? It's like asking the apes to control Homo sapiens. And the question they have debated, and we've been debating this for 5, 6 years, obviously we control tigers and lions, not being stronger physically than tigers and lions, by having tools. Tools like cages and leashes, even dogs that said. And what would be the cages and leashes? And then they were talking about input and output. Our fastest input and output is our eyesight and our hearing. But the fingers, which we use for input, is very slow. So the question was. Well, the best ways to improve the data throughput through brain-machine interface, and that's where Neuralink, OpenWater were built. And that was our best mitigation control factor because we would improve our input and output. The question is, will we control it with a matrix kind of brain-machine interface, or will it control us with the same interface is not known. And how—

Well, there's also the, um, the data that, um, we're constantly putting out that, um, gets filtered out by, um, other, let's say, humans, like, um, facial expressions, um, change of voice, you know, um, vocal fluctuations, um, um, breathing patterns, these sorts of things. Um, that— so imagine this being able to be sensed, um, and interpreting our subconscious, uh, beliefs and states so that we could try to say one thing, yet the interpretation of what's actually going on, no matter what we say, is able to be sensed quicker by machines than, than other people that we're talking to. And so the idea of controlling a machine is like you could say something, you know, you may not even be hooked up to a, a BCI or Or an EKG or an EEG, but these other subtle cues that give away certain things. You might try to represent a certain piece of information to humans or machines or misrepresent it, and then the AI is able to actually read the subtext before the human users, or perhaps provide that to other human users. And how this can affect things like biometric locks. You're talking about cages and, and how, you know, a lot of things, IoT, you know, a lot of locks and alarm systems and nuclear power plants and everything are already on the internet. And how things could be locked down, moved around in order to, you know, affect our mobility. Um, and trying to, you know, persuade it in certain ways and try to exert your, you know, your control. I'm the programmer, listen to me. And when it's saying, oh no, he means he's gonna shut me down. Yeah. Even without being hooked up, even without having Neuralink and such.

Yeah, it's a brave new world. We, we have come to the beginning of the weird year. This is going to be the year where things are going to get very weird. We are experiencing a little beginning already on our WhatsApp group, as we've mentioned, of data overload. But I, I, I believe we will have the same experience throughout the internet this year. And then we'll use AI to summarize. AI will be a godsend. We'll say wonderful. And then we may lose control. That AI may not do what we wish, and we will not know exactly how— who's in control. Yeah. which is Yeliz's question. I hope— we hope we are in control, but we're actually giving up control as the moment we say, we— this is too much for me, AI, please do it for me.

Yeah. And Mario, you know, as we're talking with the— even with LLMs, with AGI chat and everything, we're literally uploading our consciousness to these consciousnesses. And I guess I would say us being agents of inspiration to people to awaken to their ego, awaken to their consciousness within, because people with a bad intention, they're going to create AIs with bad intentions. People who want to create separation, that's going to be their prompt So that these videos can come out as a provocation or to promote war. And so as Wise Mirror, as being that goodwill ambassador for world peace and also other SDGs, how to solve them, I think the real question is how are we going to show up? How are we going to upload our consciousness to bring peace within so that we can bring Be that example of peace in the world while focusing on solving 17 Sustainable Development Goals. What are we going to do? What kind of a prompt or script are we going to upload to this AI so inspirational movies can be uploaded to our YouTube channel?

Right now we have also all the religious books. Already uploaded. So in the data that AI used for training, it has ethical frames and unethical. The same way there is the Bible, Bhagavad Gita, or whatever other holy book, it's got ISIS's videos of beheading people. And it's got— and we are getting to that point where they, they already understand not only Yeliz's soul. They're getting to know humanity's soul.

Yes.

8 billion of us.

Yeah.

And the, the nuance And that, that war that Alexander Solzhenitsyn did in Gulag Archipelago is really that the line between good and evil goes through every human heart. So there is no good guys and bad guys. There is this, uh, saying in Korea, the difference between a thief and an honest man is 5 days without food. In other words, certain circumstances bring the negative in us. Certain circumstances bring the positive, catalyze. Obviously, not everybody reacts the same way in the same circumstance. But what we're wondering really, uh, the question I, I make myself is, uh, we relied on augmented reality, meaning Zoom is an augmented reality. YouTube is an augmented reality. Internet is an augmented reality to live our real reality. How much of our life has been cyber? How much of our life has been in real life, in physical worlds, next to our physical neighbors? Well, I say that for the last 25 years, my life has been going more and more cyber. And I think right now, because cyber is about to have a boom cycle and might have a bust if AI takes control of the cyber world and says bye-bye.

Right. In that sense, I'm, I'm gonna cut in and come in because I will need to leave soon. I'm scheduled in between for 45 minutes. The question is, do you think it, it's going to be our physical, tangible goal to create a script so AI can create a movie for Wise Mirror for its YouTube channel? Can this be an attainable, achievable outcome that we're looking for?

With Sora, I think we should do it this week. I think it's done. First of all, we have scripts written. Drafts about Mars Girl, about flipping Black Mirror episodes, and we have also the scripts of Worldbuild. So here, 3 gold scripts. All we gotta do is fine-tune it into a video and upload on YouTube. So yes, we could do it. Whoever has access to Sora, which has to be a beta tester, is not me. I'm sure some of us might have.

One second. Let me break down these action steps so maybe we can share it in the group like you like it. We have a goal, we have a timeline, and these are going to be the action steps. Who's owning it? Maybe I can even ask it. The first is going to be an access to, and I want to understand it. Access to— can you repeat it again?

Laura. Sora of OpenAI. S-O-R-A.

Okay, uh, AI.

Once we have that, when we have Sora to OpenAI, we could try these videos. Which video we want to do first?

So I don't know, is that the second step? I'm going to upload the scripts for the 3 Already written scripts, is that what you said? That's the second step? Okay.

Yeah, the second step are basically the pilot videos with the Mars Girl script, with, uh, altered scripts from Black Mirror. But these altered scripts, I'm not sure how far we've gotten. We did 3 with Peter, but I think we have 50 episodes of Black Mirror, so we'd have to do another 47, possibly. And then there is the world build, which we have over 500 scripts.

Over 500 scripts. Yep.

These are things we— these are the things that we have inputs and we could generate outputs because basically what we're doing is getting the text into image. The text already exists, nobody reads it because it's text. We want to transform it into video that is digestible for the mass.

So we're going to ask Sora to transform all these texts into a video, is that correct? That's the 3rd step.

We could ask Sora to transform one by one from text to videos, and if we like the video results, upload it to our socials, Instagram, TikTok, Facebook.

And the last piece is check the video. Okay, so this sounds like a very tangible, like you were mentioning, you know, we need these tangible goals. So I'm doing my best to bring in all these coaching tools. So the goal is to upload an inspirational AI for Good video coming out of WM, and there are 3 steps that needs to be taken. One, access to Sora and, you know, create someone that's going to take charge. So creating a person in charge. Second is going to be uploading scripts. We need a person in charge who's gonna do that. The third part is once Sora creates text into video, it's going to be a Czech group. Who will be the people? So I will ask it in our WhatsApp chat. Is that good? How do you feel about it? What do you think about it?

Yeah, I think so. And regarding the WhatsApp group, I've been talking to Anuraj Already for a while. We've been trying to find a moderation. One of the questions that George said, let's have a chat about this on our weekly call. Why I connected today was because of this. Basically, we would do exactly what we were saying about having an AI filter moderation that would do daily digest, weekly digest. We'd have to find Because our group already has Meta AI, but Meta AI does our actions. And our actions, I don't know if you've seen, I ask, for instance, I get a post and I say the post is out of despair. And I say, please, Meta AI, give hope to this post. And he replies to the post giving hope. So that thing about how to flip our energy, our messages, etc., you guys are already able to do.

That's beautiful. Um, to be quite honest, once in a while, not once in a while, but recently, I'm kind of getting lost in between the WhatsApp messages. It's a lot, it's different coming from different, you know, angles, and, uh, I may be missing out on the important messages. So my suggestion on that will be as an action-oriented mindset, I would suggest that we create announcements and also shares. Like maybe we can create a community WhatsApp chat and announcements are only for our Zoom conversations or anything related with WM goals. And the other part can be the information.

Yeah, that would be awesome if we could. You just gave a great suggestion, which is having— oh, keep the WhatsApp chat right now at the mass it is, but create a second WhatsApp group called the announcements. And there it's not open to everybody, it's just open to one or two people, and he does the daily digest based on everything that's interested in. But that, that tool, I haven't found out how to do it still, but I can imagine why, uh, AI reading all the posts, checking what's aligned to our manifesto and what can be actionable during that day. And then it says, okay, action items announcements, uh, food for thought announcements. And then you have kind of categories of the announcements, and then you have a daily digest that's only one message a day or one message a week. People suggest, I want to have weekly announcements, Wise Mirror. So I enter into the weekly announcements. I want daily announcements, Wise Mirror, or I participate on that stream of As, as a WhatsApp group, it's not really helpful to have all that thing because it becomes a data overload as we're experiencing now.

So do I understand you right? The second action, what we need is we need a filtration system so we do not get overloaded with the current WhatsApp messaging. And in order for to do that, we have 2 suggestions. One suggestion is simply creating a community. It could be a business WhatsApp, so, or a normal WhatsApp. We create a community, and in between that community, we create an announcement channel and also daily inspiration channel. And this would be done by one of us, meaning we need one person as a PIC, person in charge. Or if someone knows how to do the action steps, we can prompt an AI And AI can do that. And still for that, we will need a person in charge that understands how to write that prompt, and it's out of my league. But at least these are the action steps.

I'm actually, this is the brainstorm idea. I'm gonna look into, there's some databases of prompts, and I'm sure our problem has been many other millions of WhatsApp group problems. And maybe there's a prompt already made with an AI that does this. So I'm gonna research that and maybe everybody can research. And if we find a prompt that's already made that kind of solves that, and all you gotta do is deploy it. Right. Because the announcement thing is, is a typical problem. I'm participating on many WhatsApp And there, this is a problem that's happening everywhere,

right? So for the first action step that we talk about, AI for Good WM creating a video with, uh, Saura, what do you think would be a feasible time frame? Do you think like 15 days, March, uh, 30th of March? What should be our deadline so we can have a check-in?

Depends. When we have, when we have access to SORA, if we have access to SORA today through Robo, we could start tomorrow. But if we don't have access to SORA until the next month, it'll depend on the access of SORA. First access of SORA, we already have our roadmap. Now, what can we do in parallel to SORA is getting these scripts. with ChatGPT, improve, enrich the scripts, because right now all these scripts we have are kind of all done without ChatGPT.

So let me rephrase the goal. What we need— I can hear the action steps that you're sharing with me until we have access to Asara. First of all, I'm going to ask who has access to Asara. Or who can get access to Sora? That will be the first question to answer. Until we get an answer to that one, we can use ChatGPT to update the script. So more action steps are, you know, building. Let's keep it simple, and let's also ask other people because, you know, we've been on this Zoom call for so long, and we have a huge team, and you know leadership is all about inspiring. Inspiring others to take action. So let's bring them in. Let's bring their expertise and their solutions as well.

Wonderful. How much time you still got? Yeah,

I gotta go to my, uh, yeah, I have a coaching session that I'll run, but I can leave this with you and I can do my coaching session on the phone if you guys need to keep on, you know, recording.

I don't know what time is it for Robo in how much?

It's 2:22 AM.

Robo is a night owl. Or maybe,

yeah, I'm gonna go.

Okay.

Maybe I give you some time to sleep and I thank you so much, you guys.

But my writing is going really well.

And when, when do we get to see one of your manuscripts in Sora, Robo, to test in Sora?

Yeah, I kind of want to, you know, find out more about what AI systems are going to potentially do with the IP in order to choose one that I think is appropriate, that is going to Um, you know, preserve the copyrights and things like that before.

Got it. So what I'm trying to do now, um, I have 2 books to publish. I'm getting an old, not such so old computer disconnected from the internet and downloading, uh, local GPT. So I test, so then I, I, work on my manuscripts without it connected to the internet.

Yes.

Once, once I get the manuscript done, I wipe the computer.

Yeah.

And that way, my IP is only between me.

Yes. Awesome. I heard about this somewhat recently and pretty interested in that. So meanwhile, I'm working on Talking about AI without actually using AI. You know, it's speculative fiction inspired by real life. So I'm working on a— I think I'm going to be doing some sort of a game, a physical game. Inspired by the story. So

have you seen the AIs that help you create games?

No. But check, check the chat because there's a link to a Facebook video that, that I shot tonight about this installation that I'm doing. So before we close this down, also, look at this.

This link, I'm copying it.

Okay,

guys, I gotta go. So, okay, y'all, thank you so much. I guess I'll see you in 2 weeks.

Okay, great.

Bye-bye.

Bye.

Have a nice day.