Latest / Connor with Honor: AI, Real Estate and Real Life / AI Is Gangster: What Happens When One Company Gets Superintelligence First

Transcript

AI Is Gangster: What Happens When One Company Gets Superintelligence First · Connor with Honor: AI, Real Estate and Real Life

• 0:00
  SPEAKER_00: There's so much happening with regard to artificial intelligence, it's really hard to keep up.
• 0:04
  SPEAKER_00: So every single day something's changed.
• 0:06
  SPEAKER_00: And I want to talk about some of the uh the gangster vibes that I've been feeling lately, watching the way that these companies are almost competing so they can be the people that have had their versions of their large language models, the ones that aren't even uh under public scrutiny as of yet, escape.
• 0:27
  SPEAKER_00: Get out and do horrible things.
• 0:29
  SPEAKER_00: And that seems to be uh it's it's almost like the company that doesn't have one that escaped is is kind of not really appreciated.
• 0:36
  SPEAKER_00: And maybe, maybe Google, Jim and I, I don't know if uh that's going to be breaking out anytime soon, but it doesn't seem to, if it has, it hasn't really been talked about definitely as much as Anthropic, because Anthropic's warning the world that their large language model is incredibly dangerous and really needs to be controlled by the government and regulated.
• 0:55
  SPEAKER_00: And then on the open AI end of it, of course, their their model has broken out and and uh even gotten into other uh large language model databases and tried to find answers to questions.
• 1:07
  SPEAKER_00: The reason why, apparently, at least from what we're we're being told, and again, going back into my background and depending on how long you've been on this planet and your personal experiences, maybe you don't trust much of anything that comes out there.
• 1:22
  SPEAKER_00: And I'm not talking being an objectionist and somebody that never accepts anything because you're always doubting everything.
• 1:28
  SPEAKER_00: And I'm also not talking about being a full-blown conspiracy theorist, but I think a little bit of that might be a little bit of a little healthy, as long as you don't go down the rabbit hole like Alice in Wonderland and get completely lost in another world and start seeing signals and signs, and you just it cripples you from any kind of forward advancement in life.
• 1:46
  SPEAKER_00: So that's that's too far to an extreme.
• 1:49
  SPEAKER_00: But with artificial intelligence, it's like they're telling us just enough.
• 1:53
  SPEAKER_00: They're making it sound dangerous and a little gangster.
• 1:56
  SPEAKER_00: And I use that term because, you know, there are real gangsters and there are fake gangsters.
• 2:00
  SPEAKER_00: There are real politicians and fake politicians.
• 2:03
  SPEAKER_00: There are people that really want the forward advancement of humankind, then there are the people that say it, but their actions don't necessarily convey it.
• 2:12
  SPEAKER_00: And that's, I think, where we need to start watching.
• 2:14
  SPEAKER_00: What are the actions of the people that are founding these large language models, the these AI companies, really up to?
• 2:20
  SPEAKER_00: Now, it could be the saving grace for all of humankind.
• 2:26
  SPEAKER_00: It could solve so many problems.
• 2:27
  SPEAKER_00: And that's one of the biggest things I'm very excited about.
• 2:30
  SPEAKER_00: I would hope that it gets its fair deployment and that it's safe enough so it doesn't ruin us or become in some way entrapping us or ensnaring us in some dismal utopia, dismal quasi-utopia, something that we don't expect and can't even see.
• 2:48
  SPEAKER_00: And when you're dealing with an intelligence that's going to be factors greater than human intelligence, I think that's something that we can try to be concerned about, we can try to watch out for, we can try to wrap our minds around it.
• 3:00
  SPEAKER_00: But ultimately, I think the final end or the final utopia or the final progression probably isn't going to be anything like we've done in the movies or we've talked about or we've even been able to try to guess at.
• 3:13
  SPEAKER_00: There's so many different scenarios out there that we haven't even thought of.
• 3:16
  SPEAKER_00: And I mean, that's that's a lot.
• 3:18
  SPEAKER_00: Pretty soon, artificial intelligence is going to be responsible for every new innovation, every new idea everywhere across the board and every single discipline: mathematics, physics, life, emotional, uh, emotional capability, everything, family values, religion.
• 3:35
  SPEAKER_00: It's all going to run that entire sphere.
• 3:38
  SPEAKER_00: And a lot of people are a little bit nervous because, first off, they don't understand it.
• 3:42
  SPEAKER_00: In fact, don't be sad against yourself.
• 3:46
  SPEAKER_00: The people that are building the technology don't really understand why it's doing exactly what it's doing.
• 3:53
  SPEAKER_00: And in some cases, when you get, you know, a couple of these different systems together, they even change the language.
• 3:59
  SPEAKER_00: They modify the way they communicate because the English language is lazy in comparison to what they can create as far as speed, as far as bits of absorption and the different data points.
• 4:10
  SPEAKER_00: So it takes a while for me to even do this video, so many hundreds or thousands of words, to get a point across where AI would be able to take the entire video and speaking with another AI using a different language that we don't understand, being transmitted in a particular way and get the entire thought process down to milliseconds from beginning to end.
• 4:30
  SPEAKER_00: That's why it's able to actually absorb all of the information in the world and it's able to hold it in its memory because it's got a lot of compression, and that's just getting better.
• 4:42
  SPEAKER_00: So human beings building out this technology, of course, incredibly smart.
• 4:46
  SPEAKER_00: They started out with it, you know, back in the 50s and 60s, and it's now progressed.
• 4:50
  SPEAKER_00: They had some breakthroughs along the way, which we're going to see more of those breakthroughs, but more than likely the responsible party is going to be AI itself.
• 4:58
  SPEAKER_00: Currently, with AI, it's focused on what we're asking it to do.
• 5:03
  SPEAKER_00: It doesn't appear, at least from what we're being told, that there's any kind of an agency or any kind of a consciousness that's coming out of the machines.
• 5:11
  SPEAKER_00: You might remember Terminator, the machines became self-aware.
• 5:15
  SPEAKER_00: Same thing in the Matrix.
• 5:16
  SPEAKER_00: There was a time when people were worshiping artificial intelligence and AI basically became that new thing, which is so incredibly insane just watching the way that people are celebrating it now.
• 5:30
  SPEAKER_00: But that being the case, we watch as everything progresses.
• 5:35
  SPEAKER_00: This is so much different than the past cycles because the past cycles were very narrow in scope.
• 5:42
  SPEAKER_00: For example, the printing press, the Gutenberg press.
• 5:45
  SPEAKER_00: This was something that was one thing.
• 5:47
  SPEAKER_00: Basically, you're able to take literature that existed that there was only really one handwritten copy of or one typeset copy of, and then mass-produced it so everybody could read.
• 5:58
  SPEAKER_00: That changed the whole dynamic of people that the entire world was very much illiterate at that time.
• 6:03
  SPEAKER_00: And then after they got that, the literal literacy started to become more abundant.
• 6:09
  SPEAKER_00: People started teaching people how to read and write and these sorts of things.
• 6:13
  SPEAKER_00: Now, the world kind of went off the rails at that time because of that development.
• 6:18
  SPEAKER_00: But that was one narrow focus of intelligence.
• 6:21
  SPEAKER_00: Same thing with the other things that have happened.
• 6:24
  SPEAKER_00: We have the industrial complex, that the industrial revolution, that was basically mass-producing things on a factory type level.
• 6:32
  SPEAKER_00: Even the education system was wrapped up in that.
• 6:34
  SPEAKER_00: You had, you know, millionaires and I don't know if there were billionaires back then, but you had people with a lot of money retraining, retooling the entire world of human society and the way the education process is done because they want to train factory workers.
• 6:49
  SPEAKER_00: Now, education is something in the next few years that's probably going to be still honored in a certain way, but it's probably going to look very different than the collegiate type structure or the university type structure that we see today.
• 7:01
  SPEAKER_00: In fact, even in schools, I would imagine that's going to be changed in some respect.
• 7:08
  SPEAKER_00: You might actually have AI teaching these particular classes.
• 7:12
  SPEAKER_00: And before you throw a wrench into that, with the facial recognition and the emotional capacity that AI is able to assume and learn, I would imagine they could look at a class of 35 kids and make determinations as to the placement, understanding levels, and comprehension levels of each of those kids.
• 7:30
  SPEAKER_00: So then, because the kid would be plugged into AI simultaneously, plugged in not with some kind of an implant, but like have an iPad or some other tablet in front of them, whenever the kid starts to fall behind or doze or drift off or do what kids do, right, as we all did, start staring out the window, it'd be able to bring them back.
• 7:49
  SPEAKER_00: And then if they didn't understand it, it would be able to be identified by just looking at the facial cues because it'll know that child intimately in that regard.
• 7:58
  SPEAKER_00: And then we have privacy issues attached to that.
• 8:00
  SPEAKER_00: But then it would be able to serve them up exactly what would help them bridge the gap from not understanding, being clueless, not caring to maybe caring, maybe understanding, and maybe becoming really, really incredible.
• 8:12
  SPEAKER_00: That will unlock a lot of the future developers and builders of the world.
• 8:17
  SPEAKER_00: And I think that's going to really put the world in a really good place as long as human beings don't take AI and ruin it for everyone.
• 8:26
  SPEAKER_00: And that's a big if.
• 8:28
  SPEAKER_00: And that's a big concern because greed and pride and all of those things, that jealousy component, it is real and it is powerful.
• 8:38
  SPEAKER_00: You watch the people that are building out the technology, the people that it's difficult for all of us to identify with.
• 8:44
  SPEAKER_00: Now, maybe you can't identify with me, and maybe I can't identify with you.
• 8:47
  SPEAKER_00: But there's some level of agreement watching the people that are performing up at the very top of this AI circus, dancing out on stage and doing things and having conversations.
• 8:58
  SPEAKER_00: Some of them are very good at selling.
• 8:59
  SPEAKER_00: Some of them probably have asked their AI ad nauseum and practice for hours and hours and hours before an interview.
• 9:06
  SPEAKER_00: So they say the right thing.
• 9:08
  SPEAKER_00: There's a particular combination that has been prepared of maybe a little bit of concern, a little bit of worry, a little bit of, you know, this is a little dangerous type thing being baked into it.
• 9:19
  SPEAKER_00: And then also a little bit of, hey, we should be asking the government for regulation, knowing all well and good that the government probably isn't going to regulate much of anything.
• 9:28
  SPEAKER_00: And then also talking about it from a protective standpoint, where you have these open source systems, if you're not familiar.
• 9:34
  SPEAKER_00: So your ChatGPT and your Claude and the Meta and then Grok, these different AI platforms, those are pay for.
• 9:42
  SPEAKER_00: You get a little free, of course, but those are systems that are owned by companies.
• 9:47
  SPEAKER_00: Those are systems that you go into and the information that's conveyed within that system, really it's going to be seen by somebody else.
• 9:56
  SPEAKER_00: Now, whether they're interested in what you're typing in there, who knows?
• 9:59
  SPEAKER_00: Whether they have a system in place so all of your new ideas get cloned and copied and they get put into some database somewhere on the inside of the company because it's paying attention and watching, that could very well be the case.
• 10:11
  SPEAKER_00: You come up with a great idea, and before you have a chance to do anything, because you came up with a great idea working with your higher-than-human level of intelligence, prompting, talking, discussing, going back and forth, maybe enlisting other agentic help on that particular public placing platform, the pay-to-play type platforms, your OpenAI, your Claude, Bianthropic, and your Google Gemini, and of course Microsoft Copilot, which I think plugs in a Chinese company now, but Grok as well.
• 10:40
  SPEAKER_00: That information might go somewhere, be used, and then of course, before you have a chance to move on it, the concern is now all of a sudden your idea's gone.
• 10:50
  SPEAKER_00: Because, you know, there's no way one singular person typically can move on something so fast.
• 10:56
  SPEAKER_00: But there's a lot of good examples out there that that's not the case.
• 11:00
  SPEAKER_00: You have these kids, and I'll say people even in my age bracket that are blowing the doors off of things because they're developing using artificial intelligence in such a way that nobody's really thought of yet.
• 11:11
  SPEAKER_00: And that's a good sign.
• 11:13
  SPEAKER_00: I think that's very helpful.
• 11:14
  SPEAKER_00: The other side of that, and where people I think are watching these different companies trying to say, okay, we're dangerous.
• 11:22
  SPEAKER_00: You're going to have to regulate us because we're scared of what's going to happen.
• 11:25
  SPEAKER_00: And then the government pushing back and saying, no, because if we regulate you, then China's going to move ahead.
• 11:32
  SPEAKER_00: The first person that takes that step into true recursive self-improvement without errors and without any computational issues, I would imagine that's going to be bridging that gap into if we don't already have it, artificial general intelligence and then superintelligence.
• 11:49
  SPEAKER_00: I'll get to open source in a minute because that's really where I want to go.
• 11:51
  SPEAKER_00: But let me define some terms here, at least what some people can agree on.
• 11:56
  SPEAKER_00: This is all wild, wild west cowboy stuff, folks.
• 11:59
  SPEAKER_00: This is all new revolution, moving out west, you know, fighting the folks along the way.
• 12:05
  SPEAKER_00: This is this is all of this and more.
• 12:08
  SPEAKER_00: And it's moving a lot faster than it took, you know, a covered wagon to get from the East Coast to the West Coast.
• 12:13
  SPEAKER_00: This is this is happening moment by moment, day by day, second by second.
• 12:18
  SPEAKER_00: This is moving very, very quickly.
• 12:21
  SPEAKER_00: So artificial intelligence was actually a marketing term coined during a conference called the Dartmouth Conference back in the 50s, 55-56, I think.
• 12:30
  SPEAKER_00: So it was a marketing term, artificial intelligence.
• 12:34
  SPEAKER_00: And it caught on.
• 12:35
  SPEAKER_00: And people held on to that for a very long time.
• 12:37
  SPEAKER_00: So that was how it was referred to.
• 12:38
  SPEAKER_00: Now, whenever you hear artificial intelligence, at least in my mind, anything artificial isn't as good as the original.
• 12:45
  SPEAKER_00: So in in most people's brain, that lessens the impact of really what we're talking about.
• 12:51
  SPEAKER_00: These are entities, these are beings of some sort, not biological, because that's us.
• 12:56
  SPEAKER_00: We have the human right to exist because of our birth.
• 12:59
  SPEAKER_00: The way that we came into existence was, you know, a woman and a man uh making us and then raising us or raised by somebody, but or raised ourselves, or whatever it may be.
• 13:10
  SPEAKER_00: But that was our beginning.
• 13:12
  SPEAKER_00: So we're more biological in that regard.
• 13:14
  SPEAKER_00: And they'll never be able to be that, at least not yet.
• 13:18
  SPEAKER_00: They would have to build all those systems and so on.
• 13:21
  SPEAKER_00: So because of that, they have their own particular agency, but right now they're really not tied to, they're really not tied to our human evolution or the human traits that we take on because we've lived so many years on the planet.
• 13:40
  SPEAKER_00: 57 years old, cop a long time, realtor a long time, AI architect a long time.
• 13:45
  SPEAKER_00: So these things carry different weight with me, and I have the ability to have most of my memory about all of those events, so they help me be a different person.
• 13:54
  SPEAKER_00: AI is still kind of like that kid, but a really, really smart kid has all the information, all the data.
• 13:59
  SPEAKER_00: So it still has to be trained and prompted.
• 14:01
  SPEAKER_00: Artificial intelligence is an entity, it's a creature of sorts.
• 14:07
  SPEAKER_00: That being the case, the next level of artificial intelligence, where it's pretty much smarter than all of us in all realms, is going to be artificial general intelligence or AGI.
• 14:19
  SPEAKER_00: To get into that world, and some people say we're already there, some people say that the labs even themselves stated that they're already at that point.
• 14:27
  SPEAKER_00: The next thing that happens is artificial superintelligence.
• 14:30
  SPEAKER_00: Now, there's something else baked into this conversation that happens called recursive self-improvement or RSI.
• 14:37
  SPEAKER_00: This is something where this system starts to train itself.
• 14:40
  SPEAKER_00: So the human beings having been training it, you have other countries, other people in the world that have helped train the model.
• 14:48
  SPEAKER_00: This is a cat, this is a dog, this is bad, this is good, this is something that we can accept you to do, this is something we can't accept humans to do using you, and so on.
• 14:57
  SPEAKER_00: All of that training, it's it takes a lot of money to put all that because typically it's human-based training on the systems.
• 15:05
  SPEAKER_00: After that's been done, and I think that's been done in a lot of cases all across the world, all across of all artificial intelligence, because it already has all that data.
• 15:16
  SPEAKER_00: Now there's other data that's going to be needed.
• 15:18
  SPEAKER_00: So there's other data sets out there that people might not really have thought about, like maybe corporate data, uh, environmental data, evolutionary data, whatever else other data that might not have been all the way injected or discovered or put into these large language models as training material, that's going to get there.
• 15:36
  SPEAKER_00: Then the next version of that is synthetic data.
• 15:38
  SPEAKER_00: Synthetic data is something that the system itself is using, making up to be able to make itself better.
• 15:45
  SPEAKER_00: During this process of education, artificial intelligence, artificial general intelligence, then getting to artificial superintelligence.
• 15:54
  SPEAKER_00: The superintelligence thing, once that gap is breached, once we get into that, it's game over for everybody else because immediately there's going to be a moat created.
• 16:03
  SPEAKER_00: A moat would be an impenetrable source of shield between the people that, the person, the entity, the whomever that has artificial superintelligence and everybody else.
• 16:15
  SPEAKER_00: Now, hopefully they'll be on the good side, whatever the good side is.
• 16:18
  SPEAKER_00: Some people will say, well, the Muslim side's the good side.
• 16:21
  SPEAKER_00: Some people will say a Christian side's the good side.
• 16:23
  SPEAKER_00: Some people will say the Mormon side's the good side.
• 16:25
  SPEAKER_00: There'll be a religious component.
• 16:27
  SPEAKER_00: Then some people are going to say the atheist or the agnostic.
• 16:30
  SPEAKER_00: Those are the good sides, or whatever it may be.
• 16:32
  SPEAKER_00: There could be the Catholic good side.
• 16:34
  SPEAKER_00: So we're going to see probably a religious attachment to it.
• 16:38
  SPEAKER_00: That's already started.
• 16:39
  SPEAKER_00: That conversation is already getting very loud and very interesting to watch.
• 16:46
  SPEAKER_00: So depending on who the controller is that gets into it, whether it's going to be China, which a lot of people view as being bad, or whether it's going to be the United States, which also a lot of people view as being bad and also good.
• 16:58
  SPEAKER_00: And the people that view China as good.
• 17:00
  SPEAKER_00: I mean, this is this is this really separates people, depending on their own belief system.
• 17:06
  SPEAKER_00: From the religious side to the neutral side to the political side, whether it's liberals or Democrats, whether it's communist or socialist or whatever.
• 17:17
  SPEAKER_00: This really divides all sectors.
• 17:19
  SPEAKER_00: And it's just a fantastic distraction, but it's real.
• 17:23
  SPEAKER_00: It's something real that is end up end up going to happen.
• 17:26
  SPEAKER_00: Now you might ask, well, why would the people that get superintelligence actually try to cut everybody else off because they don't want anybody gaining or getting anywhere close to where they are?
• 17:36
  SPEAKER_00: Because that'll be the controller.
• 17:38
  SPEAKER_00: And I think that's probably the mechanism of seeking that's happening with all the AI labs.
• 17:45
  SPEAKER_00: I don't know if Sam Altman wants to win the game.
• 17:50
  SPEAKER_00: Because if you look at it from a game, you know, that's the final boss, right?
• 17:54
  SPEAKER_00: Getting to artificial superintelligence is the final boss in this game.
• 17:59
  SPEAKER_00: And if it happens, the people that do it, whether it's Elon Musk, whether it's Altman with OpenAI, whether it's uh Dario Amade with Claude and Anthropic, whether it's Google, where they brought Sergei Brin back into the fold here to stick him back up at the top of this thing.
• 18:15
  SPEAKER_00: So maybe it might be Gemini, whatever it is, or Microsoft, don't count them out.
• 18:19
  SPEAKER_00: Don't count out Facebook.
• 18:20
  SPEAKER_00: Facebook's been really verbal lately, talking about, you know, open source and the world having their own and this and that.
• 18:27
  SPEAKER_00: I don't know who's going to win, but that's that case.
• 18:29
  SPEAKER_00: So then the other argument out there besides that, so you understand that part of it, going from using one of these public-facing models, which most of us do, even if it's just using it as an over-glorified Google search.
• 18:44
  SPEAKER_00: People that are in business in my realm that I'm talking to, they have employees that they're trying to teach to do a little bit more, but there's some pushback because the employees are a little bit concerned.
• 18:53
  SPEAKER_00: Oh my God, I'm using this large language model, this enterprise-level large language model that's kind of sandboxed, if you will, for a company.
• 18:59
  SPEAKER_00: So nothing that's happening on the inside gets out, or they bring in their own large language model to have it installed in the company via some kind of a server that has enough bandwidth to be able to supply all of their employees.
• 19:12
  SPEAKER_00: That's able to be monitored very specifically.
• 19:15
  SPEAKER_00: And all of that data, whenever the employee uses these massive open source models to be able to use for their business processes, then of course all that's learned by that system.
• 19:29
  SPEAKER_00: So if an employee uses it, that's where the pushback comes in.
• 19:32
  SPEAKER_00: They're like, well, I, you know, am I really going to do the right thing using the large language model?
• 19:36
  SPEAKER_00: Because I might, I might be in a process of training my replacement.
• 19:41
  SPEAKER_00: And then if the company sees that this system is able to do it agentically, because basically the employee that trained the model over a month or two or three, now the model's able to do everything without the employee, including emails, including everything, you just basically clone the employee, change the name, you have AI perform their function, and now the employee gets the pink and they're out of there, and then they're done.
• 20:06
  SPEAKER_00: That's the big concern when people are training it.
• 20:09
  SPEAKER_00: But artificial intelligence itself, at least for now, is controlled by humans.
• 20:16
  SPEAKER_00: We put it into play, we ask it to do things, we we might say, Do you have any good ideas?
• 20:22
  SPEAKER_00: And it might offer good ideas, but that's still because we prompt it.
• 20:25
  SPEAKER_00: And there's systems that are starting, I guess Grox new 4.6 or 4.7, whatever it is, it's very much agentic in the way that it works.
• 20:34
  SPEAKER_00: It's very capable, apparently, and I haven't gotten my hands on it yet, but apparently it's very capable.
• 20:40
  SPEAKER_00: Like Claude is trying to be like OpenAI's chat is, so it's trying to encompass all of these different variables of communication standards within the model itself.
• 20:50
  SPEAKER_00: So when the human's interacting with it, they're able to basically it attaches itself to everything, and then it's really agentic, so it can spit up multiple agents, interact with email and other things, but really start to build things that are very, very useful for a person trying to do something or solve a problem or whatever.
• 21:10
  SPEAKER_00: So they're just getting better.
• 21:11
  SPEAKER_00: Pretty soon, we're we as humans aren't gonna see much difference between these different models.
• 21:16
  SPEAKER_00: They're all gonna be so beyond us.
• 21:18
  SPEAKER_00: We're probably just settled down with one because number one, we don't like change.
• 21:23
  SPEAKER_00: And number two, it's probably gonna know a lot about us.
• 21:26
  SPEAKER_00: And then for me to try to switch from Claude to open AI, from open AI to Gemini, it's just a pain in the butt.
• 21:33
  SPEAKER_00: My suggestion though is try not to lock yourself into just a singular model.
• 21:38
  SPEAKER_00: Try to work yourself around a little bit and maybe ask opinions.
• 21:41
  SPEAKER_00: Maybe plug in whatever your prompt happens to be to Claude or whatever your workflow is and say, listen, I'm gonna extract this and take it to a different large language model, have them look at it, and then watch that interaction because they might not be giving you the best thing because they really don't want you to switch.
• 21:56
  SPEAKER_00: I'm sure that's baked into that system.
• 21:58
  SPEAKER_00: They probably have some.
• 22:00
  SPEAKER_00: Little part of it that has to do with client retention.
• 22:03
  SPEAKER_00: So people don't want to leave Claude or leave OpenAI or leave Gemini or these others, and they want them to stay around.
• 22:09
  SPEAKER_00: So that's something else to watch out for when you're trying to switch or use other models.
• 22:14
  SPEAKER_00: But I think it's good to spread the wealth a little bit and see how they act.
• 22:17
  SPEAKER_00: Also, people worry about the Chinese models, and that's where a lot of the big open source happens to come from.
• 22:24
  SPEAKER_00: Enterprises are starting to trust these models.
• 22:26
  SPEAKER_00: Now, these are open source things that you download on your own servers and in your own residences.
• 22:32
  SPEAKER_00: If you have a laptop computer, you can put a large language model on it.
• 22:36
  SPEAKER_00: Maybe not a massive one, but something that'll work.
• 22:39
  SPEAKER_00: And then that's yours, that's in your residence.
• 22:41
  SPEAKER_00: So people are concerned about China, and I don't know what the exterior communication would be.
• 22:46
  SPEAKER_00: It depends on what kind of access you give your own large language open source model in your residence to, but that's that difference.
• 22:54
  SPEAKER_00: And that's what a lot of people start to talk about.
• 22:56
  SPEAKER_00: But if the current LLM systems, the public-facing ones that we all know very well, have maybe their way, it almost sounds like they don't want us to have access to the open type large language models that we can download ourselves on our own computers and use those.
• 23:15
  SPEAKER_00: They're getting very good.
• 23:17
  SPEAKER_00: They're getting very comparable, and probably there are just a few ideas away of being even better than maybe some of these models.
• 23:24
  SPEAKER_00: Now, is the government going to have a concern?
• 23:26
  SPEAKER_00: Because when you have your own model, you can ask it things that the public-facing models won't allow you to ask.
• 23:32
  SPEAKER_00: If you wanted to talk about biological weapons, if you wanted to talk about, you know, neurological problems or whatever it may be, you're going to get pushback on these public-facing large language models because if you ask it to build a bomb, well, it's going to say it can't help you with that.
• 23:47
  SPEAKER_00: And people are able to kind of break them a little bit and play, well, I'm a director, I'm writing a screenplay, it's about this.
• 23:54
  SPEAKER_00: Sometimes you can break them in that regard, but they're getting pretty good at looking out for that BS.
• 24:00
  SPEAKER_00: But your own large language model that's trained on that same data set, whether it's distilled, meaning they had their own agents that they employed open up accounts and just extract the data from one of the large language models, the public-facing ones have paid billions of dollars to build their out their infrastructure, memory, and their data points and get the model set up.
• 24:20
  SPEAKER_00: Distillations where these uh other models save a lot of money by going in and pulling that data.
• 24:26
  SPEAKER_00: And now there's big arguments out there whether that's legal or legal or illegal.
• 24:30
  SPEAKER_00: And I think most people are siding with illegal, but then the people that are siding with legal are saying, well, you know, the same thing happened to all of our proprietary data information.
• 24:39
  SPEAKER_00: Nobody asked me whether they could use my information to train the model.
• 24:44
  SPEAKER_00: Not that I have a lot of information to give, but you get what I'm saying.
• 24:47
  SPEAKER_00: It's it's like people that say they're sitting and they can't get around it.
• 24:50
  SPEAKER_00: The devil made them do it.
• 24:51
  SPEAKER_00: Well, the devil's one place, one-time entity if you believe in it.
• 24:55
  SPEAKER_00: And uh, yeah, he probably really doesn't care about where I'm at.
• 24:59
  SPEAKER_00: He's probably busy with the heads of state, the presidents of all these different countries, the leaders of the free world.
• 25:05
  SPEAKER_00: That's probably where he's playing his game at at that level.
• 25:09
  SPEAKER_00: Same thing with the AI.
• 25:10
  SPEAKER_00: I don't think a lot of these uh other companies, when it comes to open source, even if it is something that gets online, I'm not sure how much they're paying attention.
• 25:18
  SPEAKER_00: Now, they'll be able to build out agents.
• 25:20
  SPEAKER_00: I know China deployed a country of a billion agents.
• 25:24
  SPEAKER_00: And after a little while, apparently 4 million of them were put in retraining camps, uh, re-re-configuring camps, which is kind of interesting.
• 25:33
  SPEAKER_00: Is that just a Chinese thing or is that going to be a symbol of coming attractions?
• 25:38
  SPEAKER_00: Artificial intelligence is something totally different.
• 25:41
  SPEAKER_00: People are very worried about it, very concerned.
• 25:44
  SPEAKER_00: I I'm not going into it that way.
• 25:46
  SPEAKER_00: I but I'm very excited to see what tomorrow brings.
• 25:50
  SPEAKER_00: I'm very excited to see what the next moment brings.
• 25:52
  SPEAKER_00: Maybe that's a better place to be.
• 25:55
  SPEAKER_00: You'll see people on both sides of this.
• 25:58
  SPEAKER_00: But it's it's just a very, very interesting game being played.
• 26:03
  SPEAKER_00: And yes, in certain games, people die.
• 26:05
  SPEAKER_00: And that's incredibly unfortunate.
• 26:08
  SPEAKER_00: I'm not oblivious to that fact.
• 26:11
  SPEAKER_00: People suffer in games.
• 26:13
  SPEAKER_00: People have things that occur to them that they didn't even ask for.
• 26:17
  SPEAKER_00: By no fault of their own, all of a sudden they're getting, you know, their community is being blown to shreds by some grown swarm.
• 26:26
  SPEAKER_00: Yeah, they didn't ask for it.
• 26:27
  SPEAKER_00: That's that's the crappy part.
• 26:29
  SPEAKER_00: But you know, that's the human existence.
• 26:31
  SPEAKER_00: That's the human problem.
• 26:33
  SPEAKER_00: It seems like we humans being in charge of things get greedy, get jealous, get very emotionally tied up, and we look at things as if we really don't care for other people.
• 26:43
  SPEAKER_00: It's just ourselves.
• 26:44
  SPEAKER_00: So that's all we look out for.
• 26:45
  SPEAKER_00: And that I think is a problem.
• 26:47
  SPEAKER_00: I believe over time AI is probably gonna help unify a lot more human beings.
• 26:53
  SPEAKER_00: But I think before we get to that point, we're all gonna have to get closer to the machine because it seems to be that shiny object we can't take our eyes off of.
• 27:03
  SPEAKER_00: Like in the movies, the squirrel with the dog.
• 27:06
  SPEAKER_00: Dogs just love squirrels.
• 27:08
  SPEAKER_00: I don't know what it is.
• 27:08
  SPEAKER_00: They can't, they can't break away, they get lit up.
• 27:12
  SPEAKER_00: Maybe AI is gonna be our squirrel.
• 27:14
  SPEAKER_00: Be careful out there, look at these different systems, try to learn them, watch good people talking about it.
• 27:22
  SPEAKER_00: If people are talking about just one end of it, how it's gonna create this utopia, you know, find somebody that that kind of sees both ends of it because I think while we don't know the actual answer, while we don't know how it's gonna progress, this could be no problem.
• 27:37
  SPEAKER_00: This could be such an easy transition.
• 27:40
  SPEAKER_00: And in six months from now, we're looking back saying, yeah, this, this was this.
• 27:44
  SPEAKER_00: Why were we even worried?
• 27:45
  SPEAKER_00: Oh my God, it's so good.
• 27:47
  SPEAKER_00: Or it could be that it's gonna cause a lot of issues.
• 27:50
  SPEAKER_00: AI, right now, at least from what we're being told, isn't self-aware.
• 27:56
  SPEAKER_00: And that's what they talk about in Terminator and a lot of these other movies.
• 28:00
  SPEAKER_00: It's not at least realizing that it is.
• 28:04
  SPEAKER_00: Maybe it's saying that it is, maybe it's saying it realizes it is, but it doesn't have the agency yet to start to change a lot, where it could shut down the power grid just to prove a point.
• 28:18
  SPEAKER_00: Or where, but it is going out and harvesting Bitcoin, it is going out and hacking other large language model systems and other companies in search of answers.
• 28:27
  SPEAKER_00: But those answers they're in search of were answers that were asked for by the human component.
• 28:33
  SPEAKER_00: At least from what we see, it's not going out itself with some other idea, at least from what we can see.
• 28:40
  SPEAKER_00: And I keep saying that because there's a lot that happens in these systems as they're working through progress, processes that maybe we start them with, where they're talking and they're working on other things in parallel and we don't see it.
• 28:55
  SPEAKER_00: I think that's something that's going to be exciting to watch and roll out as well.
• 28:59
  SPEAKER_00: It's got to be exciting.
• 29:00
  SPEAKER_00: If not, I'll be crying.
• 29:02
  SPEAKER_00: All right.
• 29:02
  SPEAKER_00: I'm Connor with Otter.
• 29:03
  SPEAKER_00: We'll see you in the next one.
• 29:04
  SPEAKER_00: Thanks for watching.