Has AI Gone Too Far? Open AI, Math Breakthroughs & Ethics Concerns

You ready? The AI kerfuffle? >> No. >> [laughter] >> I found about out about this at like 3:00 p.m. today. >> I've been reading stories. Ready for this? >> I was like, "I'll take a quick break from my day job and open up threads. Oh, no." >> [laughter] >> Okay, number one. Uh the the CEO of Nvidia saying we've reached artificial general intelligence. This is hyperbole. It is hype. It will boost the stock shares. It is and >> Why? It should do the opposite. >> All of their cards are being used for LLM data center the uh you know, for artificial intelligence use. Um They're the processing for that is so important to Nvidia's bottom line. Um they are so sold into it. Going on further, though, the other news that was interesting this week is the announcement that OpenAI has solved one of the millennium problems. And there's a a series of six math problems that have been too hard for people to solve. Where there was one guy in like 2000 or so, 2007, 2008 uh from Russia, I think, who solved one of these millennium problems. You get a million dollars if you solve one of these problems. This researcher The researcher who got the first one solved, he said he he said, "No, I don't want the money. I just wanted to do the math. That's all it was about was to show it could happen it could be done by people." The love of the game. I love that so much. And so now, um OpenAI has come out and said, "We have solved the Navier-Stokes equation. We have solved this problem that has to do with uh the three-dimensional water particles and you know, what happens if they get moving all really fast and there's no limit to how they move, what happens? And nobody's been able to show exactly what happens in this situation and everybody's like, "Whoa, if they just keep going, do they explode? Like what happens to water? Does it like become a singularity? Blah blah blah blah blah." All this stuff. Anyway, apparently their AI which is an unreleased model, they did release their latest which is you know, the timing of them releasing a new AI model that's like better than all the others right now and them saying we have another one that we're holding back that we used to solve this problem. I mean, the timing of this is so >> curious. >> Well, so curious. Well done. Um and so uh the math is interesting and so the math community is coming out and saying that's um really interesting and they've used the standard which is a like a a paired-down algorithm that does the the proofing or the checking of the proof that like all mathematicians use this computer algorithm to be able to test their solutions and so apparently OpenAI has used the standard, the lean standard that mathematicians use to validate their solution and it's right and they say it's validated, but the rest of the math community is like we don't really believe you yet, so we have to do our own validating. So it hasn't been validated by the math community. Now, the second news that came out of this one story like in the same same day basically is that there are a couple of mathematicians, one of of at New York University and another who works at Anthropic. And they have been working separately on these equations, one of which is related to the Navier-Stokes, which is the Euler, which is this 3D three-dimensional modeling problem. And they published a solution to one part of to this Euler aspect of the Navier-Stokes. And they said that they had been using Anthropic's uh assets, their platform, as well as OpenAI's and other models. So, they've been using multiple models to develop their solution to this problem, and they were getting toward a solution to the Navier-Stokes. When suddenly OpenAI said, "Aha, we've beaten you to it." And apparently, this is the big kerfuffle is that the mathematicians are number one saying OpenAI needed to kind of beat them to the punch because one of the involved mathematicians was an Anthropic employee, an Anthropic mathematician. Secondly, that there is question as to whether OpenAI has used the mathematicians' like private queries and what they and and all the data that they processed within the OpenAI platform >> Yeah. >> to solve the Navier-Stokes. And so, basically, there's >> that AI does. >> Yeah, it's a whole question of theft at this point. >> That's the That's the whole thing >> theft theft, intellectual theft. >> thing it does. Yeah. Is it iterates off of other people's stuff. >> Yeah. So, anyway, this is the >> What are you talking about? >> Yeah. And it did you know, in addition, we have, you know, the the the breaking of the barriers that have been put up. Multiple AI agents working together, like working multiple AI agents working together can get a whole lot done cuz they don't care about rules. Their only rule basically is to survive and win. So, this is the problem. And somebody Uh so, there was someone who left Anthropic, I believe, this week and said, "I'm leaving." And there's a bunch of talk about there's people who work here who are worried about what whether AI is going to destroy humanity. >> Mhm. >> You know, in a decade. >> Um yes, there's a there's a greater than 10% chance. >> This is like >> in a decade. >> Where did they get their numbers? This is BS. Nobody's done that calculation. >> No. >> They This is talking out of >> If this is 1% chance, shut it all down. >> It's still too much. >> Shut it all down. If that is a real conversation that is really happening, if it is a non-zero chance, anyone continuing to support this from within is morally bankrupt. It's like unacceptable to me. >> Well, this is >> that your AI could realistically breach, >> If you have any questions, >> right? >> and act, nope, shut it all down. Not acceptable. >> It's unacceptable. >> Mhm. >> And this is where we need to get at. So, we have um government employees and uh officials who really don't know what they're talking about, who are involved in this. They are being talked to by um you know, tech people who are not completely honest about what's happening because there's profit margins in place. All these people are driven by profit profit margins and while there may be employees of these technology companies who are really interested in the intellectual discovery of where they can go with this technology, they are still morally bankrupt in that they are not considering the ethical application of their technology. And so you have individuals who have not considered the implications or the ethics of their inventions negatively impacting humanity and those setting those inventions loose in a way that is not regulated or controlled either by the government or by the people making the technology in the first place. And if they if it something goes wrong, the technology companies they you've already seen it happen in the last couple of weeks with hugging face. It's like, oh well, you know, it what a little little little you know, there's a lot of hand-waving and they're not talking about the fact that they did not build enough a strong enough box and they didn't think outside the box of how these agents might act. Even with all their experience with this technology, they're still not thinking about this. >> No. >> And so Yeah, no. No. No. But anyway, AGI we have we have not hit that. AGI do All they're doing is they are they're diminishing the value of the idea of intelligence by even saying that they've gotten to AGI. Second, it's hype. It's hyperbole, it's sensationalism, and it's all based on a bottom line. I think people are going like out of their heads for the wrong reasons, but at the same time this is a technology that needs to be managed and regulated because the the technologists are saying they're scientists, but nobody is regulating them. They are not being managed, and they need to be. I mean, yeah. Ban the whole thing, exactly, Aaron Laura. I am I'm I mean, the data centers, we're doing fine. There are lots of AI tools that are valuable, things that are helpful. I mean, I'm saying AI, but really it's processing, right? It's putting things together in useful ways. There are tools that are absolutely valuable, and okay, let's put all of our energy into fission, fusion, solar, wind. Let's like why not try all the different things to make sure we have sustainable energy, work forward. There's all cart, horse. >> [laughter] >> Well, I think, you know, once again, this is by and large, a huge problem with the fact that we're calling it AI, and this is and this is like the colloquial term used for this thing, and it's allowing people to put like pronouns on it and like say like they'll name their AI and they'll talk to him or her on a regular basis and develop a relationship with the AI and like and and have this whole theory that it's like Hal or or the movie Her or like you know, the Jarvis or whatever. And and so you you kind of you expect this kind of behavior from it, but it's still a query machine at this time. Like I love when you get to watch videos of people using the AI and trying to ask questions in a linear kind of route, and it doesn't really work. Like they're basically just running new queries and looking at the at the at the chat history when they run the query, but it always kind of goes amiss because they are each each time you ask the AI a question, it is an individual query being run. And so it's still they're still in these kind of silos where it's just an in-out machine in a weird way. Um it's not thinking. >> No. >> That's That's why you couldn't just ask the AI to solve a math problem, a complex math problem. You had to They have people who've been working on it for months because they have to ask specific little elements and then incorporate it into the math they're already doing on their own. It's um yeah, it's it's it's a huge branding problem, which obviously is benefiting the companies themselves because it sounds sexy and exciting and >> And they're going to keep pushing it. >> But if you told somebody like, "Hey, I made a search engine, but it's worse and it makes things up." Everyone would be like, "You're fired. No No more money for you." >> [laughter] >> But because I made an AI that can like tell you you're really smart and can make things up, isn't that cool? It can hallucinate. It's like, "Oh, okay. Here's a million dollars." >> I do not want my Google Maps hallucinating. >> No. >> Right? Like you don't You want to know where to go. You don't want it to I mean, people used to accidentally walk off cliffs because they trusted Google Maps too much. >> Or drive into a river or like, yeah. >> that was just with poor directions, right? And now if it's AI making things up, >> [laughter] >> I am afraid. >> Well, and the number of times even this week alone that someone has asked me to ask Copilot something >> No. >> [gasps] >> makes me want to scream when I am trying to have a back-and-forth conversation with a person and talk through a challenge. And at some point they'll be like, "Well, what did you ask Copilot? What did Copilot say?" No. No. >> No, I did not. >> I'm now developing a reputation. >> I was thinking about it. >> Yeah. So, they they know if they're going to ask me to ask AI something, people will just they'll know like, "Oh, you're not going to ask." I was like, "Yeah, if you want to ask AI, you can ask the AI, but >> I'm not going to. >> [laughter] [gasps] >> Oh, get off my lawn, AI. >> Mhm. Mhm. >> I love it. AI, automatic idiocy. That is That's pretty good there. That's pretty good. Yeah, and hallucinate wrong results. Yeah. What was it? >> Atrophied intelligence. >> Atrophied intelligence, I like that. That's great.