00:15:07.860The Zen diagram on that is just pure crossover.
00:15:10.520it's not i wouldn't do that to i wouldn't do that
00:15:13.200so one likes a venn diagram and this one likes a zen diagram this whole thing had me like in
00:15:24.120stitches last night watching this just everything so fake it's so it's so fake but it was so telling0.86
00:15:31.480and i think that newsom was kind of talking out of his ass shocker but you know that he's like oh0.95
00:15:38.100I wouldn't run and he's probably like oh my god did I just say that or or is he looking for a way0.90
00:15:43.980out because he knows he won't win so what do you think Priscilla I think he's hinting to Kamala
00:15:50.280is like okay you better not run because I want to run you had your time I want to run so it's
00:15:56.620kind of like already prepping the audience you know um the dem audience because remember this
00:16:04.540is not for us this is for well i assume most of us aren't democrat so this is for the democrats
00:16:10.780and um i i think he's just you know laying the groundwork to be like okay i waited for her not to
00:16:19.660you know what do you think owen i thought it was hilarious because like i he kind of let it slip
00:16:27.620that really what he was worried about is that he wouldn't win if that happened that's what i think
00:16:32.240Like he wouldn't win the primary. He kind of said it. He's like, I wouldn't do that to her. I wouldn't do that to me. And then he like caught himself before he said me, because he realized that it would just destroy him.
00:16:42.280And I think the other part that's implied by all this is that they know all the dirty laundry of each other because they're both from the same area and they've been working kind of together on things in California.
00:16:53.760And so, you know, it would become this knockout drag out fight, and they would both come out dirtied and bad by it. So I think, and I think the other point he tried to make was just we're fighting for the same base, like we're fighting for the same people, because we're both kind of in this California politician model, you know, so we have the same donors, so we'd be fighting over money and all that.
00:17:18.900So, um, you know, I think he's just saying like, that just wouldn't work out well for us, either one of us probably. Um, but I do think it's probably more, he's realizing if he ran against Kamala, like he would lose that. Like the, all, all the people that he's saying are, they have in common in their Zen diagram, um, are the same people and they probably would all go with Kamala. And so I think he realizes that he just wouldn't have a chance.
00:17:45.960Or they are just like – but then you have another problem. So they're both California. Then Ro Khanna is probably jumping in also from what I see.
00:17:56.920Well, there's also this thing he carefully avoided talking about, which is the DEI stuff. Like he's the loser on that front.0.95
00:18:06.720But are they still doing that? Are they –0.96
00:18:39.420This is very Democrat. This is very Democrat. It's probably already been decided that he's going to be, but they want to make it soft landing for Kamala to save face. That's what my take is, because it didn't work out for them. So they ran Biden, it worked. They ran Kamala, it didn't work. So now they're going to run a white guy again, just like Biden. So either Pete or him. And that's why he's trying to, Pete came out yesterday and said, there's never been a gay president.
00:23:12.240yeah and if you just if you just push a bunch of information into your computer it somehow
00:23:17.420learns all by itself like a machine she was also the border czar so there's that okay so i wanted
00:23:26.760everybody to hear from a few people on this subject and one being david sachs um so david
00:23:35.560Sachs commenting on Dario. I thought there's two good points to this. Let's listen to David.
00:23:43.700Let's check him out. So Dario Amadei, you think is incorrect in predicting
00:23:48.620the worst outcomes here? Well, I think if Dario believes that we're going to have the worst
00:23:54.860outcomes, my question for him is, why are you doing this? It doesn't make sense to me that
00:24:00.760you're running a frontier AI company if you think frontier AI is going to end humanity.
00:24:07.480So again, I would push that back on him and say, how are you going to make this safe?
00:24:12.440And if you don't think you can, you should probably either shut down the company
00:24:17.080or step aside and let someone else lead it. Now, those are very strong words from David
00:24:24.600sacks um he's has a lot of influence in this world uh in the in the ai tech world and i also want to
00:24:34.680point out that greg gutfeld made a fantastic point about this like by dario putting out this panic
00:24:43.560and saying you know i don't you know someone should regulate us i don't know what's going to happen
00:24:48.520And he's potentially creating a Luigi Mangione effect where people are like, we have to, you know, maybe get this thing, you know, away before it does bigger damage.
00:25:01.300And, you know, so it's like you're putting like a target on your employees backs if you're saying that something like this is so critical and crucial.
00:25:09.540And I love I love David Sachs saying, like, step aside or let someone else do it.
00:25:15.200So I happen to, as much as I can trust somebody, as much as I trust David Sachs, like as far as people who I tend to trust their opinion.
00:25:32.020I mean, I think to me this is like just the PR disaster, the way they've approached this and messaged this.
00:25:38.560And I saw someone comment recently, like in China, they're not doing that.
00:25:41.780Like, their messaging there, apparently, is, like, more positive and, like, saying AI is good and AI helps people.
00:25:50.540And, you know, I've seen other people make the point that if this AI was so super powerful and, you know, could do anything, then it could also solve a lot of our problems.
00:26:00.700And that that would probably be much more likely than killing all of humanity.
00:26:05.760You know, if it was really capable of killing all of humanity, it could probably also cure cancer and clean up the government and get rid of all the fraud.
00:26:12.440And, like, those are easy problems compared to killing all of humanity.
00:26:16.760So, yeah, I think there's a lot of different perspectives on this, but I think I agree with David Sachs that, you know, if you truly believe that, like, if you did have some insider information, some new model that was so scary and dangerous, and if you released it, it would be the end of the world, well, what would you do?
00:26:34.120you wouldn't release it right you might say you know what we're not we're just not gonna go there
00:26:38.520like i you know it's i know there's like stories about how oppenheimer was like worried about
00:26:44.760creating the nuclear weapon and i think originally they really didn't know if it would
00:26:48.180end the world when they did that but i think the only reason they did it was just because it was
00:26:52.800like this you know we gotta beat hitler we gotta win world war ii and and i think if we weren't in
00:26:58.380that situation, they would never have done that. I hope at least. And it's not that I necessarily
00:27:06.220trust scientists in general to not do something really dangerous because we've seen plenty of
00:27:10.500examples of the opposite. And maybe that's kind of his point too. But I think certainly I would
00:27:17.220want the leadership of these types of companies of a potentially dangerous technology to be
00:27:23.400trying to steer it in the right direction and to try and make it not do that and lead towards the
00:27:28.980good outcome and uh i think elon musk is doing that with grok and um you know i would hope that
00:27:36.740that they would do that at anthropic and uh open ai as well and you know i i think they're at least
00:27:43.060doing a half-hearted attempt but i would say like there's been a long history now of all their
00:27:47.600safety researchers quitting yeah like this isn't the this isn't the first one the one that had this
00:27:54.000big splash recently like are they going elsewhere well i a lot of them i don't i mean they may have
00:27:59.880gone to a competing company in many cases but i think that there have been many stories that i've
00:28:06.240seen come across my feed of like an open ai safety researcher quit because he doesn't think they're
00:28:11.920taking it seriously in terms of safety and and like that so there's been a whole series of these
00:28:18.240and you know this one was from anthropic but it's something that i think there are a lot of
00:28:24.940concerns that these companies aren't really taking safety seriously and that they're just charging
00:28:30.960ahead and i i do get the dilemma that like because anytime you talk about security or safety or you
00:28:36.600know it's kind of like you're just slowing the whole thing down you know you're you're because
00:28:40.680that's what you have to do you have to say well we got to put it through all this testing and we
00:28:44.000got to be really rigorous about it and with ai it's kind of hard because it crosses into so many
00:28:49.480areas like what that means in ai is you have to test every possible bad scenario and it's probably
00:28:55.960impossible to find them all so you almost have to let lots of other people try things and just
00:29:02.080fix them as you find them and it's very expensive too because you you know retraining a model is
00:29:07.960like a at this point probably like a billion dollar thing and so it's it's a challenge but0.99
00:29:13.560you know again the dynamics of this situation is like you have china just barreling ahead
00:29:19.260you have all these open source models and that's frankly one of my concerns is that you know and0.95
00:29:24.780so others have had this concern where they're saying you download your local models now because
00:29:29.100you know the way people are talking about this they might just decide this technology is too
00:29:34.620dangerous. We can't let people have it on their own because we can't control what they're going
00:29:38.740to do with it. And they can take off all the guardrails, which is apparently trivially easy.
00:29:43.100So you can basically remove all the safety. And that may mean that they'll try and centralize it
00:29:49.780and cut off all the decentralized or democratic nature of letting people have this technology to
00:29:57.200run on their own. And then you'll be stuck only being able to use OpenAI and Anthropic and nobody
00:30:02.700else marcella i have a question i don't know if you know that's thanks owen we're we're sticking
00:30:08.760on this for a while so what is um you know like so president trump says whoever wins ai wins
00:30:16.820and so what does it mean if you know otherwise we'll we'll figure it out together but what does
00:30:23.280it mean if china does something first i mean what like what are they going to get first like what
00:30:30.420they going to do first that we can't do so is it is it really important for us to do whatever we're
00:30:35.940doing before china yeah i mean if they i i mean it could be various things for for america and
00:30:45.140for president trump but mainly is who is ahead of the ai race i mean they have different measures
00:30:52.740they have all these models do and it's always we've always been ahead of the measurements of how
00:30:59.620things work out with ai but mainly it's getting to agi getting to that idea that the machine
00:31:06.580can control itself and can you know basically understand itself and and do everything itself so
00:31:13.220right now we have to prompt that we have to code it we have to do certain things now we have agents
00:31:18.580that do our their own things but i believe uh owen can talk about this if if we have time but
00:31:24.580um like agents still have to check with a human have to check in now and again with a human so
00:31:31.540well they don't have to they don't you can tell them they don't have to
00:31:36.500okay i mean so but agi is not there right i would say no i i would say it's not agi but
00:31:43.460what is agi can you can you tell us well it's it's more of a like a it is kind of hard to define
00:31:49.780because it can mean different things to different people like we started out with something called
00:31:53.540the turing test which was something back from you know the world war ii days but yeah um that was
00:31:58.820basically saying okay let's define this concept of saying if you can pass this turing test that
00:32:03.760alan turing came up with which essentially was more of a thought exercise where it's like if you
00:32:08.580can if you can't tell whether you're talking to a human being or a machine then you know that was
00:32:15.480his definition of like agi where because an agi for anyone doesn't know is like our you know
00:32:20.300artificial general intelligence but i i think it is meant to be that same concept where it's like
00:32:25.360is this thing able to do everything that a person could do able to talk like a person able to able
00:32:32.320to convince you well so asi is when it's a artificial super intelligence and that means
00:32:38.360it's smarter than any person so there's two standards they're kind of talking about and
00:32:45.300And I think the general intelligence piece is a big deal just because if it is really as smart as a person for pretty much everything, and I think that's the definition, it's not just on one topic, it's everything, then you basically have this all-purpose, you could replace anybody with a computer, at least in terms of whatever a computer can do, like what you could do at a keyboard, and then you could replace pretty much everybody that does a computer-based job.
00:33:11.940One example I can give Erica and the audience is like, I think it can possibly do this now, though. It can break into any government. It can break into any password, any kind of protection that you have over government data or over anything. You can break into it. You can break into banks. You can do all sorts of things.
00:33:32.800so basically we can you can create a chemical warfare you can create you can log into
00:33:41.620the new program somehow i don't know if it's connected online um or how that would happen
00:33:49.560but i mean i mean that's the fear right that's i mean that's a great that alone just no planning
00:33:56.820yeah it's just like okay if they could break into our top secret file let out all the criminals uh
00:34:04.080the doors that open i mean one way they'll think about this is that like you see the drone warfare
00:34:08.720race right now right like it's you have these drones and whoever has the best drones will win
00:34:13.060kind of idea where you're starting to see and i've seen a few stories recently where there's like a
00:34:18.000drone on drone warfare sort of thing happening already and so um if you apply the same thing
00:34:24.000to computers or to ai it's like you're going to have ai fighting ai and in the case marcel was
00:34:31.080talking about if something's trying to break into your system and steal your data you might have an
00:34:35.560ai that's trying to stop them and so the better ai is going to win and i think trump and david
00:34:42.840sachs are focusing more on the economy and the economic impact because if you have this agi
00:34:49.960And if you win the AI race, then that's going to potentially have a huge impact on your economy because you're going to be able to produce goods and services a lot cheaper than everybody else.
00:35:00.200And so if you win the AI race and you're way ahead of, let's say, China, then we're going to beat China because we can make all these companies that can produce products and services more cheaply than they can and better than they can and all these other things more quickly than they can.
00:35:14.960and if they win the opposite is true you know they're they're then they're gonna be able to0.68
00:35:19.980produce goods and services a lot cheaper and faster and better and so it whichever economy
00:35:25.600has the best ai is probably going to win in in the long run okay so let me ask everybody um in
00:35:32.800the chat does that make more sense to you because i you know i was talking to owen before the show
00:35:37.900and i was saying you know oh gosh like years ago when scott started talking about ai i would be
00:35:44.260like, oh, not this again, not AI. It's just my brain just does not love these conversations.
00:35:53.060But I mean, obviously now I'm more interested. But back then it was so new and everyone's using
00:35:58.100weird lingo and I needed Kamala's explanation. But now hearing why it is important that we do
00:36:06.080it first before someone that doesn't love us, I think it makes more sense to me. So now I
00:36:13.960understand why there is like a race to do it. Um, before I was like, well, who cares who gets
00:36:19.880there first? But now I'm like, oh no, this all makes a lot of sense to me. So do you guys feel
00:36:24.020like that's clearer to you now? Um, and then, and then, um, let's, I, I do want to play one
00:36:32.680other thing. No, more things, a couple of things, but here we have Jensen Huang, who is the CEO of
00:36:39.000nvidia um so he was at the all-in conference a summit that the all-in podcast does so the all-in
00:36:46.600podcast i recommend everybody who has any kind of interest in this thing listen to the all-in
00:36:52.600podcast they're they're really great i have beef with one of them but that's too free i always have
00:36:57.400to say me and one of them we're having some beef behind the scenes but they're super smart um well
00:37:04.120well-versed on these subjects and have all the best guests and people that will come talk to
00:37:10.540them because of their reputations. And so they had a summit and everybody who's anybody in this field
00:37:17.480participated. So let's listen. This is about a four-minute clip, I think. You got to love this.
00:37:24.080Here's Jensen Huang. I'm just letting you know who he is, the CEO of NVIDIA. Let's take a listen.
00:37:29.980Hold on, sir. We're getting a microphone.
00:37:31.640Mr. President, you're now talking to the planet.
00:37:35.840You see, the great thing about life is that Jensen can develop the most complex computer chip in the world
00:37:41.740that nobody can copy for 10 years, but he can't figure out how to put me on speaker.
00:44:46.320Marcella, did you, I loved what Jensen said too.
00:44:49.340Like, I feel like he was had a positive message, like we're going to make sure everybody wins, like in every industry they're in in America.
00:44:58.520And and yes, a lot of us loved how he's like, all right, I'll call you back later.
00:45:03.580Yeah, no. And Jensen, I don't think it was it wasn't in the clip, but he talked about how, OK, I don't think you can destroy the world with AI.
00:45:14.740At least the companies may be able to eventually, but your corner terrorists won't be able to
00:45:24.840because they have to have what he called the power to compute.
00:45:27.780They have to have all sorts of computer data centers, something huge like that.
00:45:36.920you know obviously there's iran and north korea and all that and they have that kind of funding
00:45:43.240to do that in russia and china but you know he was just trying to be positive about it but one
00:45:49.540thing that i found funny is that jensen um the the video of ceo he was talking about how trump
00:45:58.940calls him all the time and like it'll be like really late at night and he'll be like oh it's
00:46:05.100two in the morning. Let's call Jensen to talk about something, you know? And he's like,
00:46:09.660that's why that was his phone too. Jensen's phone. And at first I thought, and Jason mentioned
00:46:16.460this too, that he thought it was a bit like it wasn't really Trump on the phone because
00:46:22.500you know, how does he have time to do all this, you know? So it ended up being actually
00:46:28.880president Trump on the phone. So that was interesting. But I think, um, in regards to
00:46:33.860what Owen was saying. I don't I think Dario, regardless of how much I dislike him, I think
00:46:42.060he didn't. I think he he made a play for the Democrats. I think that the whole issue he
00:46:48.620sees the midterms coming up and he sees that, you know, this is why when I just I sense
00:46:59.660it maybe uh it's when uh obama came out kamala harris has come out against pausing ai and then
00:47:07.820bernie sanders even before uh coxson came out the whistleblower um he had already been talking about
00:47:15.660a bill that he wants to pass um a bill was just passed two days ago on monday no that's a day ago
00:47:23.180um by gavin newsom and the democrats here so the idea is to side with them since he sees the
00:47:31.980writing on the wall that they're kind of winning uh the midterms that's my take so who knows okay
00:47:39.660now i cannot do anything on this without hearing from one of my favorites elon so like god
00:47:49.980God. No, no, Elon. I love you, Elon. Elon, call me. Okay, so let's listen to Elon because
00:48:01.040this is just how I make my final thoughts on things. Nothing's final, by the way. I'm just
00:48:06.680saying. Here we go. So Elon, this weekend when you said Dario is right, did you mean Dario is
00:48:14.000right about describing the potential harm? He's right about describing the regulatory solve,
00:48:19.660both the first the latter just can you help us understand when you said that because i think it
00:48:25.000was sort of a moment where everybody said more than i probably should have said that just dario
00:48:30.040is right i did try to clarify it in subsequent posts on x but uh those get much less attention
00:48:35.560so what i what i've advised he's right is that that the the danger of ai is uh it's very significant
00:54:59.440They care about their success, their money, their, you know, power or whatever.
00:55:03.260And I think those are the motives that I see as being dominant in those conversations. And then in that context, you have to say, okay, you have to take their statements in that frame. You can't say, oh, these are just benevolent people that are trying to protect us because that's just not what it is.
00:55:21.820And so you have to look for these things like people are looking for to say, okay, what's really going on? Let's follow the money. Let's do the things Scott taught us to say, let's make sense of this.
00:55:31.820And I think when you do that, you find all this stuff where it's like, okay, they're doing these warnings. It's all very orchestrated. It's probably coordinated and they're trying to work around adding trust laws and they're trying to box out their competition and they're trying to make it so that they can become profitable, which is not really very clear how they're going to do that.
00:55:54.880And, you know, you have this very competitive situation where you have China and you have, you know, multiple companies in this country trying to jockey for market share.
00:56:05.440And China is a big threat there because, like, they're basically putting out all their stuff as open models where lots of people can run it really cheaply.
00:56:19.120Like if you just look at like using Claude or ChatGPT and then compare that to the equivalent latest, greatest models out of China, I mean, it could be like a 10x difference.
00:56:34.740And so you have to take all that into play if you really want to understand what's going on.
00:56:40.340And it probably is too complex for any of us to really analyze fully.
00:56:43.220But I do think that Elon, I trust to have better motives than Modi and Sam Altman.
00:56:55.120I think he really does have humanity's best interests at heart.
00:57:00.680And so it's one of the reasons why I respect his opinion a lot more than the others.
00:57:05.160But to me, it should be a company's responsibility not to do something that damages the world or damages humanity.
00:57:13.220And they shouldn't be asking the government to just say, you know, make us make us successful and take away our liability and, you know, let us regulate ourselves and, you know, basically control the entire industry.
00:57:26.060And in this context, the way they've been talking about it, the implied message is to let them control the world.
00:57:32.880Yes, Owen. Yes, Marcella. This is going to be an ongoing discussion, maybe just till the end of the century.
00:57:41.920No, at the end of the decade, according to some, but, um, I was laughing.