AI Predictions That Will Completely Change Life In 2025

0:00:04 I think 2025 is going to be wild with what we see with AI video.
0:00:09 For anyone who’s saying that these models do not reason and they’re just like parents or whatever, no.
0:00:16 So the average IQ range for adults is between 90 and 109, with 100 being the theoretical average.
0:00:23 So estimated for 03, 157. This is like top of the top smartest people in the world level.
0:00:30 Hey, welcome to the Next Way Podcast. I’m Matt Wolf. I’m here with Nathan Lanz.
0:00:36 And there has been a lot happening in the AI world, including an announcement that a lot of people
0:00:42 think is just this monumental, huge, massive announcement in the fact that we got this 03
0:00:49 model from OpenAI. Well, I guess I can’t really say we got it. We got to see it. We got to see
0:00:55 the 03 model from OpenAI. But a lot of people have been, you know, touting it as like this
0:00:59 massive breakthrough. I’ve even seen like YouTube videos give it the title of like,
0:01:06 we finally have AGI. And everybody’s saying that 03 is AGI. So in this episode, we’re going to talk
0:01:12 about that a little bit. We’re going to talk about our predictions for 2025. And Nathan’s going to
0:01:17 show off something pretty cool that he’s been working on using some of these models. So let’s
0:01:24 get into some of the like 2025 predictions. You and I obviously talk a lot. So chances are our
0:01:29 predictions are going to be fairly close to each other. But let’s go ahead and dive in.
0:01:34 Yeah, let’s start on the models, I guess. I think by the end of the year, a few things are going
0:01:40 to have happened. I think XAI is going to roll out Grok 03. And it’s probably in terms of like a
0:01:44 fast LM that you can just chat with for daily use. It’s going to be really, really, really good.
0:01:48 The fact that they’re just, they’re trying to train it on more data than almost anyone else.
0:01:53 And they’ve, you know, they’re buying more chips from NVIDIA than almost anyone else.
0:01:56 I think it’s going to be really good for like daily, just like, just chatting with AI.
0:01:58 It’s going to, I think it’s going to surprise people. But in terms of like,
0:02:02 Do you think that’ll replace some advanced voice mode for ChatGPT for you?
0:02:07 I don’t know. I don’t think so. I don’t, I’m not, I’m not sure. Like I think for power users,
0:02:13 I’m pretty convinced ChatGPT and OpenAI are going to be even further ahead by the end of 2025.
0:02:19 Like, I think there’s a decent chance that, because like 03, if it’s just 01 with more
0:02:22 compute thrown at it, well, they’ve definitely been working on other stuff. They’re not just
0:02:26 sitting there like, Hey, let’s take what we’ve created and throw more compute at it. And like,
0:02:31 that’s all, that’s all we’re doing for the year. Like, it makes sense to me that like
0:02:36 something like Orion, like a larger model underneath as the baseline will be part of
0:02:41 some new model by the end of 2025. And I think there’s a decent chance that they actually
0:02:44 release it in 2025, like by the end of 2025.
0:02:49 Well, Sam Altman hinted that he believes AGI is coming in 2025, right? So yeah,
0:02:56 yeah, like I said, like I think 01 Pro is like some kind of basic AGI already. That’s my definition.
0:02:59 People like some people think that’s like stupid. Everyone had different definitions of this stuff.
0:03:05 But I think by the time you get to 04, when not only it can like respond quickly to like basic
0:03:10 questions and have like incredibly good answers, but also if you ask it something complicated,
0:03:14 it’ll give you something back in a minute or two that’s amazing. Most people are going to call
0:03:20 that AGI. And I think that could be coming by the end of 2025. And that will make all this stuff
0:03:24 we talked about, like in terms of agents possible, like agents that go off and do market research
0:03:28 for you, come back and bring you some charts and show you what, how you should be approaching a
0:03:32 market and what’s the best strategy. That stuff will be agents, how you interact with email,
0:03:38 do email, like your assistant. That’ll probably be an agent by the end of 2025.
0:03:41 And that’s going to be a dramatic change. That’s going to be like where we’re going to start to
0:03:47 have some real questions, societal questions by the end of 2025. And it’s going to be interesting
0:03:51 to see how that plays out because, you know, I have said in the past, well, Donald Trump’s
0:03:58 administration is going to be very pro AI. But also there was that recent thing with the port,
0:04:03 what was it, the port unions, unions for the port workers, where he kind of backed down from them
0:04:06 saying that we need to protect their jobs and we shouldn’t automate the ports.
0:04:10 And we might see more of that because like there may be such
0:04:17 cultural backlash to some of the job loss that it’s, you know, it may become a pretty tense
0:04:21 conversation and by the end of 2025, because I do think we’ll start to actually see dramatic
0:04:25 changes in the job market. Like maybe everyone’s not going to fire people, but you’re going to see
0:04:30 like, oh, I’m not hiring as many engineers anymore. Like there’s less job openings, less
0:04:37 assistance hired, things like this. Yeah, I agree. I think I do think like the big buzzword of 2025
0:04:42 is going to be agents though. I think like every company is going to be talking about AI agents.
0:04:47 I think you’re going to be hearing Google, open AI, Claude, X, you know, you name it,
0:04:53 they’re all going to be talking about agentic workflows and tool use and being able to like
0:04:58 do things on your computer, on your behalf. And like, I think that’s going to be the conversation
0:05:03 of 2025. I think if there’s any one prediction that I’m fairly certain on, it’s that. I also
0:05:06 think the video models, I don’t think we’ve seen anything yet. I think, you know, eight-second
0:05:12 videos that look pretty decent. I don’t think we’ve seen anything yet. I think 2025 is going to be
0:05:16 wild with what we see with AI video. I mean, Matt, like, let’s still try to step back for a
0:05:19 second. Like, like realize, like, you know, we’ve been talking about this stuff for like over two
0:05:25 years now. Like, you know, in 2023, the videos looked horrible. I was sharing the, I was sharing
0:05:29 threads about them just because they look cool because of technology underneath them. I was
0:05:34 like, this is amazing that you’ve asked AI to dream up what a video of this scene. That’s
0:05:40 incredible. My co-founder for one of my previous startups, JR Bedard, unfortunately, he passed
0:05:44 away, but he used to be fascinated with this stuff. He would like, he would like read books about
0:05:47 like how AI was going to be able to generate arc and he would show me all this stuff. This was
0:05:52 maybe like 10 years ago. Right. Oh, that’s cool. It can make a little pattern and stuff.
0:05:55 Yeah. Yeah. Yeah. And it’s like, that’s where it was 10 years ago. And it’s like, that was like
0:05:59 the seeds of what we have now, but it was like, oh, it can make out a little, a cool pattern.
0:06:02 And I was like, oh, that’s cool. And I was like, do you think that’s going to be like a real thing?
0:06:06 Like in the next like five or 10 years, and it was like, you know, maybe in like 10 years,
0:06:11 and it’s about where it is now. And the videos looked like they were cool if you understood
0:06:16 the technology, but they looked horrible two years ago. And now are they perfect to know? But
0:06:21 like, man, they already look like like 90, 95% of what you might see in a film, like at least
0:06:27 like a short scene or something. Yeah. And, and it feels like too, that, you know, I think I may
0:06:32 have said this offhand the other day, but I believe that where the video models are right now, they’re
0:06:37 kind of like in GPT-4 level in terms of they’re just trained on massive amounts of data, but they’re
0:06:43 not doing a whole lot in terms of reasoning about that data, about that output. And so, so once you
0:06:47 start to, and this is why I think like probably Sora is probably still going to be very good.
0:06:52 Because if OpenAI is that good with their internal models, as soon as they apply that to video as
0:06:57 well, like reasoning on top of it, you’re going to get so much more control of the output of these
0:07:01 videos. You’ll be able to like, yeah, I wanted the character to turn that way. I wanted to do this,
0:07:04 and also make sure that the characters are consistent, because then you can reason upon
0:07:08 the output and make sure that it’s consistent. That quality is going to go up dramatically.
0:07:14 Well, I mean, we’ve got AI large language models now that it will basically generate an output,
0:07:20 double check its output before sending it to you. When you look at like AI art generation,
0:07:28 the big way of doing AI art early on was GANs, Generative Adversarial Networks, where it would
0:07:35 essentially, you had a discriminator and a generator, and the generator would generate a
0:07:42 piece of art, and then the discriminator would go, you would give it a prompt like draw me a cat,
0:07:45 and the generator would draw a cat, and the discriminator would go, that didn’t look like
0:07:48 a cat, try again, and it would draw another cat, and the discriminator would go, that doesn’t look
0:07:54 like, and it would go back and forth until the discriminator couldn’t tell the difference between
0:08:00 a real photo of a cat and an image of a cat. We haven’t seen that kind of tech inside of video
0:08:07 yet, so I think we’re going to have something equivalent to like a GAN, or like what we’re
0:08:12 seeing with like O3, which is kind of the same idea, right? It does a generation, double checks
0:08:16 itself, goes, all right, there’s things that needs to be fixed, double checks itself, takes it to be
0:08:21 fixed, and then finally presents the output once it’s confident in the output that it gave you,
0:08:26 right? Very, very similar conceptually. Well, I think we’re going to get to a point where
0:08:32 we do that with video models, you give it a prompt of like a unicorn flying in space with a cat
0:08:37 writing on its back, it will generate that, it will watch the video back itself and go,
0:08:40 that doesn’t look right, generate it again, that doesn’t look right, generate it again,
0:08:46 and eventually it spits out a video that looks exactly what you had in your mind of what a
0:08:51 unicorn flying through space with a cat writing on its back would look like, right? Like, I think
0:09:01 we are still at the like mid-journey v2 level of AI video, right? And if you remember, if you
0:09:07 followed along to like AI art, mid-journey from v, I think it was v3 to v4 was this massive leap,
0:09:11 right? Once v4 came out, that’s when people started fooling people on Facebook with like
0:09:16 the pope and like a big old puffer jacket and stuff like that. And it was this giant leap where
0:09:22 all of a sudden people no longer could tell what was AI generated and what wasn’t. I think we have
0:09:28 not seen that leap with video yet. And I think in 2025, we will see that leap where it gets so good
0:09:34 that, you know, people didn’t even realize it coming. And the pace at what this happened
0:09:39 is so mind-blowing to me, because like you were mentioning, very beginning of 2023, we were
0:09:45 looking at tools like model scope and zero scope, and they were generating like two second clips,
0:09:51 and they were like 280p resolution, right? Like, they were like 280p.
0:09:56 -The object is kind of jittering around. -Yeah, and like you can barely tell what they were
0:10:01 actually happening on the video. And that was like 23 months ago,
0:10:07 and now we’re getting what we get out of VO. Like, holy crap, like that was a massive leap in two years.
0:10:12 Yeah, and then the crazy thing is if you look at like public perception, people are like,
0:10:19 “Yeah, Sora sucks.” And like for me, it’s just like, “Oh my god, I was, you know,
0:10:25 one of the first people like tweeting a lot about AI video on X, and I remember the comments of,
0:10:29 “Oh, but it’ll be like 10 years before you get anything that’s even close to like a cinematic shot.”
0:10:32 -No, we’re there. -No, we’re there.
0:10:36 Yeah, if you just need like a two or three second like B-roll shot for something,
0:10:40 it might take a few re-rolls, but you can get it right now, you know?
0:10:44 Yeah, and the fact that there’s no like kind of reasoning model on top of that, like really
0:10:49 judging the output, and like as soon as you apply that, like you said, it’s going to get way better.
0:10:53 Now, one thing that with all of this is going to be interesting is the cost.
0:10:58 Like, okay, it can get way better when you reason about the output, like how much does that cost?
0:11:03 And we’ve already seen like with like, you know, okay, with AGI, they eventually want to make it
0:11:08 where everyone has access to it. But it seems likely that like people who have money are going
0:11:14 to get access to that way sooner, just because of the cost to run all of this. Like Pro is $200.
0:11:20 I think that the Chief Financial Officer of OpenAI, she said, “We’re open to experimenting with
0:11:25 things like $2,000 a month.” And I think you probably will see that, right? You’ll see like,
0:11:29 okay, there’ll always be a model that everyone can use. It’s like around like a 20 bucks a month
0:11:33 or whatever. You might have something slightly better, you know, or like, you know, significantly
0:11:39 better for like 200 bucks a month. And if you’re a professional, you might be like paying two grand.
0:11:43 Yeah, well, I mean, most of them need to have free models, right? You can use Claude for free,
0:11:48 I believe, right? You can use ChatGPT for free. You can use Grock for free now. That’s brand new.
0:11:54 Grock only became free like two weeks ago or something. Like Gemini, I’m pretty sure, yeah,
0:12:00 they have Gemini and Gemini Advanced. The Gemini Not Advanced is free, right? I think you could even
0:12:06 test the Gemini 2.0 right now for free inside of their AI studio, right? So there’s always going
0:12:10 to be like that free plan that’s like, anybody can access it, but it’s not going to quite be the
0:12:16 state-of-the-art model. And the more state-of-the-art, the further you go up that sort of tier of how
0:12:22 state-of-the-art it is, the more expensive it’s going to be. Yeah, I mean, so from using O1 Pro,
0:12:27 which in my opinion right now is the best by far, it takes a long time to give you answers back.
0:12:33 And so something I’ve been realizing is, while you’re waiting, like, what do you do? Because
0:12:37 sometimes it’s like taking like three to five minutes and it’s like, okay, you know, I go, what
0:12:40 do I do? It’s like, it’s like back in the day when you would code, it takes something would take a
0:12:45 long time to compile and you’re like, okay, what do I do while that’s compiling? I go and check
0:12:47 something. I mean, I still run into that when I’m trying to render my videos, right? It takes
0:12:52 10 minutes to render out a video that I make for YouTube or something. And I’m like, well,
0:12:58 my GPU is in use right now. So like, I can’t really do a whole lot with my computer. So I go
0:13:01 walk away and do something else, you know, right? Well, but in this case, you’re using their GPU,
0:13:07 so you can still use your computer. That’s true. And so, you know, it slightly reminds me of like,
0:13:11 so when I was younger, I think I told you this before, like probably 15, I made money playing
0:13:14 EverQuest. And there was a lot of people who made money playing EverQuest back in the day.
0:13:19 I was as far as I know, I was like a top 10 player on the game. I was like, I was a well-known
0:13:24 necromancer, dark elf necromancer on the game. And one of the strategies we would employ is, you
0:13:29 know, like if you went with a party in the game, and you were with a bunch of other people, if
0:13:33 something good dropped, you had to roll on it. And that kind of if you were doing it to make money,
0:13:37 that sucked to have to share it with those people. And you might get unlucky and you might not get
0:13:43 this stuff. So what became very common was to multi box, which was basically where you would
0:13:48 be running multiple computers, multiple accounts, and then you get a huge financial benefit from
0:13:52 doing so. And you would also make alliances with other people who were multi-boxing,
0:13:58 right? So like maybe if it was like, okay, it’s hard to run six computers, but like the, you know,
0:14:02 if we could put five people, maybe one person runs three, one runs two, and then that’s a better
0:14:08 split than six random people that I’m sharing with. And so we would do that. And you would make
0:14:12 a lot more money from doing that. And so I do wonder like, okay, in this new world, where
0:14:18 if you can spend more money, you get way better outputs, but you’re still tied to like the one
0:14:24 account, will we see like some really clever people start multi-accounting AI?
0:14:29 Yeah, yeah. Well, you know what you’re describing here too, it is also, and I don’t want to get
0:14:34 two in the weeds on this, but it also feels like a fairly decent use for like blockchain tech,
0:14:40 right? Because you can distribute the compute, right? You can have other people like you ask
0:14:45 O3 a question and it goes off and it uses somebody else’s account and they get sort of tokens back
0:14:50 for allowing to use the account kind of thing. I can see something like that where it’s almost
0:14:56 like this distributed network of you’re using other people’s O3 accounts to generate what you
0:15:00 want to generate, because they’re not currently using their account. And whenever their account is
0:15:07 inactive, you use it and you pay some sort of cryptocurrency to use it and that’s their form
0:15:12 of mining, you know? Yeah, that sounds good conceptually. I’ve worked with blockchain stuff
0:15:20 in the past. It’s often harder than it sounds, but I guess I’ll ask O3 or O4 for advice on
0:15:25 if that’s the best way to go and we’ll… Yeah, maybe AI will develop that software for us instead
0:15:31 of a human actually developing it. Do you think O3 is AGI? Like from my definition of AGI, which I
0:15:36 guess is like, can do most things humans could do. If you didn’t put all this stuff into the models
0:15:41 where they tell you that they’re AI, I’m not a human. If you didn’t have any of that in there,
0:15:47 I think you could trick most people into believing that this was a human doing stuff for you. I
0:15:54 really do. And I already thought like 01 Pro was close to that. So like based on the benchmarks
0:15:59 they’re showing for O3 where it’s past what the ARC AGI benchmark where it’s got like a
0:16:08 87.5 and the average human taking it gets like an 85. That’s for me pretty close to AGI.
0:16:13 I could call that the beginning, like a first version of AGI to me.
0:16:17 Yeah. Well, I looked it up on a live stream earlier today where we’re talking about O3
0:16:23 and that ARC AGI benchmark, the average human gets about a 76% on it.
0:16:31 Oh, really? Okay. And so if we look at the O3 model that just came out, the O3 low model,
0:16:36 what do they call it? There’s like a smaller model that used less training and then like a
0:16:43 larger model. The smaller model got a 76%. So the smaller model hit the average that
0:16:49 a normal human would hit. The larger model, the one that used the insane amounts of compute,
0:16:55 got like an 86% or something like that. So smarter than the average human.
0:17:01 Yeah, which is wild. This is it looking at complicated logical puzzles and things like
0:17:05 that and solving them. For anyone who’s saying that these models do not reason and they’re just
0:17:12 like parrots or whatever, no. There is some reasoning going on here and this is the breakthrough.
0:17:16 Like this is the fact that you now have like the large language model, but you’ve got this computing,
0:17:21 you know, test time computer, whatever they’re calling it. Like that’s the big unlock where
0:17:25 like you combine the two. It’s like the two sides of a brain. Well, I just pulled up this chart.
0:17:33 This is from the ARC prize website here about OpenAI’s O3 model. So their low model scored about
0:17:38 what an average human would score. But one thing that I find super interesting about this is the
0:17:44 cost per task down here, right? So it’s like this low model, you know, this is a logarithmic scale,
0:17:49 right? So it goes from $1 to $10 to $100 to $1,000. So if you were to like chart this,
0:17:55 it would have like a sort of curving up slope, right? Yeah. So this isn’t like
0:18:04 $11 for 76%. This is like, this is probably something like $30 per task right here, if I
0:18:11 had to guess, right? And if you look at the O3 high model, this is $1,000. If there was another line
0:18:16 drawn right here on the chart, that would be at $10,000 following the, you know, the same sort of
0:18:24 logarithmic scale. So this 88% is actually halfway between $1,000 and $10,000. So in the range of
0:18:32 somewhere around $5,000 per task to be able to get that O3 high model actually working,
0:18:38 yeah, very, very, very expensive right now to be able to use this model as of right now. I mean,
0:18:43 like my understanding, like the amazing thing about this is at least from like what people are
0:18:48 like kind of, you know, inferring from like tweets from people from OpenAI is that O3,
0:18:54 which they skipped O2 because of some trademark stuff apparently. Well, there’s O2 Arena in London,
0:18:58 right? That’s, there’s a big company called O2 that owns that arena licensing and yeah. Yeah.
0:19:05 So it seems that O3 is literally just O1 with more compute thrown at it. And if that’s the case,
0:19:08 I mean, yeah, that’s what you would expect. The cost to go up dramatically and you get,
0:19:11 it’s amazing that you can just get more performance from it though. Like you do get
0:19:17 more intelligence, the more compute you throw at it, which means that we are now in a world where
0:19:21 like all the stuff we’ve talked about in the past, we’re like, okay, race for more compute, race for
0:19:28 more energy. Yeah. Those things are majorly in play now because we’ve now proven that you probably
0:19:35 can get to beyond AGI just with more compute. That’s been proven now. And so that’s, that’s a
0:19:40 big revelation like, okay, we thought that might be the case. It appears to be the case now. It is
0:19:46 impressive. Like in this, it shows on, so here’s a benchmark in competition math, which is like
0:19:50 the kind of math that like we have no idea how to do. This is like people who are like
0:19:54 PhDs in math to solve incredibly challenging. You know, this is not just like basic algebra.
0:20:01 And so like that, it’s, you know, it’s frontier math that GPT-40 scored like a 13.4, but like
0:20:08 03 got a 96.7. So I mean, this is an incredibly high score. This is like top of the top smartest
0:20:14 people in the world level. And I found this fascinating, which this is, I think this is,
0:20:18 I’m not sure if this is like an official like who’s came up with these IQ rankings for the AI,
0:20:25 but some people ran some numbers and it was shown that like, you know, GPT-40 scored like an estimated
0:20:31 115 IQ, which is considered like a somewhat smart person, not like a super smart, like a
0:20:37 relatively smart person. 130 or over is considered a smart board, you know, maybe 140 is like
0:20:43 plus is genius. Einstein’s estimated to be like around 160 IQ. So the average IQ range for adults
0:20:51 is between 90 and 109, with 100 being the theoretical average. So 100 is about what you’d
0:20:59 expect average to be. Yeah, yeah. Yeah, in America, average is about 100. So it’s estimated for 03,
0:21:05 157. And like I said, you know, it’s estimated, it’s, you know, IQ, it’s highly debated, like
0:21:10 how useful it is. But, you know, Einstein was estimated to be 160. And so this is like pulling
0:21:16 in like close to 157. So that means we’re possibly, you know, in the ballpark where like this will
0:21:22 actually be helping us solve frontier math problems, like problems in physics and just basic understanding
0:21:26 of the universe that we live in. This may be approaching that, like by the time we get to 04,
0:21:33 maybe that is already able to do that. Also, this appears to still be not a larger model that they’ve
0:21:39 trained. This still seems to 01 and 03 are not dramatically larger models and trained. They’ve
0:21:44 just came up with this new paradigm of like applying like test time compute on top of the
0:21:50 existing models. So if that’s the case, well, then maybe what was called Orion or GPT-5 or whatever,
0:21:58 maybe that’s still in the works. And so maybe by the end of next year, not only will we possibly
0:22:03 get 03, or maybe it’ll be like a slightly less compute intensive 03. So maybe it’s like three
0:22:09 times better than 01, you know, Pro or something. Just to clarify, what you’re saying is like,
0:22:17 they’re still using like a GPT-4 level model, but they’re just throwing more compute during the sort
0:22:22 of inference phase, right? It’s already, it’s the same sort of trained model that they trained with
0:22:27 maybe GPT-4, maybe it’s some sort of like 4.5 or something, right? Like maybe it’s a slightly
0:22:33 newer model. But the inference, when you actually give it a prompt and ask a question, it’s throwing
0:22:38 a ton of compute at it to like process the question and then double check itself and then
0:22:43 double check itself and then double check itself until it’s fairly certain that it gave like the
0:22:49 proper output that you’re looking for, right? But it’s still essentially using like a GPT-4,
0:22:54 maybe a slightly better model. Yeah, yeah, that’s, that’s what people believe. And that kind of,
0:22:58 that makes sense to me, honestly. Like if they have discovered this new paradigm of like,
0:23:02 yeah, not only do you get the result from the LLM, but then you’d like reason on top of that,
0:23:07 you could see, you know, it appears you can get like major improvements just from that.
0:23:12 And if that’s the case, it could be that, yes, they do have a, the model that they’ve been training
0:23:16 for a while that’s on a, you know, has massively more amounts of data that’s been trained upon,
0:23:21 that could still be in the works. And so when Sam Altman is saying that we’re going to see
0:23:28 huge increases in 2025, that could be what it is. They still have the new model coming. And then
0:23:34 they’re just going to slap this new reasoning model on top of that. Yeah, yeah, it’s like GPT-5
0:23:40 combined with like the ’03 reasoning model. And like all of a sudden ’03 gets even further
0:23:45 up the benchmarks and does even better. And the only thing that changed was the underlying model
0:23:49 before the reasoning portion happened. Yeah, because in its baseline is way higher that it’s
0:23:55 a reasoning upon, right? Right, right, right, right. So you could in theory see dramatic increases
0:23:59 beyond what we’re seeing in these benchmarks. I mean, so you might see like, okay, yeah,
0:24:06 we get like a 200 IQ plus model by the end of next year, which obviously dramatically changes
0:24:11 things because like, you know, I’ve been testing ’01 Pro and people are kind of now saying that maybe
0:24:17 ’01 Pro is maybe, it’s basically in between ’01 and ’03. It’s like basically like when they were,
0:24:21 when they were testing ’01. It’s the ’02 that we’re not allowed to talk about. Yeah, it’s basically
0:24:26 like when you take ’01 and put more compute towards it, what happens, it’s some first version of that
0:24:32 test. Right, right. And it seems like that, like I think they showed, I forgot where the benchmark
0:24:37 was, but I think somebody did show that like there was a major difference between ’01 and ’01 Pro
0:24:43 on that ARC AGI test. And I’ve been seeing that too, like when you throw, like if you throw a lot
0:24:50 of code at ’01 Pro and tell it to reason about that code and to help rearchitect that code,
0:24:57 it can do it where ’01 fails. Because one of the things that I keep on hearing come up is like,
0:25:02 it’s a huge advancement and it’s really cool, but what are the actual practical use cases? And I
0:25:08 think the most obvious practical use case is probably coding, right? I think that’s going to be,
0:25:13 like as far as like a general consumer, the most actionable use case. But I do think
0:25:19 there’s a lot of other use cases for something like ’03 beyond that. They’re just not necessarily
0:25:23 use cases that the average general consumer is going to care that much about, right? That they’re
0:25:29 not going to try to tackle themselves using ’03. Things like, you know, trying to solve climate
0:25:36 problems or, you know, trying to cure diseases or things like that. But I just wanted to hear,
0:25:41 like from your perspective, like outside of coding, let’s just set coding aside, outside of coding,
0:25:48 what do you see like ’03 like really helping the world accomplish? I think that when you start
0:25:54 having agents powered by ’03, that’s when you’ll see like real, like regular people getting benefits
0:25:59 from these reasoning models. I do think they’re like, okay, if Orion comes out and it is what’s
0:26:04 powering like an ’04 or whatever, maybe there is like a standalone non-reasoning model too,
0:26:09 or they have some kind of smart router that knows, okay, I can respond super fast to this basic
0:26:15 query. And if it’s a hard question, let’s spend some time thinking about it. I hope that’s coming.
0:26:19 Because right now, like the ’03, like the ’01 Pro, like the average person, if they used it,
0:26:24 and they’re asking it like how to bake something, you know, or whatever they’re asking it,
0:26:28 like they’re going to have a bad experience. Like it’s going to be like, oh, this is like,
0:26:32 you can get just as good of a response out of GPT-4, basically. Yeah, or like maybe,
0:26:35 maybe ’01 Pro would be like slightly better, but like it would take so much longer that
0:26:40 like the perception would be that it was not good for sure. And so I think that the magic’s
0:26:45 going to be when it start powering the agents, you know, because then you can like, okay, I hate
0:26:50 dealing with email, let’s just have the agent handle all my email. I don’t actually want to do
0:26:55 email anymore. That thought just makes me laugh because it’s like, what happens when everybody’s
0:27:00 doing that? It’s just like, nobody’s writing or checking their emails anymore, it’s just agents
0:27:06 doing it all for them. I think that’s where we’re heading though. Like, you know, it’ll be super
0:27:12 rare that you get like a handwritten email from somebody, you know. It’s like, not a handwritten
0:27:17 letter, but like a handwritten email. Right, right. A hand typed email. A hand typed, yeah,
0:27:21 hand typed email will be a rare thing, probably in like two or three years. I think that’s where
0:27:24 we’re headed because like everybody hates email. There’s all these things that we do in our daily
0:27:30 lives that like almost everybody hates. And there haven’t been major improvement to these systems.
0:27:35 So I think the systems will continue to exist, but we’ll start to have agents actually do those
0:27:40 things for us that we don’t want to do. And so I think that’ll probably, I think in 2025, we will
0:27:45 see O3 come out and it will be what makes agents really work. And that’s when people see a change
0:27:50 in their daily lives. Once we have robots and like, reasing models are going to be fundamental to
0:27:54 having robots that can actually do things in your house that you feel safe about.
0:28:00 Yeah. And where I think like the sort of like O1 and O3 are really powerful for agents is the
0:28:04 fact that they’ll essentially double check themselves, right? Like going back to the coding
0:28:10 thing, right? It can write code for you and then look at your code and look to see if it made any
0:28:14 mistakes and then fix the code and then double check to see if it made any mistakes and then go,
0:28:18 okay, I’m pretty sure the code’s right now and give it back to you. Right. So the fact that it’s
0:28:24 sort of like constantly double checking its work is one of the biggest sort of jumps that I think
0:28:29 this O1 and O3 are creating is that, you know, the reason it’s taking so long is it’s almost
0:28:33 like kind of double checking its work and then triple checking and then quadruple checking and
0:28:37 the ones that are using insane amounts of compute, it’s like, all right, we just double check this
0:28:42 a hundred times. We’re sure this is the right answer, you know? Yeah. It’s like multiple AIs talking
0:28:46 to one another and then like finding the best, you know, and double checking everything to make
0:28:51 sure it’s all accurate. Like it hallucinates way less. Like I’ve been playing around. I’ll show
0:28:56 you something later. Like a very basic game demo. Don’t expect this is not a real engine five game
0:29:00 or anything like that. It’s very, very basic. But I’ve been playing around with creating a game
0:29:05 demo just because I got so excited when I tested O1 Pro and realized how good it was at like actually
0:29:09 architecting systems with code. But you find like it’s amazing at some things, but there’s still
0:29:14 some things that Claude’s good at and there’s some things that even like Jim and I 2.0 is good at.
0:29:19 Right. But I do find that O1 Pro is the best at architecting hard systems, like actually like
0:29:24 architecting the code, making major changes or refactors to the code and in terms of like
0:29:29 not hallucinating about helping you solve something. The more and more I tested Claude,
0:29:33 like it’d be very good at some things. Like some things like in terms of like modifying my UI or
0:29:37 stuff like that. Claude was the best for some reason, but also it would just hallucinate your
0:29:41 really weird things. Like I would like show it a bunch of code and say, “Hey, what do you think is
0:29:46 the problem with this? I’ve got a bug. Here’s the bug.” And I was like, “Oh, you need to change this.”
0:29:51 And now like I look at my code and that’s already with my code. It’s like you need to
0:29:55 change this to this. And I look at my code and it’s like, “What you’re telling me to change it to
0:30:00 is my code.” Yeah, I’ve seen that in cursor before. I’ve seen that because I have my cursor set up
0:30:06 for Claude. Was it Sonnet, right? I have it set up for Sonnet. And I’ve seen that before where it’s
0:30:11 like, “Oh, I spotted the problem. You need to change this to this.” And I’m like, “They’re the same thing.
0:30:16 You didn’t change anything.” I know. And it’s like, so that’s a crazy hallucination where it’s
0:30:19 literally like what you just pasted to it. You’d imagine that what you pasted was something else
0:30:26 and that what you pasted was a solution. Yeah, yeah. And I have not had O1 Proc ever do that.
0:30:33 So that’s a major change where like you can like O1 Proc likely catch it. It might still hallucinate,
0:30:37 but then it catches its hallucination before getting back to you, right? Because there’s a
0:30:41 lot of stuff that’s going on with the O1 model behind the scenes, right? That they’re not showing
0:30:47 you the entire thinking. So it’s very possible that it is hallucinating, then double checking
0:30:53 what it just responded with, notice it hallucinated, and then fixed itself before responding to you.
0:30:58 Yeah. Yeah. So that’s going to make these models way more reliable. Like anything that needs to be
0:31:02 reliable, they’re going to get reliable. Like because probably the more compute you throw at it,
0:31:06 the more reliable it will get. The more times it thinks about the answer before it gives it to you,
0:31:13 probably maybe it’s not going to be 100%, but we’ll get like all approach like 99.9% accuracy
0:31:16 with these models probably the next year with reasoning. So I guess the big question is like,
0:31:21 okay, for 2025, like with like models, like all the models are getting really good. And like now
0:31:25 Google’s like teasing, they’ve got like a reasoning model and they’re kind of making it sound like
0:31:30 it’s not a big deal. And they’re even saying that they’re probably going to beat OpenAI in terms of
0:31:36 coding next year. I do wonder if that’s true or if like Google’s overestimating or underestimating
0:31:41 what OpenAI has accomplished. So like OpenAI is saying like, it’s not as simple as you think,
0:31:48 like we’ve discovered some things. The thing about OpenAI though, is that I do feel like a lot of
0:31:53 the other companies could potentially catch up fairly quickly, just due to the amount of brain
0:31:58 drain they have at that company, right? Like how many people are helping develop this stuff
0:32:02 and then walking away from OpenAI like pretty close to the time they actually announce it,
0:32:07 right? We’re seeing it constantly. Yeah, maybe, but I mean, I don’t know. That could just be a
0:32:11 narrative too though, because like that’s always happened in Silicon Valley. Like talent moves
0:32:15 around because like, if you’re at a company for a while, what happens is you get some stock there
0:32:19 and the people at other companies just offer you ridiculous amounts of money. And so then it makes
0:32:25 sense just to like keep your stock, go get more stock. This has always happened in Silicon Valley.
0:32:29 Like people usually move around every two years. Like people rarely stay at a company more than
0:32:34 two years in Silicon Valley. I do wonder, like that could be the case. You could be right,
0:32:37 but it also feels to me like maybe it’s something that’s kind of been exaggerated,
0:32:41 just because it’s great for tweets and things like that. And it’s just like more of what’s
0:32:46 always happened in Silicon Valley. And there’s constantly, like who’s to say they’re not getting
0:32:52 higher quality talent coming in as other talent leaves, because now they can attract the best of
0:32:58 the best. Whereas when they started a company, you know, maybe they, you know, maybe Carpathia is
0:33:01 amazing, but maybe he’s not the best at creating this despite having a huge persona online.
0:33:07 Yeah. And so in Silicon Valley, that’s always like resulted in like issues with founders and
0:33:11 people leaving and things like that. I really think it’s going to come down to like three major
0:33:16 players. I think really it’s going to be Google, Anthropic and OpenAI. And the reason I think
0:33:20 Anthropic is in the mix is because they’re just, they’re getting infinite amounts of cash from
0:33:25 Amazon, right? So it’s like, you’ve got, you’ve got Anthropic, who’s essentially owned by Amazon
0:33:30 at this point, OpenAI, who’s essentially owned by Microsoft at this point. And then Google and
0:33:35 DeepMind, which, you know, all under the same umbrella. I think you’re really missing XAI.
0:33:40 Oh, you’re right. You’re right. I am mixing. I do think they are going to, they are going to,
0:33:43 you’re right. I am missing that one as well. I do think they’re going to do a lot, especially
0:33:48 with, you know, Elon building like the largest server farm on the planet and whatnot. So yeah,
0:33:52 yeah. And it’s going to be interesting because it sounds like Elon is thinking he’s going to
0:33:59 be able to create the largest model. But if OpenAI has discovered that like, okay,
0:34:03 having a bigger model is good and does improve something, but reasoning is actually the more
0:34:08 important part. You know, is Elon going to be able to like kind of slightly,
0:34:11 is he going to be able to catch up and figure out how they did it? Or does OpenAI really
0:34:15 have some kind of secret sauce there that you don’t take people time to figure out?
0:34:22 I do often underestimate and take for granted XAI because I feel like they’re still trying to
0:34:25 catch up, right? Like the current models that we have available to us, which are,
0:34:30 you know, GROC inside of X and then they have their new like Aurora image model.
0:34:36 Neither of like their large language model or their image model feel as good as what else is
0:34:41 available. But also I’m not counting them out because I know how much obviously Elon, I believe
0:34:46 is still the richest person on the planet. If last time I checked, right? He’s got $6 billion
0:34:53 in investment. He’s building like the largest GPU cluster on the planet, right? We’re going to see
0:34:58 them make some waves and forgot to mention, he’s best friends with Donald Trump at this point.
0:35:04 So, you know, he’s got government inroads as well. So I think, yeah, I do think we are going to see
0:35:08 a lot more out of XAI. It’s just, I always forget about them because I don’t personally use them on
0:35:11 a daily basis, you know? Right. Yeah. They don’t have the consumer adoption yet, but like people
0:35:16 are showing that like that has been going up. I think I saw a chart showing that it was going
0:35:21 up a lot in Japan, which makes sense because in Twitter, X is huge in Japan. But like I saw like
0:35:25 searches for GROC as well, where it’s like going up dramatically in Japan. I think Elon last year,
0:35:31 that is like almost like it was like a global trend. He showed that chart and people were like,
0:35:35 that’s Japan, but you know, but Japan’s very important. So like a top three market. And so I
0:35:39 am curious to see how that plays out. But yeah, I think we have at least four major players
0:35:45 going after it. And I think all of them are going to make major progress in 2025. So,
0:35:50 which will make all of them make progress faster. They’re all, you know, people have said like,
0:35:56 oh, open AI had to show 03 on the final day of the 12 days of open AI, because Google has been so
0:36:00 freaking impressive with their announcements over the last month. Do you think that they had something
0:36:05 else intended for the 12th day? And they pivoted to show off 03 because of Google like sort of
0:36:09 matching them day for day with something equally as big, if not bigger? I think it’s, I think it’s
0:36:13 possible because you know, it was also you started seeing, you know, we’ve had Logan on the podcast
0:36:17 before and he was starting to like tweet out some stuff, you know, kind of kind of hinting
0:36:23 that like Google’s not behind, maybe they’re maybe they’re ahead. And then like when 03 was
0:36:27 the benchmarks were shown for 03, like all these open AI people started sharing memes like, you
0:36:32 know, opening AI was never behind. You thought they were behind. Everybody’s saying they’re behind.
0:36:40 What are we behind? You know, the most used model and with like the stuff they have internally,
0:36:45 steal the best. Well, they needed something like that to really sort of get people excited again,
0:36:49 because I feel like during the 12 days, there was a couple cool announcements, but most of them were
0:36:57 kind of like most people, it doesn’t impact them too much. Like I thought the 100, 1800 chat GPT
0:37:05 was sort of like a gimmicky novelty sort of thing. We got Santa voice. Yeah, the Santa voice.
0:37:10 I mean, I think the big moves they made were putting out Sora, but then Vio came out like
0:37:16 two days later and sort of knocked Sora off the top of like being the best AI video tool at the
0:37:21 moment. Sora turbo, right? The other sort of might still be a lot better. I don’t know. But then
0:37:26 they also announced the vision inside of the advanced voice mode. And I think that was probably
0:37:33 the most impactful thing that they did was the advanced voice mode with vision. We also got the
0:37:40 $200 a month plan, which gave us access to the GPT or the 01 Pro. Yeah, I mean, for me, 01 Pro is
0:37:45 the biggest one. But like I said, like most people, number one, they’re not going to pay $200 a month.
0:37:49 And number two, even if they did, you can see that everybody’s getting very different results.
0:37:54 Like it depends on what you’re trying to use this for. The card coding stuff or science,
0:37:58 people are blown away. If they’re trying to use it like they use just chat GPT on a regular daily
0:38:05 basis, they’re like, this sucks. Yeah. Yeah. Yeah. Yeah. Yeah. I’m curious. I want to see this this
0:38:09 game that you’ve been talking about. It’s super, it’s super simple. So don’t expect much, but I’m
0:38:18 not a game designer at all. I’m expecting Baldur’s Gate 4 meets Elden Ring. Baldur’s Gate meets Elden
0:38:25 Ring. Let’s see it. No, no, it’s not that. But you’ve been using 01 to develop this or 01 Pro.
0:38:31 I’ve been using 01 Pro. I’ve been using Claude. I’ve been using Gemini 2.0. I’ve been using Sora.
0:38:38 Are you using an IDE like a cursor or visual studio? I’m using cursor, but I feel like I’m
0:38:42 kind of a newbie using cursors, to be quite honest. Now that they added, you can put 01
0:38:47 into cursor. I do want to try it again, because that probably better than the outputs I was getting
0:38:52 from Claude. I feel like right now, 01 is like slightly better than Sonnet. 01 Pro is significantly
0:38:56 better, but way slower. But I’ve been using everything. That’s why I was saying about the whole
0:39:06 multi-boxing thing is multi-accounting. I think the wizards of AI, they are going to be putting
0:39:10 together all these different tools and maybe multiple LLMs based on what they’re good at.
0:39:15 It’s happening already. I think you’re going to see ChatGPT give you an output,
0:39:19 and then it spits it over to Gemini to double check it, and then it spits it to Claude to double
0:39:24 check it. It goes round robin through all these various LLMs and then spits back out
0:39:31 like a sort of aggregated response based on what all the LLMs said. This is very basic,
0:39:38 so don’t judge me. This is all super ugly, like you train a character, you go into a little town,
0:39:43 which is like all mid-journey generated. It looks sick, dude. It does look really good.
0:39:51 And dude, I’ve had mid-journey. I use mid-journey for this. The special effects, I asked open01pro
0:39:57 what to do, and they’re just like, “Okay, let’s put some snow here. Okay, cool. Let’s do that.”
0:40:01 So it’s like a static image, but then some coated in snow over the front.
0:40:05 Of course, you would make it look better. There’s this one game called Deepest Dungeon,
0:40:09 where they do something like this, where you have a static image of a town,
0:40:13 and when you highlight over them, they light up and stuff like that.
0:40:16 So I’ll eventually do some kind of special effects where when you go over it, it lights up.
0:40:21 Here, I go build my party. I have not worked on the UI, so all this is just super ugly,
0:40:27 but okay, recruit my team, random. So how many hours do you have into this right now?
0:40:32 It’s a decent amount, but not a crazy amount. I have a hard time estimating,
0:40:35 because it’s been the kind of thing where I’ll do it while I’m doing other stuff.
0:40:40 I’ll have open01pro running, and I’ll just be totally doing other stuff. And I’ll be like,
0:40:44 “Hey, improve this.” Not only like improve this, but sometimes it’s like,
0:40:48 “What should we improve?” So like, yeah, you hear me some kind of like a world map,
0:40:53 you know, it’s all not that great for now, but like you’d go in Mario style.
0:40:57 We have different nodes that you go through to get to the final boss.
0:41:00 And then what I’ve kind of done gameplay-wise is I was like,
0:41:04 “Well, what’s something simple I could actually implement without like making,
0:41:07 you know, Baldur’s Gate 3, which is super complicated.”
0:41:12 And I was like, “Well, I used to play this game called Puzzle Dragon, actually Taizosan’s game,
0:41:15 where it’s like basically like a match 3 game with an RPG element.”
0:41:19 But I always felt like, “Oh, but it doesn’t have like a real RPG elements in it. Like,
0:41:22 they’re very, very light.” I was like, “What are you made of this?” So like,
0:41:27 in between Baldur’s Gate, like more RPG elements, and like a basic puzzle,
0:41:30 match 3 that everyone understands, and some people love, people hate.
0:41:33 And so I just created something like that. And then you have like,
0:41:36 you basically have like a party here, like your party with like health bars,
0:41:39 it looks really good. All this is mid-journey art.
0:41:44 This is all mid-journey. Yeah. I used, I started to use scenario a little bit,
0:41:48 but this is mostly mid-journey. And, you know, and it’s just like a very basic match game where
0:41:53 you charge stuff up, you got different abilities. I hadn’t worked on the special effects at all.
0:41:56 I literally just asked 01pro, like, “Put some special effects in there.”
0:41:59 It looks really good. It’s like a, like a-
0:42:02 All the special effects, it just came up with. I didn’t, I didn’t do any of that yet.
0:42:08 It’s almost like a Candy Crush kind of concept in the actual like puzzle game almost,
0:42:12 but then you’ve got a whole bunch of like fantasy RPG elements also.
0:42:15 Yeah, yeah. I’m not going to just play the entire game logic.
0:42:18 You’re getting hooked on it. You’re like, “I’m playing now. I’m doing good.”
0:42:23 This is not work now. Just play games. You know,
0:42:26 then you like have different choices for going to different rooms.
0:42:30 I probably would have the AI, I’d probably like actually like generate the dungeon,
0:42:31 so it’s not the same every time.
0:42:32 Yeah, yeah.
0:42:35 And then you have different, you know, different stuff you buy.
0:42:40 And so, to be honest with you, I’m kind of surprised how good it is.
0:42:44 It looks really good. I mean, everything, I just, I look pretty solid.
0:42:50 Yeah, dude. Like considering like, I’ve literally, that’s like 90% 01pro
0:42:54 with little bits of Gemini 2.0 and Claude for different uses.
0:42:58 Well, you lowered my expectations where you were like, “Yeah, don’t expect too much.”
0:43:01 And then what you showed me is better than I expected.
0:43:03 So I think it’s been kind of good for me to do this too.
0:43:05 Because, you know, I think I told you off camera.
0:43:09 Like I’ve been kind of decide like, what else should I work on beyond the podcast?
0:43:13 Like, should I be building another startup or should I be doing a YouTube?
0:43:17 And I kind of decided like, well, I think I just want to be playing more of these AI tools.
0:43:21 I think it’ll just be more useful for the podcast for me to actually have hands-on experience.
0:43:24 Maybe these things turn into real projects. Maybe they don’t, but they’re like.
0:43:27 I can see it spinning off into a real like game studio.
0:43:28 And you’re actually selling them.
0:43:32 It’s been really useful for me though, to actually understand how good, like,
0:43:35 what are these things good at currently? What are the limitations?
0:43:43 What would be really cool is if a tool like a cursor sort of knew what each model was good at,
0:43:47 and then would send whatever your prompt is to the best model for that specific thing.
0:43:50 Or you could create your own logic that’d be kind of cool too.
0:43:53 Like, okay, there’s suggestions, but like you can kind of like find, you can like,
0:43:55 yeah, when I’m doing UI, I want to use this.
0:44:01 Yeah. I can see cursor building in something like that, where it kind of,
0:44:05 it’s already model agnostic anyway. You just pick which model you want to use.
0:44:08 It doesn’t seem like too hard of an extra step to basically say,
0:44:11 if the question is related to this, send it to this model.
0:44:13 If it’s related to this, send it to this model.
0:44:16 Yeah. So this, you know, doing this like little test like demo project,
0:44:19 like maybe realize like, you know, maybe a year or two ago, I was like,
0:44:24 oh, you’re going to see like one or two people startups that hasn’t super happened yet.
0:44:28 But I feel like after like actually trying the newest stuff and like realizing how much
0:44:31 better it’s about to get, I think we probably are going to get there where you’re going to see
0:44:35 like one or two people, maybe five, you know, like a small group of people
0:44:38 creating massive projects stuff that used to would have taken,
0:44:43 you know, hundred people, thousand people, you’re going to see small teams of people
0:44:46 with the concentrated focus being able to crank out amazing projects.
0:44:52 And the fascinating thing is too, is they may be able to do multiple projects as well.
0:44:56 Like I’ve been finding like, I can totally work on other stuff while I’m doing this,
0:44:58 because it’s more like, it’s almost more like I’m like the director.
0:45:01 Yeah, yeah, yeah, yeah. You’re basically going and telling your team members,
0:45:04 all right, we need to create this next and they go and do it.
0:45:04 Yeah.
0:45:06 And you’re just steering the ship.
0:45:07 You’re not doing the things.
0:45:12 It is fun. And I think most people don’t know how to tie this stuff together
0:45:16 as well as I do. So I think that’s going to get easier and more people will know how to do that.
0:45:20 I totally agree. I mean, we were talking about this before hitting record.
0:45:23 I think a lot of SaaS companies are going to be in trouble.
0:45:28 I think it’s, you know, a lot of companies are going to be able to just like build
0:45:32 tailor-made tools for their needs without needing to go and, you know,
0:45:36 pay for a SaaS company to do the thing for them, right?
0:45:38 I’m already finding myself doing that when I have like a need.
0:45:41 Like I was talking on a previous episode of like,
0:45:44 I want something that no matter what file type I put into it,
0:45:46 it converts it to a JPEG.
0:45:51 So if I throw in a PNG, converts it to a JPEG, a WebP JPEG, AVIF file, JPEG.
0:45:52 I don’t care what file type it is.
0:45:55 If I throw it, even if I throw in a JPEG,
0:45:58 I want it to output it as a JPEG in the selected folder
0:46:02 so that I can bulk throw images in there and convert them to JPEGs.
0:46:04 I guarantee there’s already tools out there that do it,
0:46:08 but it took me three minutes to build that using cursor and Claude.
0:46:11 So I just built it real quick and, you know,
0:46:17 I don’t have to go and use some either insanely ad-ridden website for it
0:46:19 or pay for some SaaS to do it.
0:46:22 I just have a tool that does it on my desktop now
0:46:23 and it works really, really well.
0:46:26 And I think you’re going to see more and more and more of that
0:46:28 where people are just like, I need this problem solved.
0:46:32 I’m going to go and just code up a solution real quick
0:46:36 because all it is is a prompt and the prompt will code up a solution, you know?
0:46:37 Yeah, it’ll get faster and faster too.
0:46:40 Right? Like eventually this will be almost instant
0:46:43 where within a second you’ve got that output of what you wanted.
0:46:45 A tool that does what you want.
0:46:48 I think in the ’03 mini demonstration they showed
0:46:50 where like it basically set up its own server
0:46:52 and then it set up all this different stuff and it’s like…
0:46:55 It set up its own server and created…
0:46:57 This came up with a benchmark for itself
0:46:59 and then set up a server to run that or is it something like this?
0:47:00 I built a benchmark.
0:47:03 Look, I’m really good because I beat my own benchmark that I just built to test.
0:47:07 But still, it is fascinating that it’s able to do these kind of things.
0:47:10 Right? It’s mind-blowing.
0:47:11 Yeah, for sure.
0:47:14 Yeah, I mean, I think it’s exciting times right now.
0:47:17 I think 2025 is going to be even more exciting.
0:47:18 So a lot of good stuff.
0:47:22 I think this is probably a good spot to wrap this one up,
0:47:24 our first episode of the year.
0:47:28 Thank you so much for tuning in for our show in 2024
0:47:31 and hopefully you’ll be with us for 2025.
0:47:34 Make sure you subscribe wherever you listen to podcasts.
0:47:38 We prefer YouTube because we try to make this visual and show stuff on our screen
0:47:41 and show off what we’re talking about a little bit.
0:47:45 But if you prefer audio, you can also find us wherever you listen to podcasts.
0:47:47 So thanks so much.
0:47:50 Happy New Year to everybody and we’ll see you in the next one.
0:47:51 Yeah, happy New Year’s.
0:48:09 [Music]