1 00:00:00,000 --> 00:00:04,640 I'm asking our employees what they think of A.I. as a company both in the tech and this 2 00:00:04,640 --> 00:00:08,560 crave space. I'm sure there are takes that people want to share. Uncut versions will be 3 00:00:08,560 --> 00:00:14,240 available below, so let's get into it. I am here with Jonathan Horst. Emily Young. James Shrype. 4 00:00:14,240 --> 00:00:18,720 Do I get to hold this at any point? Do I have to keep leaning in? Yeah, I'm realizing. Jake Bell. 5 00:00:18,720 --> 00:00:25,360 Dan, hey, uh, Andy. Nicole. Ploof. And what do you do for the company? Right. Good. What's your 6 00:00:25,360 --> 00:00:30,960 thought in the current state of A.I.? I've only used ChatGPT before. I think for like day-to-day, 7 00:00:30,960 --> 00:00:36,880 like if I don't want to deal with like a customer representative, like, hey, I want to get refund 8 00:00:36,880 --> 00:00:42,880 off this. I was like, hey, ChatGPT, write me an email about getting a refund off this. Over-hyped. 9 00:00:42,880 --> 00:00:47,120 We're thinking of it as like it's going to just do the thing we want it to do when it's probably 10 00:00:47,120 --> 00:00:52,080 just going to be yet another tool in the toolbox. It's pretty interesting. There's a lot of cool 11 00:00:52,080 --> 00:00:57,520 stuff going on and I'm excited to see it kind of develop over the next, uh, I guess a year or 12 00:00:57,520 --> 00:01:02,720 a few months even. We don't have it? No, it's not. What we refer to as A.I. right now, have you 13 00:01:02,720 --> 00:01:08,160 ever played Mass Effect? So in the lore of Mass Effect, A.I.s are like, you don't do that. So 14 00:01:08,160 --> 00:01:13,200 instead what they have is what's called a V.I. of virtual intelligence. That is a lot like what we 15 00:01:13,200 --> 00:01:18,080 consider to be like ChatGPT and all that kind of stuff right now. It's able to answer questions. 16 00:01:18,080 --> 00:01:23,440 It's able to kind of like carry a conversation. It's able to do a lot of things, but it's not 17 00:01:23,440 --> 00:01:27,360 able to think. So it's just learning patterns and spitting out what it thinks is the next thing in 18 00:01:27,360 --> 00:01:32,960 the pattern, more or less. To me, it just seems like this big mass of horrifying knowledge and it 19 00:01:32,960 --> 00:01:37,840 just gets kind of like randomly spit out whether it's accurate or not. So for me, it's like kind 20 00:01:37,840 --> 00:01:42,080 of spooky because you never really know what you're going to get. I think a lot of fantasy stuff 21 00:01:42,080 --> 00:01:47,520 going on. It's people getting very excited about what's going to be in the future while it's actually 22 00:01:47,520 --> 00:01:52,080 the capabilities are still a bit limited. However, they're crazy compared to two years ago, 23 00:01:52,080 --> 00:01:55,840 three years ago, like the transformational, especially the A.I. art space, like that's 24 00:01:55,840 --> 00:02:00,880 already pretty damn good and pretty damn disruptive and scary. I think overall it's really cool and 25 00:02:00,880 --> 00:02:06,880 it's a great idea. We could speed up things like machine learning. We can use it to diagnose people 26 00:02:06,880 --> 00:02:10,960 who have weird symptoms and stuff because let's face it, people are limited at some point. It 27 00:02:10,960 --> 00:02:18,480 becomes bad when we basically use prior work to then create new stuff and the people that made 28 00:02:18,480 --> 00:02:23,120 that work previously don't get any credit for it. It's not maybe as terrifying as people think it is. 29 00:02:23,120 --> 00:02:27,520 Yeah, certainly, certainly for it. So since you're for it, what's one thing you're really against in 30 00:02:27,520 --> 00:02:34,320 terms of the A.I.? Probably creating images of real people as we've seen with QTC and 31 00:02:34,320 --> 00:02:38,000 Pokimane and some of these other streamers that had explicit photos of them generated and shared 32 00:02:38,080 --> 00:02:42,640 on websites, stuff like that is awful. And now people think it's a lot of fun to create the A.I. 33 00:02:42,640 --> 00:02:47,040 songs of like Linus and stuff like that online, but he didn't consent to that. And no matter what, 34 00:02:47,040 --> 00:02:51,520 that's kind of wrong and weird to think that just because he's on the internet, he's okay with it. 35 00:02:51,520 --> 00:02:55,520 That's not true. And it's something that obviously there's no way to limit, but it's something I'm 36 00:02:55,520 --> 00:03:00,320 definitely concerned about. I think probably some of the protectionism. I mean, you're spending 37 00:03:00,320 --> 00:03:07,360 millions of dollars on like cloud infrastructure, GPU time, just to actually create a reasonable 38 00:03:07,360 --> 00:03:11,840 model. People, I think, are willing to protect that rather than just open sourcing it. Adobe's 39 00:03:11,840 --> 00:03:17,440 like generative fill is kind of like interesting. They're demonstrating it in like, look at, 40 00:03:17,440 --> 00:03:22,320 we could make this image so much wider and we can fill it just from this tiny little portrait. 41 00:03:22,320 --> 00:03:26,160 We could turn it into a whole landscape and it's like, yeah, I guess that's cool, but it's also 42 00:03:26,160 --> 00:03:30,480 like, what is the picture then anymore? A.I. is difficult. There are things that A.I. is really 43 00:03:30,480 --> 00:03:35,280 good for. Say, for example, A.I. image scaling or whatever, you train it to do whatever it needs 44 00:03:35,280 --> 00:03:41,440 to do and it does it basically by failing as many times as is necessary to fail the correct way. 45 00:03:41,440 --> 00:03:46,400 And that's actually really cool because it does things in a way that is very difficult to do in 46 00:03:46,400 --> 00:03:52,480 traditional software. When you get into marketing A.I. as well, I mean, I keep saying A.I. is more 47 00:03:52,480 --> 00:03:56,240 like machine learning. So like, when you get into marketing it as like A.I. and you say, oh, well, 48 00:03:56,240 --> 00:04:01,520 chat GPT is going to take over the world, well, that's overselling it. And that's, I think it's 49 00:04:01,520 --> 00:04:05,360 by design that they're doing that because they want people to invest. They want people to think 50 00:04:05,360 --> 00:04:10,640 that we're close enough. I think it's a really interesting tool. And I think it, depending on 51 00:04:10,640 --> 00:04:18,400 how it's used, it kind of broadens a lot of channels and possibilities of being able to sort 52 00:04:18,400 --> 00:04:23,760 through a lot of information really quickly. Like for example, you can ask A.I. something and it'll 53 00:04:23,760 --> 00:04:29,840 like throw out a ton of results. But then it's also like good to keep in mind. This is always just 54 00:04:30,640 --> 00:04:36,560 from a certain pool of information. So depending on how accurate that pool of information is, 55 00:04:36,560 --> 00:04:41,680 you can't always 100% trust it. I hate writing, by the way, but like, but when you figure out how 56 00:04:41,680 --> 00:04:45,680 to write something really cohesively and like, it's like, that's really impressive. And I just 57 00:04:45,680 --> 00:04:50,880 don't know how a prompt is going to lead to that. As a writer, you hate writing. I do. 58 00:04:53,120 --> 00:04:57,520 There's a lot of worry that like A.I. would take over creative jobs. And as we are in a creative 59 00:04:57,520 --> 00:05:03,680 field, what do you think about that? I personally don't think so. Because like, A.I. can be very 60 00:05:03,680 --> 00:05:10,160 objective. But like, they probably can really be subjective. If I trust the A.I. to grab focus for 61 00:05:10,160 --> 00:05:16,000 me, sometime, if we have a face in there, they'll grab like focus like pretty well. But there is 62 00:05:16,640 --> 00:05:21,680 one person of the chance that, you know, the focus gets shifted to something else. Because 63 00:05:21,680 --> 00:05:27,200 the camera thing, that's the face. Human being in general is still better in that case. Like, 64 00:05:27,280 --> 00:05:33,120 determine which part is the important part. So that's exactly the reason why we don't use like, 65 00:05:33,120 --> 00:05:38,640 autofocus on the top down and be cam for ShortCircuit. If no new writing is made, 66 00:05:38,640 --> 00:05:42,960 like if you say, for example, writing is completely no longer a thing. Basically, 67 00:05:42,960 --> 00:05:48,160 as you iterate on things that are already iterated on, you kind of like lose bits and pieces of it, 68 00:05:48,160 --> 00:05:53,760 it'll be difficult to continue indefinitely. You might get away with a little bit for a little 69 00:05:53,760 --> 00:05:58,400 while. People are already getting away with writing paragraphs or whatever. But it still requires 70 00:05:58,400 --> 00:06:02,160 like an editing pass. It still requires people who know what they're doing. Because I don't know 71 00:06:02,160 --> 00:06:08,720 if you or like people who are watching know. But A.I., especially like chat GBT, has a very short 72 00:06:08,720 --> 00:06:14,720 memory. So if you're trying to tell a story, like in a movie or something, keeping that thing on the 73 00:06:14,720 --> 00:06:18,880 rails is going to be a really tough, a really tough operation. Creative industry is definitely 74 00:06:18,880 --> 00:06:24,800 at a risk. I think it's going to be similar to other industries, like perhaps VFX, where there's 75 00:06:24,800 --> 00:06:28,960 humans still doing the job, but there's fewer of them, because the human is doing it can do 76 00:06:28,960 --> 00:06:33,760 so much faster now. But I think we're going to see things change and adapt. I think we're going to see 77 00:06:33,760 --> 00:06:40,160 a lot more like identity driven, personality driven content, like influencers, like Linus, 78 00:06:40,160 --> 00:06:44,720 who you love and have a parasocial relationship with, that's going to just become even more 79 00:06:44,720 --> 00:06:49,280 of a thing. Because in a couple of years, the majority of the content on the internet is going 80 00:06:49,280 --> 00:06:53,280 to be from A.I. And you're not even going to know that it's from A.I. And we're going to have to 81 00:06:53,280 --> 00:06:58,880 figure out identity online. And whether that's a blockchain thing or some kind of centralized, 82 00:06:58,880 --> 00:07:03,600 I don't know, identity system, we're going to have to figure out a way to know like, oh, 83 00:07:03,600 --> 00:07:08,400 actual humans said that, not just a very convincing bot with a political agenda. I don't 84 00:07:08,400 --> 00:07:12,560 know, oh yeah, will my job be a risk? If I was in high school right now, I would be, I'd be doing 85 00:07:12,560 --> 00:07:17,440 trades for sure. Something, the robots are 15 years behind the A.I.s right now, I think. So 86 00:07:18,480 --> 00:07:22,800 something practical that you can help with would definitely be a little more secure than a purely 87 00:07:22,800 --> 00:07:27,840 creative job. Yeah, I was telling Nicole, you know, A.I. can maybe edit videos, but it can't 88 00:07:27,840 --> 00:07:32,000 do a fart reverb like I can. No way. It can't put Vine boom in at the perfect moment. 89 00:07:32,960 --> 00:07:37,280 Have you used any A.I. currently? Like, where it's like, you have chat GBT, 90 00:07:37,280 --> 00:07:44,560 the image random rangers, music, are you like that? Have you used any of it? The only thing I've really done with A.I. was when Mid Journey and those other guys were coming 91 00:07:44,560 --> 00:07:49,840 out, I was messing around on it. I was like, yeah, this is cool. Haha, like Trump eating a lemon or 92 00:07:49,840 --> 00:07:54,720 whatever. I don't know. That's like the, oh my God, why is that my example? Have you used things 93 00:07:54,720 --> 00:07:59,120 like A.I. Dungeon, the like choosing an adventure Dungeons and Dragons type game online, where it 94 00:07:59,120 --> 00:08:02,960 creates a story for you to play through. And that's a lot of fun. But otherwise, I haven't done 95 00:08:02,960 --> 00:08:07,440 really any image generation. The only thing is I created an image of Kanye West at a ring door 96 00:08:07,440 --> 00:08:12,080 bell cam. And that was, that was pretty funny. Yeah, I've played with pretty much all of them. 97 00:08:12,080 --> 00:08:19,440 And I actually run a couple at home. So I run an LLM as well as stable diffusion for image generation. 98 00:08:19,440 --> 00:08:24,880 And I run those locally. I recently just bought a couple more 30 90s and, you know, upgraded some of 99 00:08:24,880 --> 00:08:31,280 them with VRAM so that I could try some of the larger models. So they're like a chat GBT for 100 00:08:32,160 --> 00:08:37,920 but you can only talk to them for like, I don't know, two to five minutes before the 101 00:08:37,920 --> 00:08:43,200 like token inference length gets too much. And it takes like a minute to respond to you. 102 00:08:43,200 --> 00:08:48,080 But initially they're they're really quite powerful as a tech slash creator company that we are. 103 00:08:48,080 --> 00:08:53,360 Do you think that we should be using AI and like any capacity or depends on the AI and depends on 104 00:08:53,360 --> 00:08:59,440 what we're using it for. If we're talking chat GBT, I don't know that it's not like copyright 105 00:08:59,440 --> 00:09:06,000 infringing. So that's one thing. The fact that we don't know what the test like the the actual 106 00:09:06,000 --> 00:09:12,400 data sets that they train them on is a bit of a big giant question mark in terms of like the 107 00:09:12,400 --> 00:09:20,000 ethics and the I guess legality ultimately of using AI for generative purposes. On the other 108 00:09:20,000 --> 00:09:26,080 hand, like say for example, I'm like 95% of the way through a script. And you're like, 109 00:09:26,160 --> 00:09:30,800 you don't know how to close it out. You could sit there and you could scratch your head for 110 00:09:30,800 --> 00:09:38,240 an hour or two, or you could ping some concepts off of chat GBT or something and see what it looks 111 00:09:38,240 --> 00:09:42,160 like. I believe we do currently. I think the business team uses it for talking points and 112 00:09:42,160 --> 00:09:46,480 things like that. And then they just kind of clean it up. I think basically every organization 113 00:09:46,480 --> 00:09:52,320 should be using whatever new tools come to light. I mean, transcribing meetings, summarizing a 114 00:09:52,320 --> 00:09:56,320 meeting that you missed, if it can allow your team to be smaller and increase your output, 115 00:09:56,320 --> 00:10:02,240 yeah, you should be using that really specific use cases where it's going to be really amazing. 116 00:10:02,240 --> 00:10:07,120 But in terms of like broad strokes, no, it's going to like help just like that tiny little 117 00:10:07,120 --> 00:10:11,760 step like making one thing that you're trying to making one vision you have a bit easier to 118 00:10:11,760 --> 00:10:17,120 to execute or like making it better. That sounds great. We should use it as a starting point, 119 00:10:17,120 --> 00:10:21,680 but we shouldn't use it for like the final product as a tool, we should be using it as a tool, 120 00:10:21,680 --> 00:10:26,080 like you'd use a computer as a tool or a camera as a tool. If I'm not like selling 121 00:10:26,080 --> 00:10:31,200 up other people's work, I'm okay with that. It's like, I need lines to be in a car, 122 00:10:31,200 --> 00:10:36,880 where can I get the car? You know, like that car, it costs money, that costs time, that costs people's 123 00:10:36,880 --> 00:10:42,960 effort. Yeah, so in general, just like as long as the AI we're not like selling up other people's 124 00:10:42,960 --> 00:10:49,440 stuff and be able to help us like increase our production quality, increase our speed, 125 00:10:50,240 --> 00:10:56,560 I think that's a good thing to use. Not really. Maybe machine learning for the lab, but like real 126 00:10:56,560 --> 00:11:01,920 AI and using it to make, I don't know, I don't think so. I just, I hate the idea of eliminating 127 00:11:01,920 --> 00:11:06,800 jobs, especially when it's not ready. I think in a couple of years that story might change and I 128 00:11:06,800 --> 00:11:11,440 might be like, yeah, sure, I don't want to make the timestamps for every video. So run it through 129 00:11:11,440 --> 00:11:16,080 the AI generator to generate those timestamps for us like fine. But right now, nah, I'm sweating 130 00:11:16,080 --> 00:11:19,120 like crazy. So we're going to wrap it up here. Do you want to say anything to the float planers? 131 00:11:19,280 --> 00:11:25,200 I'm trying to coin this phrase now, or like something you didn't, we didn't talk about that you wanted to expand on. Tell Sammy you don't like being called float planers. 132 00:11:26,640 --> 00:11:32,400 No, I'm so glad to get this off my chest. Everything, like all, all tools are inherently 133 00:11:32,400 --> 00:11:38,160 not good or evil. It is what you use it for. That is, this is the most important. So like AI 134 00:11:38,160 --> 00:11:42,080 every job scaling is amazing. For example, it's just one of those things that's just 135 00:11:43,600 --> 00:11:48,800 so universally useful. Try it at home. It's fun at home. It's a fun little exercise. It's 136 00:11:48,800 --> 00:11:54,320 really, really simple to get going. And it's a fun little playground and you don't have to like 137 00:11:56,000 --> 00:12:01,280 spend money on chat GPT. I mean, chat GPT is still the best. But if you play with it at home, 138 00:12:02,480 --> 00:12:07,600 yours is you can play with it as much as you want. It's still, it's still flawed. It's still very 139 00:12:07,600 --> 00:12:14,640 flawed. It's all no, like how to commit the perfect crime. It's got answers for that somewhere. You 140 00:12:14,640 --> 00:12:19,760 just got to convince it. I would love real AI. I want a Jarvis, you know, a Jarvis that goes like, 141 00:12:19,760 --> 00:12:23,280 Oh, I got you fam. I got this ready for you before like an actual personality, 142 00:12:23,280 --> 00:12:28,240 an actual artificial intelligence that's effectively a human. That'll be scary. And it means like 143 00:12:28,240 --> 00:12:33,120 deleting an AI is like deleting a sentient life. And that's why I does not truly exist now is 144 00:12:33,120 --> 00:12:37,760 because like a sentient life is not something that's easy to do. For a lot of like the medical 145 00:12:37,760 --> 00:12:46,800 environment, I am pretty like pro AI because like some people are doing research on like data and 146 00:12:46,800 --> 00:12:54,640 stuff like for data driven stuff. Like I think AI is definitely winning and also quite like useful. 147 00:12:54,640 --> 00:13:00,480 Maybe like if the AI research enough like paper, we can cure cancer, you know, you never know. So 148 00:13:00,480 --> 00:13:05,440 in that perspective, I think it's great. Generally excited. Generally excited. 149 00:13:08,000 --> 00:13:15,440 Yeah, cut this out. No, this is uncut version. All right, we'll cut it there. Thank you very 150 00:13:15,440 --> 00:13:22,480 much. Bye flow planers. James leave him, but I'm gonna keep waving. I'm committed a bit. Bye.