1 00:00:00,000 --> 00:00:08,640 Wait, so people prefer and like AI-generated images over the real art, but same group of 2 00:00:08,640 --> 00:00:12,960 people also say that it's going to be bad. 3 00:00:12,960 --> 00:00:33,680 Yeah. 4 00:00:33,680 --> 00:00:37,800 Ladies and gentlemen, boys and girls, children of all ages, dogs, cats, robots, and everybody 5 00:00:37,800 --> 00:00:43,660 in between, especially you indistinguishable AI-generated mid-journey models, welcome to 6 00:00:43,660 --> 00:00:45,920 HTTTA. 7 00:00:45,920 --> 00:00:50,960 I am your host, Wes the SynthBind SynthBind Wes, and I am joined by the graceful, the 8 00:00:50,960 --> 00:00:57,400 grandiose, the garden of genius herself, the gregarious and glowing, filled with glittering 9 00:00:57,400 --> 00:01:01,200 secrets about everything AI, the glorious go-to-go. 10 00:01:01,200 --> 00:01:03,200 G, how are you this week? 11 00:01:03,200 --> 00:01:11,240 Hi, I'm so glorious that I need to tell you I've been catfished and thanks to you. 12 00:01:11,240 --> 00:01:12,240 Yeah. 13 00:01:12,240 --> 00:01:17,520 Well, let's get into that right away because that's quite a shift from the mercurial intros 14 00:01:17,520 --> 00:01:19,060 that I tried to purvey here. 15 00:01:19,060 --> 00:01:21,720 I had gota, I found this site. 16 00:01:21,720 --> 00:01:23,320 What's the name of the site again? 17 00:01:23,320 --> 00:01:26,760 It's kitefished.com. 18 00:01:26,760 --> 00:01:27,760 Katfished.com. 19 00:01:27,760 --> 00:01:31,400 We'll put the link in the show notes and the YouTube description. 20 00:01:31,400 --> 00:01:35,280 So what it is, is it's a 17 question test. 21 00:01:35,280 --> 00:01:40,920 All you have to do is say, is the person a real person or an AI-generated person? 22 00:01:40,920 --> 00:01:43,360 And you're going to go through it and see how you do. 23 00:01:43,360 --> 00:01:47,160 So go to, I'm curious to see how you scored out of the 17 pictures. 24 00:01:47,160 --> 00:01:49,400 I scored 14. 25 00:01:49,400 --> 00:01:50,880 That's pretty good. 26 00:01:50,880 --> 00:01:51,880 That's pretty good. 27 00:01:51,880 --> 00:01:55,760 I took it as well and got a 15 of 17. 28 00:01:55,760 --> 00:02:01,260 Now we probably have an unfair disadvantage given that we're constantly playing with these 29 00:02:01,260 --> 00:02:02,480 tools and using them. 30 00:02:02,480 --> 00:02:07,760 So what were some of the things to you that stood out when evaluating these 17 different 31 00:02:07,760 --> 00:02:10,400 images as if the person was real or fake? 32 00:02:10,400 --> 00:02:14,120 I never looked so analytically into a woman's body. 33 00:02:14,120 --> 00:02:17,120 I'm like, is this armpit correct? 34 00:02:17,120 --> 00:02:19,220 That's a nice looking armpit on that one person. 35 00:02:19,220 --> 00:02:20,220 This is going to be interesting. 36 00:02:20,220 --> 00:02:22,240 In the art academy, I used to draw people. 37 00:02:22,240 --> 00:02:27,120 So I actually can draw people hyper-realistically. 38 00:02:27,120 --> 00:02:32,400 And this actually was my first gig in high school earning money and going to music festivals 39 00:02:32,400 --> 00:02:36,560 because I would offer to draw pencil hyper-realistic portraits. 40 00:02:36,560 --> 00:02:38,280 Oh, that's awesome. 41 00:02:38,280 --> 00:02:40,000 Anatomical things of the face. 42 00:02:40,000 --> 00:02:47,680 I should know good, but if this is completely failure on my side, then that says a lot how 43 00:02:47,680 --> 00:02:50,080 good these models have become. 44 00:02:50,080 --> 00:02:51,080 Yeah. 45 00:02:51,080 --> 00:02:53,720 Thing that I noticed because it's not so much in the faces. 46 00:02:53,720 --> 00:02:58,200 We can tell if the face is a little off in certain ways. 47 00:02:58,200 --> 00:03:03,400 The size of the eyeballs sometimes mid-journey has an issue with. 48 00:03:03,400 --> 00:03:07,880 It'll make one eyeball bigger than the other and not the whole eye socket, but just the 49 00:03:07,880 --> 00:03:11,040 scope and scale of the eyeball itself. 50 00:03:11,040 --> 00:03:16,920 We have this, it's called uncanny valley where it's basically like we naturally look at faces 51 00:03:16,920 --> 00:03:17,920 all the time. 52 00:03:17,920 --> 00:03:21,380 So our minds are used to seeing those things. 53 00:03:21,380 --> 00:03:26,320 But then I find that when you're taking a test like this to try to detect them, we can 54 00:03:26,320 --> 00:03:32,420 kind of overfit sometimes that I think like too many things are real or AI. 55 00:03:32,420 --> 00:03:34,840 This is how I would from now on will judge. 56 00:03:34,840 --> 00:03:38,760 If eyebrows are not symmetrical, that's a real human being. 57 00:03:38,760 --> 00:03:40,680 You need to have imperfections. 58 00:03:40,680 --> 00:03:41,680 Exactly. 59 00:03:41,680 --> 00:03:44,080 But mid-journey gets this wrong too. 60 00:03:44,080 --> 00:03:48,480 So anything to do with pairs of stuff, it can mess up. 61 00:03:48,480 --> 00:03:52,760 So one thing to check if you're looking at pictures of women is their earrings. 62 00:03:52,760 --> 00:03:57,680 If they're wearing earrings, sometimes if they're wearing different ones, which would 63 00:03:57,680 --> 00:04:00,120 probably happen, it's AI generated. 64 00:04:00,120 --> 00:04:02,120 For me was chin, not double chin. 65 00:04:02,120 --> 00:04:05,320 Like it's very like man's chin kind of fish. 66 00:04:05,320 --> 00:04:06,320 Yeah, the little chin bud. 67 00:04:06,320 --> 00:04:07,320 Like there's the shadow. 68 00:04:07,320 --> 00:04:08,920 Yeah, exactly. 69 00:04:08,920 --> 00:04:15,320 And one more aspect for me was how the background light hits hair. 70 00:04:15,320 --> 00:04:22,920 It just seems as you would in Photoshop select a person and paste it in a different background. 71 00:04:22,920 --> 00:04:28,240 If you think like from the color theory, like each light bounces differently from different 72 00:04:28,240 --> 00:04:29,420 surfaces. 73 00:04:29,420 --> 00:04:35,400 So if you take a person with black hair, but the background was white, and now you place 74 00:04:35,400 --> 00:04:41,120 the same person on the black background, it's going to feel something is wrong because how 75 00:04:41,120 --> 00:04:44,560 the white light bounced on the dark hair. 76 00:04:44,560 --> 00:04:46,240 But now with AI, right? 77 00:04:46,240 --> 00:04:50,240 AI you can just say change background. 78 00:04:50,240 --> 00:04:56,640 But those may be these micro details, which we as humans recognize from nature, how light 79 00:04:56,640 --> 00:04:58,340 bounces from objects. 80 00:04:58,340 --> 00:05:04,900 This is something at least for me, I always look at is there a light bouncing on the object, 81 00:05:04,900 --> 00:05:06,120 what it should be. 82 00:05:06,120 --> 00:05:10,120 And they figured this out in video games because you have stuff in the past few years, like 83 00:05:10,120 --> 00:05:15,640 ray tracing, where they're actually able to, with mathematical equations, map out how light 84 00:05:15,640 --> 00:05:20,800 is going to bounce off and reflect and refract on certain objects in a space. 85 00:05:20,800 --> 00:05:23,680 But I don't know if that happens in mid journey or anything like that. 86 00:05:23,680 --> 00:05:28,760 The only fact is from this point on, it's going to be just getting better and better. 87 00:05:28,760 --> 00:05:32,800 And I think the survey that we'll also put a link to in the show notes, they surveyed, 88 00:05:32,800 --> 00:05:36,000 I'm trying to find the exact number here, but a bunch of people, again, they showed 89 00:05:36,000 --> 00:05:38,600 them these quads of four people that looked similar. 90 00:05:38,600 --> 00:05:40,640 You know, it was done right in your backyard there. 91 00:05:40,640 --> 00:05:42,080 Yeah, it was done in Germany. 92 00:05:42,080 --> 00:05:49,000 So aside from the test for the images, they also did a survey after just on their perceptions 93 00:05:49,000 --> 00:05:52,260 of AI, if they heard about things, what their feeling was. 94 00:05:52,260 --> 00:05:57,960 So again, they were over a thousand test subjects were shown multiple images in quads. 95 00:05:57,960 --> 00:06:00,280 So there was four images at once. 96 00:06:00,280 --> 00:06:05,000 Everybody looked kind of similar and they had to identify the one real image in there, 97 00:06:05,000 --> 00:06:06,500 the one real person. 98 00:06:06,500 --> 00:06:11,760 The moment you said Germany, I'm kind of not surprised that a lot of people fell for it. 99 00:06:11,760 --> 00:06:12,760 You think so? 100 00:06:12,760 --> 00:06:13,760 It's crazy. 101 00:06:13,760 --> 00:06:17,240 I just met on a weekend some of my friends and they never heard of it. 102 00:06:17,240 --> 00:06:20,520 They are like, your YouTube videos were the first touch point. 103 00:06:20,520 --> 00:06:21,520 We were like, what? 104 00:06:21,520 --> 00:06:22,520 Like, what is this? 105 00:06:22,520 --> 00:06:24,920 Like I can actually use this? 106 00:06:24,920 --> 00:06:31,160 Yeah, if you go on a street, I can almost promise to you that 99% of people never heard 107 00:06:31,160 --> 00:06:32,160 of it. 108 00:06:32,160 --> 00:06:33,160 That's interesting. 109 00:06:33,160 --> 00:06:38,480 So again, Germany and then you have to look in the population, the average age and all 110 00:06:38,480 --> 00:06:45,000 these things and also the relationship with technology and personal data plus the history 111 00:06:45,000 --> 00:06:46,720 playing a role in that. 112 00:06:46,720 --> 00:06:53,200 So I'm just kind of wonder what's going to happen when actually EU wakes up and Germany 113 00:06:53,200 --> 00:06:55,000 wakes up to all of this. 114 00:06:55,000 --> 00:06:58,680 Well, let me give you some stats from this digital insight survey, because I think some 115 00:06:58,680 --> 00:07:02,880 of them are kind of intriguing and maybe kind of aligned with what you'd expect. 116 00:07:02,880 --> 00:07:06,920 And you can offer some probably a little better insight, obviously, because this survey is 117 00:07:06,920 --> 00:07:08,880 done on your doorstep here. 118 00:07:08,880 --> 00:07:13,640 So the main thing that kind of brought up these AI images, if they're distinguishable 119 00:07:13,640 --> 00:07:18,740 or indistinguishable from one another, only about 8% of Germans could even accurately 120 00:07:18,740 --> 00:07:23,800 identify when presented four images, which one of those was real. 121 00:07:23,800 --> 00:07:28,320 One of the interesting ones was 94% of survey participants agreed. 122 00:07:28,320 --> 00:07:35,360 So that kind of tracks with that 8% that it's really difficult to tell and about 57% of 123 00:07:35,360 --> 00:07:42,120 people, so almost two thirds, actually preferred AI generated images, artwork, everything in 124 00:07:42,120 --> 00:07:44,820 that space to ones that had been done by humans. 125 00:07:44,820 --> 00:07:50,360 So it's interesting because the contrary point to that that came out is only about one third 126 00:07:50,360 --> 00:07:55,440 think that it's going to be a good thing for the creative process, for creativity with 127 00:07:55,440 --> 00:07:59,680 another 18% saying it's going to be just detrimental. 128 00:07:59,680 --> 00:08:00,680 You know? 129 00:08:00,680 --> 00:08:08,600 Wait, so people prefer and like AI generated images over the real art, but same group of 130 00:08:08,600 --> 00:08:13,560 people also say that it's going to be bad. 131 00:08:13,560 --> 00:08:19,240 But what we know that it's the truth, like the reality is in what we as human, what we 132 00:08:19,240 --> 00:08:21,720 like and prefer and we just lean to that. 133 00:08:21,720 --> 00:08:27,680 Or I think it's just we don't want to admit that, hey, you know, there's no one that created 134 00:08:27,680 --> 00:08:28,680 this thing. 135 00:08:28,680 --> 00:08:29,680 I can relate to another person. 136 00:08:29,680 --> 00:08:34,280 I can't relate to a machine that did this in the same way with a bunch of math problems. 137 00:08:34,280 --> 00:08:39,440 So even though I like the advertisement of the model or the shoe or the painting that 138 00:08:39,440 --> 00:08:44,380 the AI preferred more, I don't think that's good for anybody who's going to actually be 139 00:08:44,380 --> 00:08:47,040 a model or painter or something like that. 140 00:08:47,040 --> 00:08:50,720 It's so funny how history repeats itself. 141 00:08:50,720 --> 00:08:55,720 If you look when cameras were invented and what people said, that first thing like of 142 00:08:55,720 --> 00:09:02,680 course photography, now you can capture people how they are in reality in the moment, but 143 00:09:02,680 --> 00:09:09,760 there was so much fear that now anybody can take a picture of you and catch you secretly. 144 00:09:09,760 --> 00:09:16,040 But then there was a whole other aspect that to have your portrait taken, you had to hire 145 00:09:16,040 --> 00:09:19,120 an artist who would paint you realistically. 146 00:09:19,120 --> 00:09:24,840 However, that realistically never really happened because as I shared with you from my experience 147 00:09:24,840 --> 00:09:30,800 when people asked me to draw hyper realistically their girlfriends or family members, I would 148 00:09:30,800 --> 00:09:37,140 always get like, oh, could you make her eyelashes longer or could you do this or that? 149 00:09:37,140 --> 00:09:42,120 Can you make it a little bit better version that the person who receives it would like 150 00:09:42,120 --> 00:09:43,120 it? 151 00:09:43,120 --> 00:09:48,360 So the same with kings and queens and all these lords that when we hired an artist to 152 00:09:48,360 --> 00:09:55,160 make a painting of them, they wanted something that they, how they would like to be portrayed 153 00:09:55,160 --> 00:09:59,040 and suddenly cameras came who do reality. 154 00:09:59,040 --> 00:10:05,360 And suddenly there was instances where people are like, oh, so this Lord is actually fat 155 00:10:05,360 --> 00:10:11,840 and with a beard, but all what we saw that he's on a horse riding in armor, right? 156 00:10:11,840 --> 00:10:16,880 And it's kind of just gives me the same sentiment that this split between, oh, we actually 157 00:10:16,880 --> 00:10:20,360 prefer this as example, realistic image. 158 00:10:20,360 --> 00:10:24,840 And now we actually can see the truth, but it's bad. 159 00:10:24,840 --> 00:10:26,000 Yeah. 160 00:10:26,000 --> 00:10:29,680 That's definitely a divergence because we already can't tell. 161 00:10:29,680 --> 00:10:32,440 So you can't tell why does it matter? 162 00:10:32,440 --> 00:10:38,820 But then once you know your opinion somehow is changed, to me, that's not an objective 163 00:10:38,820 --> 00:10:40,120 way to assess something. 164 00:10:40,120 --> 00:10:42,600 You can't objectively assess art anyway. 165 00:10:42,600 --> 00:10:47,880 It's a subjective thing, but to just have your opinion switch because it was made by 166 00:10:47,880 --> 00:10:54,880 a machine versus a person, I think is not a barometer that art or creativity should 167 00:10:54,880 --> 00:10:56,720 necessarily be judged by. 168 00:10:56,720 --> 00:11:02,360 In academia, I learned very, very early one big lesson. 169 00:11:02,360 --> 00:11:03,960 The saying goes like that. 170 00:11:03,960 --> 00:11:09,260 How good is your art is equal to how many eyeballs have seen it. 171 00:11:09,260 --> 00:11:14,960 So if you think about it, it basically says that it's equals to the eyeballs because if 172 00:11:14,960 --> 00:11:21,400 your art has been seen by 10 million people, there is going to be a fraction of people 173 00:11:21,400 --> 00:11:26,080 who don't like it completely, but just a fraction of people who like it is bigger than your 174 00:11:26,080 --> 00:11:28,840 art been seen by a thousand people. 175 00:11:28,840 --> 00:11:34,600 So I think it's the same with AI art that if you show, and this is why those AI arts 176 00:11:34,600 --> 00:11:36,840 win competitions, right? 177 00:11:36,840 --> 00:11:41,680 Because at the end of the day, how many of those people who actually like it, and that's 178 00:11:41,680 --> 00:11:42,680 the truth. 179 00:11:42,680 --> 00:11:45,200 That's what determines how good is the art. 180 00:11:45,200 --> 00:11:46,200 Yeah. 181 00:11:46,200 --> 00:11:49,360 The results of the study are pretty interesting and they made it very kind of just about not 182 00:11:49,360 --> 00:11:54,040 only the image detection, but also just like general AI awareness. 183 00:11:54,040 --> 00:11:58,400 Only about half of people in the study, the thousand people they certified, had even heard 184 00:11:58,400 --> 00:12:03,400 of these generative tools that they realized they could do stuff for business or art or 185 00:12:03,400 --> 00:12:08,000 music and still only about half were interested in learning more about them. 186 00:12:08,000 --> 00:12:12,480 So I have friends that have said to them, it feels kind of like when crypto and Bitcoin 187 00:12:12,480 --> 00:12:13,480 came out. 188 00:12:13,480 --> 00:12:14,480 It feels the same. 189 00:12:14,480 --> 00:12:19,160 And I'm like, yeah, but the barrier to entry to that was so much higher because if you 190 00:12:19,160 --> 00:12:23,760 wanted to trade crypto or buy crypto, especially in the early days, you had to get a computer. 191 00:12:23,760 --> 00:12:25,800 You had to load special software on that. 192 00:12:25,800 --> 00:12:31,360 You had to get a wallet to put your keys in offline or store all these fancy things that 193 00:12:31,360 --> 00:12:35,520 no one had any frame of reference for, you know, to be trading it. 194 00:12:35,520 --> 00:12:38,800 You weren't able to trade it normally in the US for the longest time. 195 00:12:38,800 --> 00:12:41,560 So I thought that that comparison was a little different. 196 00:12:41,560 --> 00:12:44,820 But then again, I wasn't all in on crypto either. 197 00:12:44,820 --> 00:12:47,240 Like I am some of these AI tools. 198 00:12:47,240 --> 00:12:52,880 The kind of sentiment I'm sensing, at least here in Europe, that people get a little bit 199 00:12:52,880 --> 00:12:56,560 tired of how media shapes all of this. 200 00:12:56,560 --> 00:13:02,960 And I also point fingers to the content creators that there is this whole hype and we go from 201 00:13:02,960 --> 00:13:06,600 crypto to NFTs to metaverse. 202 00:13:06,600 --> 00:13:08,600 Now the whole thing is AI. 203 00:13:08,600 --> 00:13:12,680 And it feels like these hypes are half a year to a year cycle. 204 00:13:12,680 --> 00:13:17,960 And then everyone kind of forgets or it crashes or people lose a lot of money. 205 00:13:17,960 --> 00:13:25,220 You know, so I'm kind of not surprised that the moment AI comes into the attention cycle, 206 00:13:25,220 --> 00:13:29,320 people go like, oh, okay, let me wait and see, is this a real thing? 207 00:13:29,320 --> 00:13:34,840 The problem is, or maybe not a problem, that AI is a technology. 208 00:13:34,840 --> 00:13:40,480 And if you actually look in the history, that's been worked on and thought about and applied 209 00:13:40,480 --> 00:13:43,400 in the use cases every day. 210 00:13:43,400 --> 00:13:47,360 That technology is solid and real and it's happening right now. 211 00:13:47,360 --> 00:13:55,400 But of course, this gap between something developed to mass awareness, to alignment, 212 00:13:55,400 --> 00:13:58,520 to policymaking is insane, huge. 213 00:13:58,520 --> 00:14:01,360 And it's kind of this like exponential growth of AI. 214 00:14:01,360 --> 00:14:05,880 So the moment, let's say, I don't want to say majority, but at least what you said, 215 00:14:05,880 --> 00:14:12,400 like, okay, half, half of population of let's say Germany knows about chat GPT or that it 216 00:14:12,400 --> 00:14:16,440 exists, we are going to be on GPT-5 or 6. 217 00:14:16,440 --> 00:14:23,600 And you know, this kind of like a lag effect that you really don't understand what's happening, 218 00:14:23,600 --> 00:14:24,600 what's going on. 219 00:14:24,600 --> 00:14:29,320 Like even for us, what it feels being in the middle of it, it happened so fast. 220 00:14:29,320 --> 00:14:35,760 For me, it's scary how this sentiment can influence attitude towards it because I just 221 00:14:35,760 --> 00:14:42,040 heard again on a weekend friends ad that their companies blocked chat GPT just plain blank. 222 00:14:42,040 --> 00:14:46,600 I do want to share before we get on, because I know we talk about how do you do detecting 223 00:14:46,600 --> 00:14:48,600 AI or fake or real images. 224 00:14:48,600 --> 00:14:53,000 I do want to tell this one story about how auto GPT actually really helped me out. 225 00:14:53,000 --> 00:14:54,640 And this was kind of surprising. 226 00:14:54,640 --> 00:14:55,640 Okay. 227 00:14:55,640 --> 00:15:02,600 So as you know, we're week four in this podcast, GNI are our noobs, self-proclaimed and champions 228 00:15:02,600 --> 00:15:06,560 of our new noobdom in the world of podcasting. 229 00:15:06,560 --> 00:15:15,000 There's a seedy underbelly that I have now found, but was saved from by auto GPT. 230 00:15:15,000 --> 00:15:20,760 So I've been doing some podcasts myself, met some great people during the course of doing 231 00:15:20,760 --> 00:15:25,080 their podcasts about, you know, I did one, it's called the Coder Kids podcast. 232 00:15:25,080 --> 00:15:29,400 We had a great discussion about how parents, I'm a parent myself, you know, should use 233 00:15:29,400 --> 00:15:34,840 these tools and approach them with their kids, how these might change education. 234 00:15:34,840 --> 00:15:39,320 That's done a couple other podcasts as well, but I did one last Friday was my first like 235 00:15:39,320 --> 00:15:41,480 five minute interview, right? 236 00:15:41,480 --> 00:15:46,280 That was for a live, supposedly streaming podcast radio station. 237 00:15:46,280 --> 00:15:51,960 I will not name it for purposes of this story, but this is also just a sample of some of 238 00:15:51,960 --> 00:15:52,960 the things I've found. 239 00:15:52,960 --> 00:15:57,920 So there's a lot of podcasts out there that are set up as commodities in and of themselves 240 00:15:57,920 --> 00:16:00,280 where it's like, Hey, yeah, you can be a guest. 241 00:16:00,280 --> 00:16:01,920 Anybody can come be a guest on my podcast. 242 00:16:01,920 --> 00:16:05,720 You just got to pay me, you know, certain amount of money to come be the guest because 243 00:16:05,720 --> 00:16:10,200 I've got the audience and they have a whole cycle set up like this where they're just 244 00:16:10,200 --> 00:16:16,720 getting people just, you know, all these different funnels with advertisements playing off people's 245 00:16:16,720 --> 00:16:20,200 emotions like vanity and be like, we absolutely want you on this thing. 246 00:16:20,200 --> 00:16:21,200 Come on down. 247 00:16:21,200 --> 00:16:24,760 I didn't think this was the case here because the ones that I had run into before that I 248 00:16:24,760 --> 00:16:26,760 was like, Oh, you know, let me cold call them. 249 00:16:26,760 --> 00:16:28,160 See, I think I'd be a good fit. 250 00:16:28,160 --> 00:16:33,400 The ones I'd run into before would lead with that would say, Hey, yep, you can come on. 251 00:16:33,400 --> 00:16:34,400 You just pay this fee. 252 00:16:34,400 --> 00:16:36,160 I'm like, all right, that's just, that's not for me. 253 00:16:36,160 --> 00:16:38,320 It's a little scammy or whatever. 254 00:16:38,320 --> 00:16:41,640 So this week I did the five minute interview. 255 00:16:41,640 --> 00:16:42,640 It was fine. 256 00:16:42,640 --> 00:16:47,840 I thought, and you know, they posted on their site, their site has from the outside, 113 257 00:16:47,840 --> 00:16:50,440 five star reviews, five stars on Apple podcast. 258 00:16:50,440 --> 00:16:53,720 If you Google it, it's in the top 3% of podcasts. 259 00:16:53,720 --> 00:16:56,360 So it's like, Oh, this is a great opportunity. 260 00:16:56,360 --> 00:16:57,360 So I get a call today. 261 00:16:57,360 --> 00:17:00,320 I did the interview last Friday, forgot about it. 262 00:17:00,320 --> 00:17:01,320 Right. 263 00:17:01,320 --> 00:17:02,320 All right. 264 00:17:02,320 --> 00:17:03,320 It was fine. 265 00:17:03,320 --> 00:17:04,320 I went and listened to it. 266 00:17:04,320 --> 00:17:05,320 It's like Kate listening to myself, you know, sure. 267 00:17:05,320 --> 00:17:06,320 You're the same way. 268 00:17:06,320 --> 00:17:07,320 But he's like, I can't do it. 269 00:17:07,320 --> 00:17:08,320 I sound so stupid. 270 00:17:08,320 --> 00:17:14,120 Even though everyone's like, you sound so great on these podcasts. 271 00:17:14,120 --> 00:17:15,120 Yeah. 272 00:17:15,120 --> 00:17:16,120 So I'm like, can't listen to it. 273 00:17:16,120 --> 00:17:17,400 Totally forgotten about it. 274 00:17:17,400 --> 00:17:20,760 It exists in the greater internet, but whatever. 275 00:17:20,760 --> 00:17:24,080 So he's called today and they're like, Hey, is this Wes? 276 00:17:24,080 --> 00:17:30,080 This is the, one of the partners and co-executive producers of the entire podcast network that 277 00:17:30,080 --> 00:17:34,240 you're on and this show, you know, was that your first podcast that you've ever been on? 278 00:17:34,240 --> 00:17:37,680 I'm like, actually no, I podcast myself. 279 00:17:37,680 --> 00:17:42,240 But they're like, well, for a five minute interview, you know, we have all these polling 280 00:17:42,240 --> 00:17:47,520 and demographics and you tested a 9.3, you were a 9.3. 281 00:17:47,520 --> 00:17:50,080 We don't get that for five minute interviews. 282 00:17:50,080 --> 00:17:55,440 And I'm just all like, Oh, well, of course you don't cause I, I am the master of AI. 283 00:17:55,440 --> 00:18:00,920 No, you've got radio voice DJ talking about AI. 284 00:18:00,920 --> 00:18:05,440 So this person now has got my, he got, got her both her hooks in. 285 00:18:05,440 --> 00:18:06,440 Right. 286 00:18:06,440 --> 00:18:08,440 And she starts going through her script. 287 00:18:08,440 --> 00:18:09,440 Right. 288 00:18:09,440 --> 00:18:10,960 So, Hey, you know, here's what I'd like to do. 289 00:18:10,960 --> 00:18:15,560 We don't have anybody on our channel that is an AI expert. 290 00:18:15,560 --> 00:18:16,640 Right. 291 00:18:16,640 --> 00:18:17,640 We want to book you again. 292 00:18:17,640 --> 00:18:23,400 We want to give you next 10 weeks of half hour segments, you know, in a certain time, 293 00:18:23,400 --> 00:18:25,440 you're going to get interviewed by one of our celebrity hosts. 294 00:18:25,440 --> 00:18:29,920 They sent me this site and one of their celebrity hosts has got 14 Emmy awards and is like a 295 00:18:29,920 --> 00:18:32,160 newscaster in New York. 296 00:18:32,160 --> 00:18:34,600 And I'm like, that's pretty legit, I guess. 297 00:18:34,600 --> 00:18:35,600 Right. 298 00:18:35,600 --> 00:18:38,320 So they're like, you're this person's going to interview you for the next 10 weeks. 299 00:18:38,320 --> 00:18:40,040 You have creative control over the show. 300 00:18:40,040 --> 00:18:42,920 And I'm like, I don't think I was that good on the podcast. 301 00:18:42,920 --> 00:18:45,120 Like, you know, and going through. 302 00:18:45,120 --> 00:18:48,000 And then they're like, we're going to do video interviews too. 303 00:18:48,000 --> 00:18:50,200 There's going to be sponsorship deals associated with this. 304 00:18:50,200 --> 00:18:52,360 And I'm like, is any of this real life? 305 00:18:52,360 --> 00:18:53,760 Like what is happening right now? 306 00:18:53,760 --> 00:18:59,080 I've done three, now four podcast episodes with you and been a guest on three others, 307 00:18:59,080 --> 00:19:02,040 you know, I thought we're aiming to be there. 308 00:19:02,040 --> 00:19:06,320 I'm in the tap 1% of podcasters yet, you know, about anything. 309 00:19:06,320 --> 00:19:07,840 Well, you don't know. 310 00:19:07,840 --> 00:19:08,840 Okay. 311 00:19:08,840 --> 00:19:13,720 I'm looking what is the catch here because it sounds too good to be true. 312 00:19:13,720 --> 00:19:14,720 Exactly. 313 00:19:14,720 --> 00:19:19,480 So for me it would be red alerts because I've been invited to so many podcasts. 314 00:19:19,480 --> 00:19:25,720 I don't go, but because it's red alert all around and I'm very cautious, but I'm listening. 315 00:19:25,720 --> 00:19:28,680 At least in those instances, they're inviting you. 316 00:19:28,680 --> 00:19:32,320 I reached out, did a little survey because it was just like, okay, you know, yeah, come 317 00:19:32,320 --> 00:19:33,920 on five minute segment. 318 00:19:33,920 --> 00:19:34,920 That's great. 319 00:19:34,920 --> 00:19:35,920 And then did it, forgot about it. 320 00:19:35,920 --> 00:19:40,720 But anyway, so I'm apparently going to get 12 different video calls that they're going 321 00:19:40,720 --> 00:19:44,800 to post and their show airs on channel 117 on Comcast in your area. 322 00:19:44,800 --> 00:19:48,680 And I'm like, beyond TV, like what, what? 323 00:19:48,680 --> 00:19:52,520 And they're like, we guarantee you'll have at least 25 leads. 324 00:19:52,520 --> 00:19:56,520 And then, oh yeah, by the way, we're going to get you a billboard in Times Square, New 325 00:19:56,520 --> 00:19:57,520 York. 326 00:19:57,520 --> 00:20:00,280 And I'm like, what is going on right now? 327 00:20:00,280 --> 00:20:02,600 Like, as is this real? 328 00:20:02,600 --> 00:20:04,720 I'm like, am I now the face? 329 00:20:04,720 --> 00:20:05,720 Is this a cam? 330 00:20:05,720 --> 00:20:06,720 Yes. 331 00:20:06,720 --> 00:20:09,280 Let me tell you one night. 332 00:20:09,280 --> 00:20:13,920 So then they go, it's only going to cost $5,000 to set all this up. 333 00:20:13,920 --> 00:20:16,240 This is normally a package that would be 25 grand. 334 00:20:16,240 --> 00:20:18,480 We're so interested in you. 335 00:20:18,480 --> 00:20:22,920 And I'm like, it's not, this is not how I imagine this would work. 336 00:20:22,920 --> 00:20:26,480 But for 20 minutes, I am just spoon feeding me. 337 00:20:26,480 --> 00:20:32,000 I'm like, yes, I'm now the champion of AI and prompt engineering news for the world 338 00:20:32,000 --> 00:20:38,000 to see, you know, like just playing on all those emotions of vanity, pride, and you know, 339 00:20:38,000 --> 00:20:40,200 it just all pumped up. 340 00:20:40,200 --> 00:20:41,840 It says Times Square billboard. 341 00:20:41,840 --> 00:20:46,680 And you go on their site, they got pictures of everybody there with their face back on 342 00:20:46,680 --> 00:20:47,960 a billboard at Times Square. 343 00:20:47,960 --> 00:20:51,760 And I'm just like, everything about this looks legit. 344 00:20:51,760 --> 00:20:53,520 Everything is such a scheme. 345 00:20:53,520 --> 00:20:54,520 Yeah. 346 00:20:54,520 --> 00:20:58,320 So anyway, so they say it's only going to cost this $5,000 best bet. 347 00:20:58,320 --> 00:21:03,560 So at this point, I'm a little put off by that, but I had seen that in some of these 348 00:21:03,560 --> 00:21:04,880 other podcasts. 349 00:21:04,880 --> 00:21:10,120 However, my sense of their legitimacy at this point was like, and maybe that's not a bad 350 00:21:10,120 --> 00:21:11,120 idea. 351 00:21:11,120 --> 00:21:12,720 Like I want to get Times Square billboard. 352 00:21:12,720 --> 00:21:16,960 It's $150 for 30 seconds on Times Square billboards, like 12 grand a day or something like that. 353 00:21:16,960 --> 00:21:20,360 You know, like if I'm going to pay $5,000, I'm going to be in Times Square. 354 00:21:20,360 --> 00:21:22,560 I'm going to go point myself and be like, yes, right. 355 00:21:22,560 --> 00:21:23,560 That's me. 356 00:21:23,560 --> 00:21:25,460 I'm famous, you know, or whatever. 357 00:21:25,460 --> 00:21:29,420 That's just one of those like typical American like measures of success. 358 00:21:29,420 --> 00:21:30,420 Something in Times Square. 359 00:21:30,420 --> 00:21:35,400 Times Square is the center of the most Americana place like in New York. 360 00:21:35,400 --> 00:21:39,240 So when they say it's only going to cost $5,000, throws up a little red flag. 361 00:21:39,240 --> 00:21:43,680 But being the hustler and businessman that I am, I'm like, so, hey, why don't we, instead 362 00:21:43,680 --> 00:21:49,400 of exchanging any pecuniary stuff, you know, making any kind of deal like this, why don't 363 00:21:49,400 --> 00:21:52,680 we talk about what I can offer you guys? 364 00:21:52,680 --> 00:21:58,680 If you have this social media and podcast and everything kind of network, surely there's 365 00:21:58,680 --> 00:22:03,800 a ton of AI tools that we could deploy or develop for y'all that would be beneficial. 366 00:22:03,800 --> 00:22:07,960 And I mean, what business owner at that point isn't at least a little interested on saying, 367 00:22:07,960 --> 00:22:13,380 oh, I can have someone do all the social media content for three different shows across multiple 368 00:22:13,380 --> 00:22:17,600 different platforms, like all at once with some, I don't need to pay these four other 369 00:22:17,600 --> 00:22:19,440 people that I have doing it. 370 00:22:19,440 --> 00:22:21,240 You know, I can go down. 371 00:22:21,240 --> 00:22:23,080 Business doesn't want to have less overhead. 372 00:22:23,080 --> 00:22:26,080 So I'm doing my pitch saying, hey, well, how about this? 373 00:22:26,080 --> 00:22:28,600 Let's figure out a contract, do a little quid pro quo. 374 00:22:28,600 --> 00:22:32,200 I'll help you get some of these tools set up and established. 375 00:22:32,200 --> 00:22:33,480 Then you can leverage them. 376 00:22:33,480 --> 00:22:36,040 You put me in a time square billboard, you know? 377 00:22:36,040 --> 00:22:41,320 So it eventually, they kind of brushed that off a little too quickly in my opinion and 378 00:22:41,320 --> 00:22:42,320 was ready. 379 00:22:42,320 --> 00:22:44,320 We're like, all right, so I'm just going to, we're going to get you booked. 380 00:22:44,320 --> 00:22:46,800 I'm going to transfer you over to this person. 381 00:22:46,800 --> 00:22:48,080 They're going to take your credit card information. 382 00:22:48,080 --> 00:22:50,720 I'm like, that's not how any of this stuff works. 383 00:22:50,720 --> 00:22:52,820 That's not how any of this stuff normally works. 384 00:22:52,820 --> 00:22:55,120 So I said, yeah, this is very flattering. 385 00:22:55,120 --> 00:22:56,120 I got to talk to you. 386 00:22:56,120 --> 00:22:58,920 I'm like, I don't want to talk about this just because I'm like, this is not spelling 387 00:22:58,920 --> 00:22:59,920 right. 388 00:22:59,920 --> 00:23:05,200 So rather than look at this myself, I'm like, let me try to form an unbiased opinion. 389 00:23:05,200 --> 00:23:12,640 So I get on God mode, auto GPT, and I act as a podcast producer saying, hey, you're 390 00:23:12,640 --> 00:23:13,800 a podcast producer. 391 00:23:13,800 --> 00:23:18,720 You've just been offered this deal by this company with this podcast, right? 392 00:23:18,720 --> 00:23:19,720 Can you scour the internet? 393 00:23:19,720 --> 00:23:21,120 Tell me if it's legit. 394 00:23:21,120 --> 00:23:25,580 Hey, scour the internet, look at different reviews and comments on those reviews. 395 00:23:25,580 --> 00:23:31,680 If you see things like best thing ever, life changing words like that, label those as positive. 396 00:23:31,680 --> 00:23:35,440 You see things like ripoff scam label those as negative. 397 00:23:35,440 --> 00:23:40,320 So it goes and does that because on the surface everything kind of smells right. 398 00:23:40,320 --> 00:23:43,440 And they've done a careful job at shaping that whole world. 399 00:23:43,440 --> 00:23:51,080 I get the auto GPT results and naturally treated to literally the exact same story that happened 400 00:23:51,080 --> 00:23:52,080 to me. 401 00:23:52,080 --> 00:23:57,640 They told me three days after the interview, Ted, I had a 9.3 rating and I'm like, oh, 402 00:23:57,640 --> 00:23:59,720 script, totally. 403 00:23:59,720 --> 00:24:03,960 But thankfully I had enough sense that I was able to at least let go, I only take a step 404 00:24:03,960 --> 00:24:07,240 back and reconsider and then use those AI tools. 405 00:24:07,240 --> 00:24:10,520 It also then scoured their Apple podcast reviews. 406 00:24:10,520 --> 00:24:13,920 A bunch of them were fake because they all had date stamps at the same time. 407 00:24:13,920 --> 00:24:18,280 So it's like, okay, so you paid a click farm to give you a bunch of five star podcasts. 408 00:24:18,280 --> 00:24:24,680 Plus what podcast is going to have 113 five star reviews and then seven one star reviews 409 00:24:24,680 --> 00:24:30,280 only that are just like beware scam, you know, all this kind of stuff. 410 00:24:30,280 --> 00:24:34,800 That was an example for me on how I had auto GPT kind of bring me down to earth a little 411 00:24:34,800 --> 00:24:37,640 bit to kind of ground me after that. 412 00:24:37,640 --> 00:24:42,040 But I was almost like, I should forward these people to like one of those like ethical hacking, 413 00:24:42,040 --> 00:24:46,960 like you know, scammer time wasting channels that you see on YouTube. 414 00:24:46,960 --> 00:24:52,440 Because I'm like, this is so predatory, you know, was I can't promise you a billboard 415 00:24:52,440 --> 00:24:58,360 in New York, maybe in Berlin, but I think if we focus on actually this podcast and telling 416 00:24:58,360 --> 00:25:00,940 the stories, all of these good things will happen. 417 00:25:00,940 --> 00:25:07,360 But I can tell you, I have flashbacks first of all from us because I was almost caught 418 00:25:07,360 --> 00:25:16,880 in very similar thing just way back when I was 23 and a female founder of a startup in 419 00:25:16,880 --> 00:25:22,640 Denmark and I got contacted by US agency that Oh my God, we will tell your story. 420 00:25:22,640 --> 00:25:24,280 We will bring you on the shows. 421 00:25:24,280 --> 00:25:27,760 We will fly to Copenhagen to interview your team. 422 00:25:27,760 --> 00:25:29,160 And this is so amazing. 423 00:25:29,160 --> 00:25:34,600 So basically, and then the moment a guy was saying, Oh, and your hands are so nice to 424 00:25:34,600 --> 00:25:35,840 hold product. 425 00:25:35,840 --> 00:25:36,840 I was like, what? 426 00:25:36,840 --> 00:25:38,840 That's a little weird. 427 00:25:38,840 --> 00:25:39,840 Right? 428 00:25:39,840 --> 00:25:40,840 It's weird. 429 00:25:40,840 --> 00:25:45,560 And then they're like, and that Tai Lopez is in our network and he's holding this event. 430 00:25:45,560 --> 00:25:47,360 I was like, okay, that's it done. 431 00:25:47,360 --> 00:25:52,640 And for me it's like, but we went through kind of three calls really. 432 00:25:52,640 --> 00:25:54,280 And I was like, thank God. 433 00:25:54,280 --> 00:26:00,160 Now looking back, I was like, Oh my God, thank God, not so stupid 23 year old and didn't 434 00:26:00,160 --> 00:26:01,440 get clicked on that. 435 00:26:01,440 --> 00:26:08,200 But so many people get my email box now from YouTube first way was like so many scams. 436 00:26:08,200 --> 00:26:09,680 I don't click any links. 437 00:26:09,680 --> 00:26:13,720 I don't know, but I can imagine how many people do. 438 00:26:13,720 --> 00:26:19,440 And I can think of many users also of these auto GPT tools where you automate emails and 439 00:26:19,440 --> 00:26:21,120 send emails with certain links. 440 00:26:21,120 --> 00:26:22,120 There's going to be that. 441 00:26:22,120 --> 00:26:23,120 Right. 442 00:26:23,120 --> 00:26:24,120 How about that catfish story? 443 00:26:24,120 --> 00:26:26,440 No, this is so crazy. 444 00:26:26,440 --> 00:26:28,480 I was in so deep. 445 00:26:28,480 --> 00:26:33,120 I was just like in shock, just like I made it for everybody. 446 00:26:33,120 --> 00:26:37,400 And thankfully my wife is just such a grounding force for all this stuff. 447 00:26:37,400 --> 00:26:38,400 Yeah. 448 00:26:38,400 --> 00:26:40,160 You know, brings me back down to her. 449 00:26:40,160 --> 00:26:42,480 She's like, I'm seeing a couple of scammy reviews here. 450 00:26:42,480 --> 00:26:46,120 I'm like, Oh, but there's all these other things. 451 00:26:46,120 --> 00:26:50,000 So I went, did the auto GPT thing and thankfully gave me some more insight. 452 00:26:50,000 --> 00:26:55,960 That's very interesting story that auto this God mode thing, which was also like I out 453 00:26:55,960 --> 00:27:01,720 of all of them, I liked that one the most, but actually can use to also scout internet 454 00:27:01,720 --> 00:27:07,400 very fast to identify certain scammy things, which is like fantastic. 455 00:27:07,400 --> 00:27:11,640 And I just, I wanted something devoid of my own feelings about that at that point, you 456 00:27:11,640 --> 00:27:12,640 know? 457 00:27:12,640 --> 00:27:17,800 So even, even if I saw a negative reviews, I didn't want to be like, Oh, but you know, 458 00:27:17,800 --> 00:27:20,120 rationalizing myself out of it, but maybe not. 459 00:27:20,120 --> 00:27:21,520 I could have a different experience. 460 00:27:21,520 --> 00:27:22,520 You know, I'm special. 461 00:27:22,520 --> 00:27:27,800 I can tell you like this is, you know, well, just don't take it as bragging. 462 00:27:27,800 --> 00:27:28,800 Okay. 463 00:27:28,800 --> 00:27:34,880 No, but a developing like an eye for things and number one thing, if it sounds too good 464 00:27:34,880 --> 00:27:36,760 to be true, it is. 465 00:27:36,760 --> 00:27:44,200 And I literally in my business email, I have self promo scams, fake products, like all 466 00:27:44,200 --> 00:27:47,580 these things go down and it's a lot. 467 00:27:47,580 --> 00:27:52,360 And what I get pissed that I still have to go through all the emails coming to you with 468 00:27:52,360 --> 00:27:59,620 all the promotion and like, all we have is token NFT, like just everything. 469 00:27:59,620 --> 00:28:01,800 And as I said, I never click on links. 470 00:28:01,800 --> 00:28:04,800 If you send me and you're like, Oh, this is our plugin. 471 00:28:04,800 --> 00:28:05,800 Click here. 472 00:28:05,800 --> 00:28:09,000 I'm not going to nothing or like video editors. 473 00:28:09,000 --> 00:28:15,280 So this is like, again, video editor sending Dropbox files, Dropbox things. 474 00:28:15,280 --> 00:28:19,600 And I know this is like number one thing you click on a Dropbox immediately downloads a 475 00:28:19,600 --> 00:28:21,920 thing and then that's it. 476 00:28:21,920 --> 00:28:28,480 One number thing on this fakeness and auto GPT, what is kind of fascinating to me with 477 00:28:28,480 --> 00:28:31,880 like agents, that it creates multiple different agents. 478 00:28:31,880 --> 00:28:37,640 And what if these multiple agents can continue living? 479 00:28:37,640 --> 00:28:39,240 I'm talking like continuous mode. 480 00:28:39,240 --> 00:28:45,040 For example, can auto GPT one of the agents have Instagram account and Instagram personality 481 00:28:45,040 --> 00:28:50,360 plus create fake images like the ones on catfish. 482 00:28:50,360 --> 00:28:52,680 But then let's take one step forward. 483 00:28:52,680 --> 00:28:59,160 I also back in the days, like a year ago made a video about will architects or AI create 484 00:28:59,160 --> 00:29:01,580 content in metaverse. 485 00:29:01,580 --> 00:29:06,760 And I had this idea that actually you will have this like companions which are going 486 00:29:06,760 --> 00:29:08,680 to be AI driven. 487 00:29:08,680 --> 00:29:15,800 So now imagine this like AI agents having a real 3D body face that you can actually 488 00:29:15,800 --> 00:29:17,680 talk with them. 489 00:29:17,680 --> 00:29:22,680 What if you're like just talking and there is like screens and this faces who actually 490 00:29:22,680 --> 00:29:25,280 interact with you and you see them working. 491 00:29:25,280 --> 00:29:31,920 So it's more like we as people become like this managers with maybe five, 10 assistants 492 00:29:31,920 --> 00:29:33,180 in the future. 493 00:29:33,180 --> 00:29:39,480 And we are overviewing and kind of helping them navigate and providing feedback just 494 00:29:39,480 --> 00:29:43,000 basically becomes like parents to these little agents. 495 00:29:43,000 --> 00:29:44,480 Yeah, it's baby AGI. 496 00:29:44,480 --> 00:29:45,960 We've got to be parents. 497 00:29:45,960 --> 00:29:46,960 Right. 498 00:29:46,960 --> 00:29:52,260 So yeah, the future, no one knows what type of future we are looking into. 499 00:29:52,260 --> 00:29:58,700 So going forward, Wes, do you think now that you will fall for scams like the one you fell 500 00:29:58,700 --> 00:30:01,880 for or the images? 501 00:30:01,880 --> 00:30:08,400 And will we see on this podcast on upcoming episodes or you're transitioning to some big, 502 00:30:08,400 --> 00:30:10,080 big podcasting studio? 503 00:30:10,080 --> 00:30:13,000 Now I'm right here with you faithful listeners. 504 00:30:13,000 --> 00:30:14,000 Yay! 505 00:30:14,000 --> 00:30:16,200 And gee, there's no way I could ever leave you behind. 506 00:30:16,200 --> 00:30:17,960 You're not leaving me. 507 00:30:17,960 --> 00:30:22,400 I am definitely, I'm constantly amazed every day. 508 00:30:22,400 --> 00:30:26,560 If something hits the right emotional heartstrings, then yeah, I may, you know, they probably 509 00:30:26,560 --> 00:30:30,320 play me like a fiddle again, but ultimately I'm still here with you all. 510 00:30:30,320 --> 00:30:34,520 So with that, I am Wes the Synthbind for Go2Go. 511 00:30:34,520 --> 00:30:37,920 We're going to close out this week with a happy prompting everybody. 512 00:30:37,920 --> 00:30:41,960 Happy prompting everybody. 513 00:30:41,960 --> 00:30:49,380 Thanks for listening to How to Talk to AI with your hosts Go2Go and Wes the Synthmind. 514 00:30:49,380 --> 00:30:56,280 As always, you can check out the show notes and links at howtotalkto.ai. 515 00:30:56,280 --> 00:30:58,640 That's all for this week's episode. 516 00:30:58,640 --> 00:31:12,840 Happy prompting everyone.