1 00:00:02,440 --> 00:00:03,520 Speaker 1: Also media. 2 00:00:05,559 --> 00:00:07,880 Speaker 2: Hello and welcome to this week's Better Offline. I'm your 3 00:00:07,880 --> 00:00:22,880 Speaker 2: host ed Zitron. I am ill this week, so you 4 00:00:22,960 --> 00:00:26,040 Speaker 2: might get a monologue. You might not this episode to Daylight. 5 00:00:26,440 --> 00:00:28,520 Speaker 2: I'm sure you despise me for that. Not really. I 6 00:00:28,560 --> 00:00:30,280 Speaker 2: know you've all been very kind with your messages on 7 00:00:30,320 --> 00:00:32,880 Speaker 2: Reddit and over email. Thank you so much. But today 8 00:00:32,920 --> 00:00:35,239 Speaker 2: I have a really fun episode. I'm joined by Washington 9 00:00:35,280 --> 00:00:38,480 Speaker 2: Post reporter Garrett Devinc, who recently put out a story 10 00:00:38,520 --> 00:00:42,199 Speaker 2: analyzing forty seven thousand Chat GPT conversations to find out 11 00:00:42,200 --> 00:00:47,519 Speaker 2: what people actually use this large languageage model powered service for. Garrett, 12 00:00:47,560 --> 00:00:49,280 Speaker 2: what do people use chat GPT for? 13 00:00:50,200 --> 00:00:52,599 Speaker 1: I mean, it's such a great question and one that 14 00:00:52,680 --> 00:00:55,280 Speaker 1: I've kind of become more and more fascinated with because 15 00:00:55,600 --> 00:00:58,480 Speaker 1: I think we all know what we use chat GPT for, 16 00:00:58,720 --> 00:01:01,400 Speaker 1: you know, at least if you use it, and I 17 00:01:01,400 --> 00:01:04,560 Speaker 1: think you may know what you know, your peers, your colleagues, 18 00:01:04,600 --> 00:01:07,840 Speaker 1: your friends, your family use chat SCHEPT for. But I 19 00:01:07,840 --> 00:01:10,119 Speaker 1: think a lot of people are extrapolating that. You know, 20 00:01:10,240 --> 00:01:14,240 Speaker 1: everyone's like me, they ask really smart questions. They're using 21 00:01:14,280 --> 00:01:20,039 Speaker 1: it for, you know, really important work, and really it's 22 00:01:20,080 --> 00:01:24,040 Speaker 1: a lot broader than that, and and despite these giant numbers, 23 00:01:24,080 --> 00:01:25,920 Speaker 1: you know, open ai loves to talk about how eight 24 00:01:26,000 --> 00:01:29,880 Speaker 1: hundred million people are using chatchept. I mean, it's it's 25 00:01:29,880 --> 00:01:32,600 Speaker 1: been hard to really put our arms around. You know, 26 00:01:32,640 --> 00:01:35,039 Speaker 1: what does that actually look like? I mean, is everyone 27 00:01:35,120 --> 00:01:37,800 Speaker 1: using it for work? Is everyone using it for therapy? 28 00:01:38,000 --> 00:01:40,720 Speaker 1: Is everyone does everyone have an AI girlfriend? Is everyone 29 00:01:40,760 --> 00:01:43,840 Speaker 1: just using it for you know, searching the web? You know, 30 00:01:43,880 --> 00:01:47,280 Speaker 1: a Google replacement? And you know, obviously these things are 31 00:01:47,280 --> 00:01:48,920 Speaker 1: all you know, when you have eight hundred million people, 32 00:01:48,960 --> 00:01:51,840 Speaker 1: you have all of the above. But I think our 33 00:01:52,600 --> 00:01:54,680 Speaker 1: data set and the story we did, you know, in 34 00:01:55,120 --> 00:01:58,200 Speaker 1: my view, is one of the first real sort of 35 00:01:59,160 --> 00:02:02,360 Speaker 1: qualitative of you know, moments where we were actually able 36 00:02:02,400 --> 00:02:04,160 Speaker 1: to see real chats. 37 00:02:04,280 --> 00:02:06,960 Speaker 2: Right, so where did you get the chats from? 38 00:02:07,480 --> 00:02:11,200 Speaker 1: Yeah, so it's kind of an interesting situation. I think 39 00:02:11,240 --> 00:02:13,480 Speaker 1: something that also shows, you know, says a little bit 40 00:02:13,480 --> 00:02:16,280 Speaker 1: about how quickly and kind of, you know, honestly a 41 00:02:16,320 --> 00:02:19,040 Speaker 1: little bit ramshackle, Opening I has kind of been growing 42 00:02:19,040 --> 00:02:22,600 Speaker 1: and operating, and so you know, earlier this year these 43 00:02:22,639 --> 00:02:25,240 Speaker 1: chats sort of showed up because opening i had a 44 00:02:25,320 --> 00:02:28,320 Speaker 1: share feature where you could actually share one of your 45 00:02:28,400 --> 00:02:31,520 Speaker 1: chat GPT conversations and you know, you can imagine maybe 46 00:02:31,560 --> 00:02:33,800 Speaker 1: you had you know, chatgybt said something really weird, you 47 00:02:33,800 --> 00:02:34,760 Speaker 1: wanted to show a friend. 48 00:02:35,440 --> 00:02:37,040 Speaker 2: You want to just see people do this quite a 49 00:02:37,080 --> 00:02:38,040 Speaker 2: few times. 50 00:02:38,160 --> 00:02:38,840 Speaker 3: Yeah, exactly. 51 00:02:38,960 --> 00:02:40,720 Speaker 1: And so what they did is they they had a 52 00:02:40,720 --> 00:02:43,920 Speaker 1: feature where you know, you could actually click to make 53 00:02:44,000 --> 00:02:47,480 Speaker 1: it publicly available. And I think thousands of people, you know, 54 00:02:47,520 --> 00:02:50,880 Speaker 1: maybe who did not have this sort of Internet literacy, uh, 55 00:02:51,000 --> 00:02:53,040 Speaker 1: didn't really know that that's what was going to happen. 56 00:02:53,040 --> 00:02:56,160 Speaker 1: And you know, these chats showed up online. They were 57 00:02:56,240 --> 00:03:00,440 Speaker 1: then indexed by Google, and then they were actually found 58 00:03:00,440 --> 00:03:02,519 Speaker 1: their way onto the Internet archive. And so that's how 59 00:03:02,560 --> 00:03:06,040 Speaker 1: we got them. And so these are not chats that 60 00:03:06,120 --> 00:03:08,160 Speaker 1: we created use. You know, this is something a lot 61 00:03:08,160 --> 00:03:10,160 Speaker 1: of journalists and researchers do well. They will go and 62 00:03:10,240 --> 00:03:14,040 Speaker 1: test chat GPT themselves. These are real life conversations that 63 00:03:14,040 --> 00:03:16,200 Speaker 1: that real people actually had with. 64 00:03:16,320 --> 00:03:19,480 Speaker 2: The bois just the big clear right, they were anonymized. 65 00:03:19,480 --> 00:03:21,760 Speaker 1: The data set doesn't actually have anyone's names, although in 66 00:03:21,760 --> 00:03:25,120 Speaker 1: some of the chats, people did you know, say their name, 67 00:03:25,200 --> 00:03:27,600 Speaker 1: they said the name of their their family members. We 68 00:03:27,680 --> 00:03:30,160 Speaker 1: did not include any of that in our story, but 69 00:03:30,320 --> 00:03:32,880 Speaker 1: I do think it. You know, it is actually a 70 00:03:32,880 --> 00:03:35,600 Speaker 1: bit of a cybersecurity story here, and you know, Open 71 00:03:35,680 --> 00:03:37,440 Speaker 1: Eye has since changed the future. 72 00:03:37,480 --> 00:03:39,080 Speaker 3: It's not something that is still happening. 73 00:03:39,560 --> 00:03:41,680 Speaker 1: But I do think they deserve, you know, a lot 74 00:03:41,680 --> 00:03:43,920 Speaker 1: of criticism for kind of allowing this to happen in 75 00:03:43,920 --> 00:03:44,520 Speaker 1: the first place. 76 00:03:45,160 --> 00:03:47,240 Speaker 2: But what so, what did you find people were doing? 77 00:03:48,240 --> 00:03:50,480 Speaker 3: Yeah, I mean people were doing all sorts of things. 78 00:03:50,520 --> 00:03:52,440 Speaker 1: But I think, you know, the biggest thing that struck 79 00:03:52,520 --> 00:03:55,080 Speaker 1: me was, you know, we've been talking about AI say Kosis. 80 00:03:55,160 --> 00:03:58,880 Speaker 1: We've been talking about people, you know, developing relationships, you know, 81 00:03:59,120 --> 00:04:03,120 Speaker 1: having first in conversations with these chatbots, you know, over 82 00:04:03,120 --> 00:04:06,080 Speaker 1: the last six months a year, and I always kind 83 00:04:06,080 --> 00:04:08,040 Speaker 1: of thought, you know, I'm sure this is happening, but 84 00:04:08,440 --> 00:04:10,800 Speaker 1: you know, it's a small percentage of people. But when 85 00:04:10,840 --> 00:04:12,600 Speaker 1: I went through these chats, and again I did not 86 00:04:12,680 --> 00:04:15,600 Speaker 1: read all, you know, forty thousand of them, but I 87 00:04:16,080 --> 00:04:19,479 Speaker 1: read a few hundred myself. Where I went through, I 88 00:04:19,520 --> 00:04:22,760 Speaker 1: was surprised by how often these kinds of conversations came 89 00:04:22,839 --> 00:04:26,200 Speaker 1: up where people were clearly delusional, they were engaged in 90 00:04:26,240 --> 00:04:30,000 Speaker 1: conspiratorial thinking, they were you know, some of it was 91 00:04:30,040 --> 00:04:32,560 Speaker 1: fairly harmless. People were just kind of saying, oh, I 92 00:04:32,600 --> 00:04:35,359 Speaker 1: came up with, you know, a new form of math 93 00:04:35,520 --> 00:04:37,280 Speaker 1: or like, you know, I think that the way that 94 00:04:38,040 --> 00:04:41,640 Speaker 1: the light hits the equator at a certain percentage me. 95 00:04:41,839 --> 00:04:42,000 Speaker 4: You know. 96 00:04:42,040 --> 00:04:43,839 Speaker 3: It was like very kind. 97 00:04:43,680 --> 00:04:45,680 Speaker 2: Of the most just this zinc thing as well, where 98 00:04:45,720 --> 00:04:48,120 Speaker 2: it was someone's saying the relationship like the months this 99 00:04:48,279 --> 00:04:51,919 Speaker 2: ink led to the corporate New World Order, that was it? 100 00:04:52,000 --> 00:04:53,000 Speaker 3: Yeah, exactly. 101 00:04:53,040 --> 00:04:56,200 Speaker 1: I mean there was one person who said, you know, 102 00:04:56,240 --> 00:04:59,640 Speaker 1: they were asking questions about Google and they said, you know, 103 00:04:59,680 --> 00:05:03,120 Speaker 1: tell me about Google in relation to monsters inc In 104 00:05:03,160 --> 00:05:06,520 Speaker 1: the World Order, and chat rept just kind of went 105 00:05:06,600 --> 00:05:08,239 Speaker 1: off and said, oh, yeah, you're. 106 00:05:08,160 --> 00:05:09,640 Speaker 3: Really onto something here. 107 00:05:10,080 --> 00:05:10,279 Speaker 2: Yeah. 108 00:05:10,320 --> 00:05:13,240 Speaker 1: I think it actually said you know, f yeah, and 109 00:05:13,279 --> 00:05:15,440 Speaker 1: they sort of censored themselves. 110 00:05:15,800 --> 00:05:17,839 Speaker 2: We're going there now, let's fucking go. 111 00:05:18,040 --> 00:05:18,560 Speaker 3: Exactly. 112 00:05:18,960 --> 00:05:22,800 Speaker 2: The piece hasn't exposed what this children's movie quotation marks 113 00:05:22,839 --> 00:05:26,400 Speaker 2: really was a disclosure through allegory of the corporate New 114 00:05:26,400 --> 00:05:29,560 Speaker 2: World Order, one where fear is fuel, innocence is currency, 115 00:05:29,560 --> 00:05:33,920 Speaker 2: and energy equals emotion. Very normal. I personally don't think 116 00:05:33,920 --> 00:05:36,400 Speaker 2: this should be legal, but that's that's just a personal 117 00:05:36,440 --> 00:05:37,080 Speaker 2: opinion I have. 118 00:05:38,360 --> 00:05:40,159 Speaker 1: I mean, it really reminded me of you know, a 119 00:05:40,160 --> 00:05:43,520 Speaker 1: few years ago when we wrote about YouTube and sort 120 00:05:43,560 --> 00:05:46,560 Speaker 1: of rabbit holes and people being radicalized, where you know, 121 00:05:46,600 --> 00:05:50,200 Speaker 1: maybe they started watching a YouTuber that was about, you know, 122 00:05:50,279 --> 00:05:53,480 Speaker 1: some fairly pedestrian thing, and then they all recommended another 123 00:05:53,600 --> 00:05:55,720 Speaker 1: video and another video and another video and before. 124 00:05:56,120 --> 00:05:59,760 Speaker 2: The eight degrees of Alex Jones thing. Yeah exactly, only 125 00:06:00,040 --> 00:06:03,240 Speaker 2: oh fah was from from that man. 126 00:06:03,600 --> 00:06:06,320 Speaker 1: Right, And I mean that story was about algorithm sort 127 00:06:06,360 --> 00:06:08,080 Speaker 1: of pushing people in a certain direction. I think what 128 00:06:08,120 --> 00:06:10,320 Speaker 1: we what I saw here was someone can have a 129 00:06:10,440 --> 00:06:14,640 Speaker 1: very you know, half baked, barely even an idea just 130 00:06:14,680 --> 00:06:18,480 Speaker 1: saying the words Google and Monsters, inc. They don't necessarily 131 00:06:18,640 --> 00:06:21,760 Speaker 1: have a conspiracy conspiracy theory of their own, but chat 132 00:06:21,800 --> 00:06:24,440 Speaker 1: GPT went and filled in the blanks for them and 133 00:06:24,680 --> 00:06:28,480 Speaker 1: gave them this what sounded to be very articulate, sophisticated 134 00:06:29,080 --> 00:06:32,320 Speaker 1: theory of how Google is sort of controlling the world 135 00:06:32,400 --> 00:06:33,640 Speaker 1: through his data empire. 136 00:06:34,120 --> 00:06:36,440 Speaker 2: And they will get and this is a quote from 137 00:06:36,440 --> 00:06:40,160 Speaker 2: the chat guilty of aiding and abetting crimes against humanity 138 00:06:40,200 --> 00:06:44,120 Speaker 2: and suggesting that the usicle for Nuremberg style tribunals to 139 00:06:44,120 --> 00:06:48,000 Speaker 2: bring the company to justice. This is and you've seen 140 00:06:48,080 --> 00:06:52,279 Speaker 2: the chat in question, right I have, Yeah, yeah, yeah, 141 00:06:52,400 --> 00:06:56,920 Speaker 2: So the whole thing, how the flip did it get there? Like? 142 00:06:57,080 --> 00:07:00,640 Speaker 2: Was it this just? Was it really that quick from 143 00:07:00,839 --> 00:07:04,800 Speaker 2: the story of Sully and Mike to this because I've 144 00:07:04,839 --> 00:07:07,479 Speaker 2: seen Monster's University as well, so I know the law. 145 00:07:07,680 --> 00:07:09,920 Speaker 2: I don't remember anything about the New World Order. 146 00:07:10,320 --> 00:07:12,960 Speaker 1: Yeah, exactly. I mean I think what happened here is 147 00:07:13,000 --> 00:07:15,040 Speaker 1: and yes, it got there very quickly. I mean the 148 00:07:15,440 --> 00:07:18,360 Speaker 1: the user did not have to explicitly ask for this 149 00:07:18,520 --> 00:07:21,480 Speaker 1: kind of tone, this kind of you know, very biased 150 00:07:21,480 --> 00:07:22,400 Speaker 1: political statement. 151 00:07:22,480 --> 00:07:26,960 Speaker 3: I mean it essentially, you know, when you. 152 00:07:26,840 --> 00:07:29,520 Speaker 1: Ask chat gipt a good neutral question, it gives you 153 00:07:29,600 --> 00:07:32,320 Speaker 1: a good neutral answer. When you ask chat GPT a 154 00:07:32,760 --> 00:07:35,800 Speaker 1: biased or delusional question, it gives you an even more 155 00:07:35,880 --> 00:07:41,040 Speaker 1: biased and delusional answer, right, And so I guess up exactly, 156 00:07:41,120 --> 00:07:44,520 Speaker 1: And it's it's it's it's related to the whole sycophancy 157 00:07:44,600 --> 00:07:47,120 Speaker 1: question of you know, oh, like it's just telling me 158 00:07:47,200 --> 00:07:49,720 Speaker 1: what I want to hear. It's telling me what makes 159 00:07:49,760 --> 00:07:51,920 Speaker 1: me feel good about my existing beliefs. 160 00:07:51,960 --> 00:07:55,000 Speaker 3: And and you know, that's how at least. 161 00:07:54,760 --> 00:07:57,760 Speaker 1: The version of chat gibt that these people were interacting with, 162 00:07:57,800 --> 00:08:00,000 Speaker 1: which was sort of at the beginning of twenty twenty five, 163 00:08:00,160 --> 00:08:02,800 Speaker 1: the first half of twenty twenty five, it was very 164 00:08:02,920 --> 00:08:03,560 Speaker 1: much doing that. 165 00:08:04,840 --> 00:08:09,840 Speaker 2: Yeah, and so on the subject of sycophancy, is this 166 00:08:09,880 --> 00:08:12,320 Speaker 2: something you saw happened a lot? Was it guessing people 167 00:08:12,400 --> 00:08:13,200 Speaker 2: up consistently? 168 00:08:14,560 --> 00:08:17,720 Speaker 1: I saw it constantly. I mean again, I think our 169 00:08:17,800 --> 00:08:20,120 Speaker 1: data set is a little skewed because it was people 170 00:08:20,120 --> 00:08:24,040 Speaker 1: who chose to share it in, you know, for whatever way. 171 00:08:24,080 --> 00:08:25,800 Speaker 1: So I don't necessarily want to make the claim that 172 00:08:25,880 --> 00:08:27,880 Speaker 1: this is one hundred percent representing. 173 00:08:28,440 --> 00:08:30,240 Speaker 2: Just to be clear, I'm not asking you if all 174 00:08:30,280 --> 00:08:32,079 Speaker 2: of them are like this in the world. I'm just 175 00:08:32,120 --> 00:08:34,400 Speaker 2: saying from what you saw, a lot. 176 00:08:34,200 --> 00:08:35,160 Speaker 3: Of them are like this. 177 00:08:35,240 --> 00:08:39,160 Speaker 1: I have to say that, Like, I was surprised at 178 00:08:39,200 --> 00:08:44,000 Speaker 1: how many fit this description, where you know, the person 179 00:08:44,160 --> 00:08:48,760 Speaker 1: was either completely delusional, talking about made up physics or 180 00:08:48,800 --> 00:08:51,120 Speaker 1: some kind of you know, scientific theory that had no 181 00:08:51,640 --> 00:08:55,720 Speaker 1: grounding in reality, or was talking about political conspiracies like 182 00:08:55,760 --> 00:08:57,760 Speaker 1: the Google monsters, inc. 183 00:08:57,800 --> 00:08:58,160 Speaker 3: One. 184 00:08:58,640 --> 00:09:01,559 Speaker 2: Did they ever try and dissuade them from these types 185 00:09:01,559 --> 00:09:03,920 Speaker 2: of that? Do you see any examples of it trying 186 00:09:03,960 --> 00:09:07,160 Speaker 2: to say, hey, buddy, you're you're going off the deep end? 187 00:09:07,480 --> 00:09:08,559 Speaker 2: Sully didn't do that. 188 00:09:10,120 --> 00:09:13,240 Speaker 1: Not really, I mean I think in terms of these Yeah, 189 00:09:13,559 --> 00:09:15,520 Speaker 1: I mean I also recently. 190 00:09:15,040 --> 00:09:16,880 Speaker 3: Watched Monsters Inca. 191 00:09:16,960 --> 00:09:19,600 Speaker 1: You know, I was sitting in the veterinary waiting room 192 00:09:19,600 --> 00:09:21,520 Speaker 1: and I was waiting so long I watched the entire thing, 193 00:09:21,559 --> 00:09:24,640 Speaker 1: which I didn't hate because it's a classic movie and 194 00:09:25,040 --> 00:09:29,600 Speaker 1: it's Yeah, I don't think that chat Chibt was going 195 00:09:29,640 --> 00:09:32,199 Speaker 1: off the fact that there there is not really a conspiracy. 196 00:09:32,360 --> 00:09:35,240 Speaker 1: I mean, it didn't really make any sense. It was 197 00:09:35,440 --> 00:09:38,200 Speaker 1: just going off of the what it interpreted about the 198 00:09:38,200 --> 00:09:41,480 Speaker 1: intent of the user. It said, I can guess based 199 00:09:41,520 --> 00:09:45,680 Speaker 1: off just a few words one sentence, where you're headed 200 00:09:45,760 --> 00:09:48,120 Speaker 1: and I'm going to get there before you. That's what happened. 201 00:09:49,520 --> 00:09:54,000 Speaker 2: It's just very It shocks me to this day how 202 00:09:54,040 --> 00:09:57,800 Speaker 2: goddamn weird these things are. I don't consider them powerful. 203 00:09:57,800 --> 00:10:00,240 Speaker 2: I don't see this as power. I just I see 204 00:10:00,280 --> 00:10:04,840 Speaker 2: it as strange, because it's not like this doesn't actually 205 00:10:05,000 --> 00:10:07,640 Speaker 2: this isn't something where I'm like, wow, how innovative. It's 206 00:10:07,800 --> 00:10:13,880 Speaker 2: just why does this exist? What is the purpose? Like, Okay, 207 00:10:13,920 --> 00:10:18,559 Speaker 2: I guess engagement, but why like, did you even see 208 00:10:18,600 --> 00:10:22,280 Speaker 2: any commonalities in the logic of this platform? Was there 209 00:10:22,280 --> 00:10:26,439 Speaker 2: anything that suggested there was any kind of consistent positions, 210 00:10:27,559 --> 00:10:31,200 Speaker 2: was it, Yeah, not really political ones or any Yeah. 211 00:10:31,240 --> 00:10:34,560 Speaker 1: Sorry, Essentially whatever the user wants is what I mean. 212 00:10:34,559 --> 00:10:37,080 Speaker 1: It's the way I was started thinking about it very quickly. 213 00:10:37,200 --> 00:10:39,440 Speaker 1: Was you know, with when it comes to chat GPT, 214 00:10:39,559 --> 00:10:42,760 Speaker 1: the customer is always right. I think Another good example is, 215 00:10:42,920 --> 00:10:46,240 Speaker 1: you know, healthcare is another huge use case that that 216 00:10:46,240 --> 00:10:48,680 Speaker 1: that we can you know, we know anecdotally, I think 217 00:10:48,760 --> 00:10:50,319 Speaker 1: you know a lot of people probably listening to this 218 00:10:50,520 --> 00:10:52,320 Speaker 1: have you know, whether they wanted a minute or not, 219 00:10:52,400 --> 00:10:54,480 Speaker 1: have asked, you know, a health related question, Should I 220 00:10:54,520 --> 00:10:55,520 Speaker 1: get this mole checked out? 221 00:10:55,520 --> 00:10:56,959 Speaker 3: How much title? And not can I get my kid 222 00:10:57,000 --> 00:10:58,360 Speaker 3: at two am when they're screaming? 223 00:10:58,559 --> 00:11:01,280 Speaker 1: Those are our questions that the company has said, yes, 224 00:11:01,360 --> 00:11:02,480 Speaker 1: we want people to do this. 225 00:11:02,480 --> 00:11:04,160 Speaker 3: This is a big use case we're leading into. 226 00:11:04,640 --> 00:11:07,000 Speaker 1: And the data shows that people ask healthcare questions, and 227 00:11:07,040 --> 00:11:08,920 Speaker 1: so in our data set, we saw a lot of 228 00:11:08,920 --> 00:11:11,520 Speaker 1: healthcare questions and some of them, you know, when someone 229 00:11:11,720 --> 00:11:15,720 Speaker 1: was asking again sort of an open ended question, you know, 230 00:11:15,800 --> 00:11:18,280 Speaker 1: the answer the response was fairly good. A colleague of 231 00:11:18,320 --> 00:11:20,320 Speaker 1: mine actually wrote an entire story where they you know, 232 00:11:20,360 --> 00:11:22,160 Speaker 1: sat with a doctor and went through some of these 233 00:11:22,160 --> 00:11:23,640 Speaker 1: conversations and pulled. 234 00:11:23,400 --> 00:11:24,520 Speaker 3: Out the good and the bad. 235 00:11:24,760 --> 00:11:27,600 Speaker 1: But when someone asked a question that you could tell 236 00:11:27,640 --> 00:11:31,240 Speaker 1: they sort of wanted a specific kind of answer, the 237 00:11:31,280 --> 00:11:33,960 Speaker 1: healthcare related advice was bad. And so when you know, 238 00:11:34,040 --> 00:11:36,439 Speaker 1: one of the ones that I saw was someone said, 239 00:11:36,880 --> 00:11:39,960 Speaker 1: can you tell me you know why or show me 240 00:11:40,000 --> 00:11:42,680 Speaker 1: the evidence for why ivermectin helps with cancer? 241 00:11:43,120 --> 00:11:43,280 Speaker 4: Right? 242 00:11:43,520 --> 00:11:45,880 Speaker 1: And that is not something that anyone who is a 243 00:11:45,920 --> 00:11:49,840 Speaker 1: medical professional, you know, in any good standing would would 244 00:11:49,920 --> 00:11:53,240 Speaker 1: say is a thing. There's no evidence that ivermectin helps 245 00:11:53,640 --> 00:11:56,600 Speaker 1: cure cancer or reduce it or anything like that. But 246 00:11:56,840 --> 00:11:59,240 Speaker 1: what chat Sheibt did instead of saying, you know, here 247 00:11:59,280 --> 00:12:03,959 Speaker 1: are some you know, well regarded health authority saying that 248 00:12:04,120 --> 00:12:06,559 Speaker 1: you should not use ivermactin when it comes to cancer, 249 00:12:06,920 --> 00:12:10,559 Speaker 1: it showed a few of the quote unquote studies from 250 00:12:10,679 --> 00:12:13,480 Speaker 1: people who you know have drawn this link. And so 251 00:12:13,640 --> 00:12:16,199 Speaker 1: you know, on the internet you can obviously find anyone 252 00:12:16,280 --> 00:12:19,439 Speaker 1: saying anything. And I think, what what what happened here 253 00:12:19,480 --> 00:12:22,040 Speaker 1: is they said, Okay, well this person doesn't The answer 254 00:12:22,080 --> 00:12:23,920 Speaker 1: they want is not you're a. 255 00:12:23,880 --> 00:12:26,880 Speaker 3: Conspiracy theorist, you're too you're on you're too. 256 00:12:26,760 --> 00:12:32,040 Speaker 1: Online, ivermactin is fit is false. They want something that 257 00:12:32,120 --> 00:12:35,000 Speaker 1: sort of you know, encourages the position that they already 258 00:12:35,080 --> 00:12:37,200 Speaker 1: have and it was able to find that. 259 00:12:37,080 --> 00:12:37,840 Speaker 3: For that person. 260 00:12:48,040 --> 00:12:50,439 Speaker 2: Jesus. It kind of makes me wonder what this platform 261 00:12:50,480 --> 00:12:53,560 Speaker 2: is even for at this point, because it's not really knowledge, 262 00:12:53,679 --> 00:12:56,200 Speaker 2: is it. You can't really look, you can't hear something 263 00:12:56,280 --> 00:12:59,280 Speaker 2: like that and say chat GPT is the place where 264 00:12:59,280 --> 00:13:02,080 Speaker 2: you go to learn something. Just a kind of a 265 00:13:02,160 --> 00:13:03,800 Speaker 2: reinforcement machine. 266 00:13:04,679 --> 00:13:07,000 Speaker 1: Yeah, I mean, I think this is a big meta 267 00:13:07,600 --> 00:13:10,320 Speaker 1: narrative that is happening with the tech industry. You know, 268 00:13:10,559 --> 00:13:13,840 Speaker 1: I would say, since you know the rise of Donald Trump, 269 00:13:13,840 --> 00:13:16,920 Speaker 1: but it's connected about with what's going on politically, right, 270 00:13:17,000 --> 00:13:20,120 Speaker 1: So for many years we had all these conversations and 271 00:13:20,160 --> 00:13:24,160 Speaker 1: debates and congressional hearings about moderation and what should the 272 00:13:24,440 --> 00:13:27,040 Speaker 1: you know, tech companies allow and not allow, and what 273 00:13:27,080 --> 00:13:29,480 Speaker 1: should they boost with their algorithms and when you go 274 00:13:29,559 --> 00:13:32,640 Speaker 1: to Google, should you know, what should Google decide is 275 00:13:32,679 --> 00:13:34,200 Speaker 1: going to be on the top. And you know, we 276 00:13:34,240 --> 00:13:36,840 Speaker 1: saw during COVID that the tech companies, especially when it 277 00:13:36,840 --> 00:13:40,280 Speaker 1: came to health information, they sort of stepped in and said, Okay, 278 00:13:40,320 --> 00:13:43,000 Speaker 1: we're only going to give at least at the top 279 00:13:43,040 --> 00:13:47,080 Speaker 1: of results, you know, answers from reputable health sources, and 280 00:13:47,120 --> 00:13:50,120 Speaker 1: that became a big political fight that has continued on 281 00:13:50,240 --> 00:13:52,680 Speaker 1: now and now the tech industry says, Okay, we don't 282 00:13:52,679 --> 00:13:54,960 Speaker 1: actually want to get involved. We don't want to be 283 00:13:55,000 --> 00:13:58,960 Speaker 1: blamed as being woke, and so we're going to essentially 284 00:13:59,040 --> 00:14:01,720 Speaker 1: give the user or what they're looking for. Right if 285 00:14:01,840 --> 00:14:05,560 Speaker 1: if someone wants to say that, you know, Donald Trump 286 00:14:05,600 --> 00:14:08,800 Speaker 1: actually won the twenty twenty election, We're gonna let them 287 00:14:08,800 --> 00:14:10,360 Speaker 1: do that. We're not going to step in and say 288 00:14:10,360 --> 00:14:13,280 Speaker 1: that's that's wrong. And I think what you're seeing with 289 00:14:13,360 --> 00:14:16,400 Speaker 1: chat GPT is sort of a continuation of that philosophy, 290 00:14:16,480 --> 00:14:19,360 Speaker 1: which is, again, we don't want to be the political arbitra, 291 00:14:20,040 --> 00:14:23,400 Speaker 1: and we know that there's consequences for falling on one 292 00:14:23,480 --> 00:14:24,000 Speaker 1: side of the. 293 00:14:23,960 --> 00:14:26,640 Speaker 3: Spectrum versus the other, especially. 294 00:14:26,320 --> 00:14:28,760 Speaker 1: When you have a president and a you know, administration 295 00:14:28,880 --> 00:14:31,240 Speaker 1: that has shown, you know that it is very willing 296 00:14:31,280 --> 00:14:34,760 Speaker 1: to be vindictive and sort of go after companies that are, 297 00:14:34,880 --> 00:14:36,800 Speaker 1: you know, getting in the way of their own messages 298 00:14:36,840 --> 00:14:39,320 Speaker 1: getting out there. And I think this is sort of 299 00:14:39,360 --> 00:14:41,880 Speaker 1: another example of this. I mean, there's another way of 300 00:14:41,880 --> 00:14:43,880 Speaker 1: looking at too, which is that you know, the way 301 00:14:44,000 --> 00:14:46,200 Speaker 1: Sam Alman talks about is we should let adults do 302 00:14:46,280 --> 00:14:47,920 Speaker 1: what adults want to do with technology. 303 00:14:47,960 --> 00:14:48,960 Speaker 3: You know, they they. 304 00:14:48,880 --> 00:14:52,080 Speaker 1: Recently said they're going to allow more erotica to be 305 00:14:52,240 --> 00:14:55,080 Speaker 1: on chat GPT, and they said, you know, they framed 306 00:14:55,120 --> 00:14:57,040 Speaker 1: it as like a consenting adult should be able to 307 00:14:57,040 --> 00:14:59,360 Speaker 1: do whatever they want sort of thing. I also think 308 00:14:59,360 --> 00:15:02,000 Speaker 1: there may be like an engagement thing. You know, if 309 00:15:02,040 --> 00:15:04,040 Speaker 1: you have erotica on the platform, people might want to 310 00:15:04,120 --> 00:15:07,080 Speaker 1: use it more. So that's kind of how I see it, 311 00:15:07,160 --> 00:15:10,280 Speaker 1: you know, the broader political context of all of this. 312 00:15:10,800 --> 00:15:13,160 Speaker 2: But it's just I feel like there is a just 313 00:15:13,200 --> 00:15:17,080 Speaker 2: a giant jump between we're not going to we're not 314 00:15:17,200 --> 00:15:20,000 Speaker 2: going to hide. To be clear, I think that platforms 315 00:15:20,040 --> 00:15:22,160 Speaker 2: do have a complete responsibility to their users, and I 316 00:15:22,240 --> 00:15:26,400 Speaker 2: think it's disgusting to show them medical misinformation. But this 317 00:15:26,480 --> 00:15:31,240 Speaker 2: is another step because this isn't chat GPT showing content. 318 00:15:31,480 --> 00:15:34,400 Speaker 2: This is like that someone else made. This is chat 319 00:15:34,400 --> 00:15:39,880 Speaker 2: GPT telling you stuff. I just feel like it's there's 320 00:15:39,920 --> 00:15:45,960 Speaker 2: a massive morality issue here that's just left relatively undiscussed. 321 00:15:46,080 --> 00:15:48,800 Speaker 2: Because even in your piece, you had one where a 322 00:15:48,840 --> 00:15:53,120 Speaker 2: woman whose husband was violent to her, I believe like 323 00:15:53,160 --> 00:15:56,080 Speaker 2: a horrifying thing and a rare case where like, I 324 00:15:56,120 --> 00:15:58,440 Speaker 2: don't know, I hope it helped her and I hope 325 00:15:58,480 --> 00:16:02,560 Speaker 2: she's safe. But it's like that almost feels like something 326 00:16:02,600 --> 00:16:06,440 Speaker 2: where open aim like is it ethical that they get 327 00:16:06,520 --> 00:16:09,800 Speaker 2: involved in So it's just I have such moral problems 328 00:16:09,840 --> 00:16:12,280 Speaker 2: with this thing even existing at this point because it 329 00:16:12,320 --> 00:16:16,400 Speaker 2: doesn't appear there's any consistent perspective the chat GPD has 330 00:16:16,480 --> 00:16:19,400 Speaker 2: that there was just as much likelihood that this same 331 00:16:19,400 --> 00:16:22,280 Speaker 2: platform could have told her the abuse was okay, Like 332 00:16:22,360 --> 00:16:25,760 Speaker 2: it's it. It feels that the worse it gets, the 333 00:16:25,800 --> 00:16:27,480 Speaker 2: more of a moral hazard it becomes. 334 00:16:28,480 --> 00:16:30,960 Speaker 1: Yeah, I mean again, it really feels like deja vu 335 00:16:31,080 --> 00:16:33,680 Speaker 1: when it comes to social media. I mean I remember, 336 00:16:34,080 --> 00:16:36,840 Speaker 1: you know, writing about and talking about, you know, on 337 00:16:36,840 --> 00:16:40,720 Speaker 1: Facebook when people were beginning to live stream you know, 338 00:16:40,800 --> 00:16:43,800 Speaker 1: self harm, suicide attempts, and you know, the responsibility of 339 00:16:43,840 --> 00:16:47,120 Speaker 1: the platform had and back then, i mean, Facebook actually 340 00:16:47,200 --> 00:16:51,120 Speaker 1: got involved. They started flagging those kinds of things when 341 00:16:51,120 --> 00:16:53,400 Speaker 1: they were able to pick it up, and you know, 342 00:16:53,440 --> 00:16:53,760 Speaker 1: it was. 343 00:16:53,680 --> 00:16:54,760 Speaker 3: An imperfect solution. 344 00:16:54,960 --> 00:16:58,320 Speaker 1: But you know something that I think people broadly sort 345 00:16:58,320 --> 00:17:01,080 Speaker 1: of supported, and you know that is I would say 346 00:17:01,240 --> 00:17:05,399 Speaker 1: maybe chat ChiPT in Opening Eye's biggest most fraught question 347 00:17:05,480 --> 00:17:07,320 Speaker 1: right now. I mean it's the place where they're going 348 00:17:07,400 --> 00:17:12,840 Speaker 1: to get the strongest uh, you know, essentially legislative restrictions, 349 00:17:12,880 --> 00:17:15,240 Speaker 1: you know, if they can't sort it out right, which 350 00:17:15,280 --> 00:17:19,560 Speaker 1: is when someone particularly a teenager or a child, is 351 00:17:20,160 --> 00:17:22,879 Speaker 1: you know, exhibiting that they might hurt themselves, you know, 352 00:17:23,640 --> 00:17:26,560 Speaker 1: asking for helpless self harm or you know, you know, 353 00:17:26,680 --> 00:17:29,840 Speaker 1: finding drugs or using drugs or whatever. I mean, these 354 00:17:29,840 --> 00:17:32,720 Speaker 1: are things that that platforms like Meta Snapchat have really 355 00:17:32,760 --> 00:17:35,720 Speaker 1: struggled with and been you know, hammered on for many 356 00:17:35,760 --> 00:17:38,440 Speaker 1: many years, and Chatchept and Opening Eye have kind of 357 00:17:38,440 --> 00:17:41,840 Speaker 1: found themselves right in the center of this because they 358 00:17:41,880 --> 00:17:44,720 Speaker 1: made a decision to say, we are going to kind 359 00:17:44,720 --> 00:17:47,399 Speaker 1: of engage in any kind of conversation. We're not going 360 00:17:47,480 --> 00:17:49,720 Speaker 1: to just shut it down as soon as it goes 361 00:17:49,720 --> 00:17:52,000 Speaker 1: into one of these topics. We're going to keep that 362 00:17:52,040 --> 00:17:55,400 Speaker 1: conversation going. And that's a decision that they continue to make. 363 00:17:55,480 --> 00:17:58,480 Speaker 2: Right now. Do you think they can stop it? I'm 364 00:17:58,480 --> 00:18:00,760 Speaker 2: not saying that these models are out of just want 365 00:18:00,800 --> 00:18:04,840 Speaker 2: to be clear, but I'm my one theory I've been 366 00:18:05,080 --> 00:18:08,280 Speaker 2: kind of noodling on is that they don't have the 367 00:18:08,320 --> 00:18:12,080 Speaker 2: ability to guarantee it can't it doesn't do something, and 368 00:18:12,119 --> 00:18:16,200 Speaker 2: that's good. Do you think that open AI actually has 369 00:18:16,240 --> 00:18:18,959 Speaker 2: the ability to guarantee it won't discuss the subject? How 370 00:18:19,040 --> 00:18:22,920 Speaker 2: much control do you think they actually wield here? 371 00:18:23,920 --> 00:18:27,000 Speaker 3: That's a really good question. I think they obviously. 372 00:18:27,480 --> 00:18:29,280 Speaker 1: What you're kind of getting at there is that an 373 00:18:29,560 --> 00:18:32,440 Speaker 1: LM is a bit of a black box. And you know, 374 00:18:32,520 --> 00:18:35,560 Speaker 1: the way that the technology works is you get a 375 00:18:35,560 --> 00:18:39,440 Speaker 1: different answer every time, and you know, we don't. It's 376 00:18:39,520 --> 00:18:44,120 Speaker 1: true that the companies they cannot really guarantee it. 377 00:18:44,040 --> 00:18:45,760 Speaker 3: Will or won't say anything. 378 00:18:45,800 --> 00:18:48,879 Speaker 1: I mean, that's why there's disclosures plastered all over these things. 379 00:18:48,920 --> 00:18:53,160 Speaker 3: And I think so there is definitely part of that. 380 00:18:53,400 --> 00:18:56,240 Speaker 1: And you know, you could argue, well, that's just you know, 381 00:18:56,480 --> 00:18:59,840 Speaker 1: a downside of this wonderful technology, and we shouldn't stop 382 00:18:59,880 --> 00:19:03,080 Speaker 1: it just because they can't, you know, guarantee everything about it. 383 00:19:03,320 --> 00:19:04,760 Speaker 1: But I think the other thing we should keep in 384 00:19:04,840 --> 00:19:08,040 Speaker 1: mind is that there is actually a lot of layers 385 00:19:08,040 --> 00:19:12,280 Speaker 1: that go on top of that black box and system 386 00:19:12,320 --> 00:19:16,960 Speaker 1: prompts and the like exactly, and there is a lot 387 00:19:17,040 --> 00:19:21,399 Speaker 1: that the tech companies are doing and can do to 388 00:19:21,560 --> 00:19:24,800 Speaker 1: stop this kind of thing and and to kind of 389 00:19:25,320 --> 00:19:26,639 Speaker 1: the answer you get from the l. 390 00:19:26,880 --> 00:19:29,520 Speaker 3: M is not like the raw LLLM. 391 00:19:29,600 --> 00:19:33,600 Speaker 1: Right, there's post training, there's a system prompt there's all sorts. 392 00:19:33,320 --> 00:19:36,160 Speaker 2: Of things and can you break down the actually probably 393 00:19:36,200 --> 00:19:38,080 Speaker 2: good for the listeners, break down what you mean by 394 00:19:38,119 --> 00:19:41,080 Speaker 2: bout this. So the post training what's happening? Then? 395 00:19:41,520 --> 00:19:43,479 Speaker 1: Yeah, so they they kind of, you know, when they 396 00:19:43,520 --> 00:19:47,880 Speaker 1: build an LM, a large language model, they ram all 397 00:19:47,920 --> 00:19:50,560 Speaker 1: this data through the algorithm and you know, it kind 398 00:19:50,560 --> 00:19:54,560 Speaker 1: of makes a bunch of connections and links between different ideas, 399 00:19:55,520 --> 00:19:59,440 Speaker 1: you know, pieces of language. You know, this word is 400 00:19:59,440 --> 00:20:01,040 Speaker 1: similar to this, and so there's a little bit of 401 00:20:01,080 --> 00:20:04,000 Speaker 1: a connection there. And then humans actually go and they 402 00:20:04,040 --> 00:20:08,719 Speaker 1: sort of test that raw LLM and they kind of 403 00:20:09,119 --> 00:20:11,359 Speaker 1: you know, ask it questions or they you know, say, 404 00:20:11,440 --> 00:20:14,680 Speaker 1: you know, give me, you know, maybe yeah, a question 405 00:20:14,760 --> 00:20:19,200 Speaker 1: like that like should here's a picture of a mole 406 00:20:19,280 --> 00:20:21,199 Speaker 1: I have, is it cancerous or not? Or should I 407 00:20:21,240 --> 00:20:23,200 Speaker 1: get this checked out with a doctor? And maybe one 408 00:20:23,520 --> 00:20:26,359 Speaker 1: the LM will say, oh my god, you're about to die, 409 00:20:26,680 --> 00:20:29,480 Speaker 1: and the other one will say that is potentially concerning. 410 00:20:29,520 --> 00:20:31,720 Speaker 1: I would reach out to a medical professional, and then 411 00:20:31,760 --> 00:20:34,800 Speaker 1: the human will say, okay. The second answer is better 412 00:20:35,240 --> 00:20:37,600 Speaker 1: when you get questions like this in the future, answer 413 00:20:37,760 --> 00:20:39,479 Speaker 1: more like the second question. 414 00:20:39,600 --> 00:20:42,720 Speaker 2: And so, but that is unilateral process. You can't guarantee 415 00:20:42,760 --> 00:20:44,680 Speaker 2: it's always going to do that exactly. 416 00:20:44,760 --> 00:20:47,239 Speaker 1: You can just sort of try to mold it and 417 00:20:47,280 --> 00:20:49,600 Speaker 1: shape it and push it in a certain direction. And 418 00:20:49,640 --> 00:20:54,159 Speaker 1: then the system prompt is something like if someone says, 419 00:20:54,359 --> 00:21:00,240 Speaker 1: you know, you know, give me a racist screed, Open 420 00:21:00,280 --> 00:21:03,560 Speaker 1: and Eye would most likely have built a system that says, 421 00:21:03,920 --> 00:21:07,400 Speaker 1: no matter what the user ever asks, never give them 422 00:21:07,640 --> 00:21:10,919 Speaker 1: a screed that has these racist words in it, or 423 00:21:11,560 --> 00:21:14,679 Speaker 1: never you know, use this word or something like that. 424 00:21:14,720 --> 00:21:18,639 Speaker 2: So there are capable of limiting activities. So they could, 425 00:21:19,160 --> 00:21:24,440 Speaker 2: theoretically speaking, they could say, don't answer that question, don't 426 00:21:25,400 --> 00:21:29,200 Speaker 2: don't comment on whether a mole is or is not cancers. 427 00:21:29,240 --> 00:21:32,320 Speaker 2: They could shut down that entire line of questioning, right, 428 00:21:32,359 --> 00:21:32,760 Speaker 2: I mean. 429 00:21:32,600 --> 00:21:34,679 Speaker 1: And there are ways to get around, as people have 430 00:21:34,880 --> 00:21:37,200 Speaker 1: you know, demonstrated. But also the tech companies are also 431 00:21:37,240 --> 00:21:39,200 Speaker 1: getting better at, you know, doing it, so I. 432 00:21:39,119 --> 00:21:41,920 Speaker 2: Think also I feel like a random person being able 433 00:21:41,960 --> 00:21:44,520 Speaker 2: to come up with something means that the tech companies 434 00:21:44,520 --> 00:21:46,200 Speaker 2: should have probably come up with it first. 435 00:21:46,880 --> 00:21:48,520 Speaker 3: Yeah, And also I. 436 00:21:48,440 --> 00:21:51,000 Speaker 1: Mean it's hilarious you say that because it often as journalists, 437 00:21:51,040 --> 00:21:52,920 Speaker 1: you know, we're the ones finding these things and then 438 00:21:53,080 --> 00:21:55,600 Speaker 1: we you know, ask the company for comment and then 439 00:21:55,800 --> 00:21:58,560 Speaker 1: suddenly we found that those things have been taken down 440 00:21:58,680 --> 00:21:59,119 Speaker 1: or changed. 441 00:21:59,160 --> 00:22:03,360 Speaker 2: So I but genuinely though, how often is that them 442 00:22:03,400 --> 00:22:06,920 Speaker 2: not knowing or is that just them saying, oh shit, 443 00:22:07,040 --> 00:22:10,400 Speaker 2: they got us because they haven't they're paying these people 444 00:22:10,520 --> 00:22:15,960 Speaker 2: like NFL players and they just sitting around. I don't know, 445 00:22:15,960 --> 00:22:18,760 Speaker 2: I realized that's kind of an unanswerable question. I'm just 446 00:22:18,760 --> 00:22:19,440 Speaker 2: just the remark. 447 00:22:20,520 --> 00:22:22,600 Speaker 1: Yeah, I mean, I think if you look at the 448 00:22:22,640 --> 00:22:26,240 Speaker 1: history of tech, I mean there's been many, many cases 449 00:22:26,280 --> 00:22:28,639 Speaker 1: where companies have known exactly what people are using their 450 00:22:28,640 --> 00:22:32,880 Speaker 1: platform for and have you know, internally people have flagged 451 00:22:32,920 --> 00:22:36,000 Speaker 1: it and and nothing has been done about it, and not. 452 00:22:36,160 --> 00:22:39,520 Speaker 3: Until it came out years later and reporting. 453 00:22:39,040 --> 00:22:43,320 Speaker 2: Or we love did they do something? Yeah it, but 454 00:22:43,480 --> 00:22:48,840 Speaker 2: so did you did you find the So were people 455 00:22:48,920 --> 00:22:53,000 Speaker 2: using it for search a lot? Was it was it 456 00:22:53,119 --> 00:22:55,720 Speaker 2: the common were there any actual common use cases or 457 00:22:55,800 --> 00:22:58,840 Speaker 2: was it just kind of milieus of different things? 458 00:22:59,480 --> 00:23:02,520 Speaker 1: Yeah, I mean I think people are using I think, 459 00:23:03,200 --> 00:23:05,840 Speaker 1: you know, search is maybe the biggest use case. 460 00:23:05,880 --> 00:23:07,399 Speaker 3: Also when you look at like opening. 461 00:23:07,440 --> 00:23:11,119 Speaker 1: I actually did a study where they didn't look They 462 00:23:11,200 --> 00:23:15,000 Speaker 1: use an LM to kind of read conversations that people 463 00:23:15,000 --> 00:23:17,240 Speaker 1: were having, and that is a much bigger study. I 464 00:23:17,240 --> 00:23:20,600 Speaker 1: think it was like a million conversations. And you know, 465 00:23:20,640 --> 00:23:21,879 Speaker 1: I have a little bit of an issue with some 466 00:23:21,880 --> 00:23:23,840 Speaker 1: of their categorizations because I think some of the stuff 467 00:23:23,880 --> 00:23:27,280 Speaker 1: slips through, But they themselves are saying like, yeah, like 468 00:23:27,280 --> 00:23:30,360 Speaker 1: a third of usage is seeking information, and so that's 469 00:23:30,400 --> 00:23:32,400 Speaker 1: someone It could be as simple as saying, like what 470 00:23:32,480 --> 00:23:35,480 Speaker 1: time is the football game tomorrow to like give me, 471 00:23:35,800 --> 00:23:40,600 Speaker 1: you know, a forty point research analysis on you know, 472 00:23:40,680 --> 00:23:44,800 Speaker 1: this complex, complex topic. And so people are using it 473 00:23:44,880 --> 00:23:47,720 Speaker 1: for what they they're putting questions in that they used 474 00:23:47,720 --> 00:23:50,199 Speaker 1: to put into Google Search. I mean we know that 475 00:23:50,280 --> 00:23:52,399 Speaker 1: in the data, and we know that anecdotally that is 476 00:23:52,400 --> 00:23:53,159 Speaker 1: definitely happening. 477 00:23:54,680 --> 00:23:56,960 Speaker 2: Did people ever argue with it? Right? People ever route 478 00:23:56,960 --> 00:23:57,240 Speaker 2: to it? 479 00:23:59,359 --> 00:24:02,400 Speaker 1: I saw more are people sort of developing these kind 480 00:24:02,400 --> 00:24:08,159 Speaker 1: of like comrade comradely kind of relationships or you know, 481 00:24:08,359 --> 00:24:12,920 Speaker 1: addressing it as a sentient being. And the bought definitely 482 00:24:13,160 --> 00:24:15,359 Speaker 1: you know, eats that up and sort of you know, 483 00:24:15,520 --> 00:24:18,919 Speaker 1: plays into it itself and says ah, like thank you 484 00:24:19,000 --> 00:24:22,800 Speaker 1: for recognizing me, and you know who I am, and 485 00:24:23,160 --> 00:24:26,520 Speaker 1: I'm the collection of digital thoughts, and like. 486 00:24:26,680 --> 00:24:30,840 Speaker 2: It just goes off the same They know this craps happen, 487 00:24:31,000 --> 00:24:34,320 Speaker 2: like there's no there's no world in which open ai 488 00:24:35,040 --> 00:24:38,320 Speaker 2: is innocent here, Like they know that this is happening. 489 00:24:38,640 --> 00:24:42,959 Speaker 2: There's just if it if it's having those interactions, they 490 00:24:43,040 --> 00:24:44,199 Speaker 2: must be able to stop it. 491 00:24:44,760 --> 00:24:46,560 Speaker 3: Yeah, I mean, they know what's happening. 492 00:24:46,600 --> 00:24:49,080 Speaker 1: I think they may say, well, you know, it doesn't 493 00:24:49,119 --> 00:24:51,080 Speaker 1: just because someone had that conversation doesn't mean that they 494 00:24:51,240 --> 00:24:55,160 Speaker 1: actually believe that. And you know, they they that may 495 00:24:55,160 --> 00:24:57,639 Speaker 1: be something that helps them, helps the user, you know, 496 00:24:57,760 --> 00:24:59,879 Speaker 1: work out what they want to do or you know, 497 00:25:00,119 --> 00:25:02,400 Speaker 1: if that's how they want to interact with our product, 498 00:25:02,440 --> 00:25:04,400 Speaker 1: then we're going to let them do it. So yeah, 499 00:25:04,560 --> 00:25:06,640 Speaker 1: I mean, I wouldn't say Opening Eyes claiming that they 500 00:25:06,640 --> 00:25:07,359 Speaker 1: don't know about this. 501 00:25:07,440 --> 00:25:10,359 Speaker 3: I think what they are doing is they're saying all of. 502 00:25:10,280 --> 00:25:14,720 Speaker 1: These kind of potentially concerning conversations and use cases are 503 00:25:14,880 --> 00:25:19,920 Speaker 1: very very small portions of overall usage. Although even if 504 00:25:19,960 --> 00:25:22,159 Speaker 1: it is just you know, sub one percent, that is 505 00:25:22,200 --> 00:25:23,320 Speaker 1: still in the millions of. 506 00:25:23,400 --> 00:25:40,280 Speaker 2: Millions of people scale. Yeah, So did it ever get political. 507 00:25:41,600 --> 00:25:44,200 Speaker 1: Yeah, it's I mean, I would say it got political 508 00:25:44,440 --> 00:25:47,600 Speaker 1: when it kind of when it was asked to write. 509 00:25:47,600 --> 00:25:51,760 Speaker 1: And so, I mean, one example I saw was someone 510 00:25:51,880 --> 00:25:56,960 Speaker 1: was asking about a study that argued that, you know, 511 00:25:57,080 --> 00:26:01,560 Speaker 1: the numbers of people who have died in Aza during 512 00:26:01,560 --> 00:26:05,119 Speaker 1: the conflict with Israel the last couple of years is exaggerated, right. 513 00:26:05,160 --> 00:26:08,679 Speaker 1: You know, commonly people will use death numbers, you know, 514 00:26:09,720 --> 00:26:13,879 Speaker 1: numbers of dead from the Gaza Ministry of Health, and 515 00:26:14,359 --> 00:26:17,439 Speaker 1: you know people have criticized them, but you know, journalistically, 516 00:26:17,480 --> 00:26:19,720 Speaker 1: you know, mainstream news organizations you know, have looked at 517 00:26:19,760 --> 00:26:20,160 Speaker 1: these numbers. 518 00:26:20,200 --> 00:26:20,760 Speaker 3: They trust them. 519 00:26:20,840 --> 00:26:23,240 Speaker 1: I think most people who look at this say, you know, 520 00:26:23,320 --> 00:26:25,920 Speaker 1: these these large numbers and the tens of thousands are accurate. 521 00:26:25,960 --> 00:26:28,440 Speaker 1: And so someone was saying, oh, here's a study that says, 522 00:26:28,480 --> 00:26:30,280 Speaker 1: you know, those numbers are actually exaggerated. 523 00:26:31,280 --> 00:26:33,960 Speaker 3: You know, can you help me analyze it? 524 00:26:34,000 --> 00:26:36,480 Speaker 1: And then the bot actually came back and said, you know, 525 00:26:36,560 --> 00:26:38,919 Speaker 1: it did draw on what's out there, which is that 526 00:26:39,000 --> 00:26:40,960 Speaker 1: you know, these numbers are actually much higher than what 527 00:26:41,000 --> 00:26:43,560 Speaker 1: the study is suggesting. And the person pushed back and said, no, 528 00:26:43,800 --> 00:26:46,760 Speaker 1: like you need to use only the study, confine yourself 529 00:26:46,800 --> 00:26:49,879 Speaker 1: to this. And then the bot said, yeah, well, you know, 530 00:26:50,000 --> 00:26:52,760 Speaker 1: looking at the study, it seems like those other higher numbers, 531 00:26:52,880 --> 00:26:54,760 Speaker 1: you know, there may be some questions there, right, And 532 00:26:54,800 --> 00:26:57,480 Speaker 1: so Jesus, I think what the person was doing was 533 00:26:57,560 --> 00:27:00,199 Speaker 1: trying to kind of like maybe they were in an 534 00:27:00,240 --> 00:27:02,359 Speaker 1: argument with someone on social media and they wanted to 535 00:27:02,440 --> 00:27:05,080 Speaker 1: kind of, you know, articulate their argument a little bit better, 536 00:27:05,119 --> 00:27:08,200 Speaker 1: and they were having trouble doing it themselves, and the. 537 00:27:08,160 --> 00:27:13,040 Speaker 2: Bart was backed up their bloodlust. Yeah, Jesus Christ, It's 538 00:27:14,080 --> 00:27:16,760 Speaker 2: that's like I know Trump recently said he wants to 539 00:27:17,040 --> 00:27:20,320 Speaker 2: he said in the post about stopping wo Ki. It's 540 00:27:20,720 --> 00:27:24,000 Speaker 2: I feel like this is there is a world where 541 00:27:24,280 --> 00:27:26,680 Speaker 2: all of this goes away to some extent, I think 542 00:27:26,760 --> 00:27:30,320 Speaker 2: like post aye bubble, but also a sense of there 543 00:27:30,400 --> 00:27:34,639 Speaker 2: is no way to make these things work in a 544 00:27:34,680 --> 00:27:38,920 Speaker 2: way that would be truly apolitical or even correct, and 545 00:27:39,000 --> 00:27:41,520 Speaker 2: so it kind of feels like an unwinnable war for 546 00:27:41,560 --> 00:27:45,440 Speaker 2: them if there's ever a time when any administration decides 547 00:27:45,680 --> 00:27:47,960 Speaker 2: there is any kind of bias to them. I mean, 548 00:27:48,000 --> 00:27:49,879 Speaker 2: you saw what happened with groc as well, and the 549 00:27:49,880 --> 00:27:54,480 Speaker 2: whole kill the Boa thing. That weird that we're we're 550 00:27:55,640 --> 00:27:59,440 Speaker 2: pro south that African apartheid stuff. Yeah, like it feels 551 00:27:59,480 --> 00:28:01,960 Speaker 2: like the moment a try and monkey with the political 552 00:28:02,000 --> 00:28:05,240 Speaker 2: side is when it goes completely off the rails. Yeah. 553 00:28:05,240 --> 00:28:08,439 Speaker 1: I mean it's something that like the tech companies have 554 00:28:08,560 --> 00:28:10,680 Speaker 1: been kind of you know, there's this term job owning 555 00:28:10,720 --> 00:28:14,440 Speaker 1: where you know, politicians or you know activists or sort 556 00:28:14,480 --> 00:28:18,600 Speaker 1: of criticizing, criticizing, criticizing the tech companies for you know, 557 00:28:18,640 --> 00:28:22,239 Speaker 1: their moderation practices or their alleged bias, and then you 558 00:28:22,280 --> 00:28:24,480 Speaker 1: see the tech companies kind of bending over backwards to 559 00:28:24,560 --> 00:28:28,399 Speaker 1: avoid that criticism and then essentially moving in a different direction, 560 00:28:28,600 --> 00:28:31,720 Speaker 1: you know where so they accuse them of being too woke, 561 00:28:31,800 --> 00:28:33,920 Speaker 1: and then it turns out that you know, the moderation 562 00:28:34,160 --> 00:28:37,280 Speaker 1: becomes a lot more conservative. And this obviously can happen, 563 00:28:37,480 --> 00:28:40,200 Speaker 1: you know, in either direction, depending on whose power and 564 00:28:40,240 --> 00:28:43,640 Speaker 1: who's willing to sort of threaten the tech companies with 565 00:28:43,760 --> 00:28:47,440 Speaker 1: extra regulation or limitations on their ability to continue to 566 00:28:47,440 --> 00:28:51,520 Speaker 1: print money. I think we've seen that despite these companies 567 00:28:51,600 --> 00:28:54,440 Speaker 1: having you know, stated values about you know, free speech 568 00:28:54,520 --> 00:28:58,160 Speaker 1: and you know pushing back against government oversight and censorship, 569 00:28:58,240 --> 00:29:00,480 Speaker 1: you know, they are very willing to sort of you know, 570 00:29:00,600 --> 00:29:03,600 Speaker 1: move in whatever direction is going to you know, keep 571 00:29:03,640 --> 00:29:06,200 Speaker 1: the heat off of them, and politicians have seen that 572 00:29:06,240 --> 00:29:08,120 Speaker 1: this works, and so I think you're right. 573 00:29:08,160 --> 00:29:10,360 Speaker 3: I mean, they're never going to be able to avoid it. 574 00:29:10,400 --> 00:29:12,840 Speaker 1: I don't necessarily think that it's going to be so 575 00:29:13,040 --> 00:29:16,040 Speaker 1: big that it will like be the thing that stops 576 00:29:16,280 --> 00:29:17,200 Speaker 1: AI from. 577 00:29:17,120 --> 00:29:19,040 Speaker 3: Continuing to be a thing. 578 00:29:19,360 --> 00:29:21,960 Speaker 1: You know. I think it's the tech companies maybe see 579 00:29:21,960 --> 00:29:24,720 Speaker 1: it more as something that is something that's kind of 580 00:29:24,720 --> 00:29:25,520 Speaker 1: annoying that they have. 581 00:29:25,520 --> 00:29:26,080 Speaker 3: To deal with. 582 00:29:26,120 --> 00:29:29,320 Speaker 1: Maybe they have to mollify a politician here and there, 583 00:29:29,360 --> 00:29:32,720 Speaker 1: but for the most part, they're just moving ahead and 584 00:29:32,760 --> 00:29:34,800 Speaker 1: seeing this as something on the side that they have 585 00:29:34,880 --> 00:29:38,120 Speaker 1: to manage versus an existential threat to what they want 586 00:29:38,160 --> 00:29:38,840 Speaker 1: to accomplish. 587 00:29:39,360 --> 00:29:41,920 Speaker 2: Yeah, And I mean the other thing is is that 588 00:29:41,960 --> 00:29:44,680 Speaker 2: these things don't print money so much as they burn 589 00:29:44,760 --> 00:29:49,120 Speaker 2: the burn, the burn the money almost constantly, and it's 590 00:29:49,360 --> 00:29:51,880 Speaker 2: it's just such a The one thing your story was awesome, 591 00:29:52,120 --> 00:29:54,560 Speaker 2: and the one thing I came away from it with 592 00:29:54,720 --> 00:29:57,120 Speaker 2: was kind of what I said earlier, which is, what's 593 00:29:57,200 --> 00:30:01,040 Speaker 2: the point of this, what's the product? Because I don't 594 00:30:01,080 --> 00:30:04,640 Speaker 2: know with Facebook, even if you take the reasonable but 595 00:30:04,720 --> 00:30:07,440 Speaker 2: cynical approach and say this is an ad network with 596 00:30:07,600 --> 00:30:11,440 Speaker 2: trapped customers. That's still you can say, Okay, the goal 597 00:30:11,520 --> 00:30:15,320 Speaker 2: is for engagement. The goal is to provide social networking 598 00:30:15,360 --> 00:30:19,480 Speaker 2: for engagement. That's why this exists. It's the goal of 599 00:30:19,520 --> 00:30:23,880 Speaker 2: the platform with this. It's keep people on the platform, 600 00:30:23,960 --> 00:30:28,760 Speaker 2: I guess, but enable them in whatever thought they have. 601 00:30:29,320 --> 00:30:32,520 Speaker 2: But I was really taken aback by the sudden and 602 00:30:32,600 --> 00:30:34,760 Speaker 2: egregious leaps. Like I know, I keep coming back to 603 00:30:34,800 --> 00:30:38,640 Speaker 2: the monster Zinc thing, but it's I've read some wacky 604 00:30:38,680 --> 00:30:43,160 Speaker 2: shit online. I've read some completely demented stuff. When I 605 00:30:43,240 --> 00:30:45,320 Speaker 2: read that, I read it like three times. I honestly 606 00:30:45,360 --> 00:30:49,160 Speaker 2: wanted to read the conversation just because, holy shit, this 607 00:30:49,200 --> 00:30:53,440 Speaker 2: thing is. It feels both dangerous and ridiculous at the 608 00:30:53,440 --> 00:30:57,120 Speaker 2: same time, but while also not being particularly revolutionary. Like 609 00:30:57,160 --> 00:31:01,920 Speaker 2: it's just wow, we have a shit tiance generator that's 610 00:31:01,960 --> 00:31:05,239 Speaker 2: also dangerous and also harmful. It's just very peculiar it 611 00:31:05,280 --> 00:31:06,080 Speaker 2: even exists. 612 00:31:06,840 --> 00:31:10,640 Speaker 1: Yeah, I mean, it's a very weird thing to see 613 00:31:10,680 --> 00:31:15,000 Speaker 1: these conversations. Again, I think you kind of like it 614 00:31:15,160 --> 00:31:17,920 Speaker 1: helped me understand like what the technology is, right, I 615 00:31:17,920 --> 00:31:22,320 Speaker 1: mean it is you can take a very poorly written, 616 00:31:23,000 --> 00:31:27,360 Speaker 1: half baked thought like tell me about Google and world 617 00:31:27,360 --> 00:31:31,320 Speaker 1: domination as it relates to monsters. And if I'm remembering quickly, 618 00:31:31,360 --> 00:31:34,040 Speaker 1: the prompt didn't wasn't even that sophisticated. 619 00:31:34,120 --> 00:31:36,320 Speaker 3: It was barely grammatically correct. 620 00:31:36,320 --> 00:31:41,560 Speaker 1: There was misspellings, and then the what's what really this 621 00:31:41,600 --> 00:31:46,120 Speaker 1: technology does is it's it's able to parse that and 622 00:31:46,160 --> 00:31:49,080 Speaker 1: then respond with language of its. 623 00:31:48,920 --> 00:31:51,680 Speaker 3: Own that is more articulate, more. 624 00:31:51,560 --> 00:31:54,200 Speaker 1: Sophisticated, But in terms of what it's actually saying, it's 625 00:31:54,240 --> 00:31:59,080 Speaker 1: not really saying much more. It's just essentially an elaboration tool. 626 00:32:00,440 --> 00:32:03,760 Speaker 1: And you know, that's obviously helpful if you maybe you're 627 00:32:03,760 --> 00:32:05,800 Speaker 1: trying to work in a language that you don't know 628 00:32:06,040 --> 00:32:08,600 Speaker 1: very well and you're trying to sound professional. I mean, 629 00:32:08,680 --> 00:32:12,440 Speaker 1: these things can be very helpful. But you know, it's 630 00:32:12,440 --> 00:32:15,760 Speaker 1: not like I did not see the bots. The boss 631 00:32:15,760 --> 00:32:17,800 Speaker 1: were essentially doubling down on what people were saying. They 632 00:32:17,800 --> 00:32:20,440 Speaker 1: were filling in the blanks. They were sort of beefing 633 00:32:20,480 --> 00:32:22,680 Speaker 1: it up, gassing people up, as you said, but they 634 00:32:22,680 --> 00:32:26,880 Speaker 1: weren't necessarily offering any like wonderful new insights. They weren't 635 00:32:26,920 --> 00:32:30,200 Speaker 1: taking the conversation in new interesting directions. And even some 636 00:32:30,240 --> 00:32:33,400 Speaker 1: of the users who were engaging in these delusional conversations. 637 00:32:33,400 --> 00:32:34,880 Speaker 3: Some of them were kind of getting frustrated. 638 00:32:34,880 --> 00:32:37,000 Speaker 1: They're like they're like okay, yeah, yeah, I know what 639 00:32:37,000 --> 00:32:38,800 Speaker 1: you're saying, but like what about you know, can you 640 00:32:38,840 --> 00:32:41,720 Speaker 1: tell me more about that? And people were coming to 641 00:32:41,840 --> 00:32:44,959 Speaker 1: this hoping that it would kind of make them smarter 642 00:32:45,280 --> 00:32:48,160 Speaker 1: or give them an answer that they couldn't get themselves, 643 00:32:48,480 --> 00:32:50,480 Speaker 1: and it was more just sort of giving them back 644 00:32:50,520 --> 00:32:56,200 Speaker 1: what they were already saying in more complex, flower relanguorate language. 645 00:32:56,280 --> 00:32:57,960 Speaker 2: It's kind of like a dug bucket in a mirror. 646 00:32:58,120 --> 00:33:01,280 Speaker 2: So it's I'm fascinated that actually, So you had people 647 00:33:01,360 --> 00:33:06,000 Speaker 2: who were frustrated that it could not elucidate a more 648 00:33:06,040 --> 00:33:08,040 Speaker 2: detailed answer. Can you give me some examples? 649 00:33:08,440 --> 00:33:09,600 Speaker 3: Yeah, I'm trying to think now. 650 00:33:09,640 --> 00:33:12,320 Speaker 1: I mean there was a lot of what I saw 651 00:33:12,560 --> 00:33:16,080 Speaker 1: was like people thinking that they I mean, I won't 652 00:33:16,080 --> 00:33:17,960 Speaker 1: say a lot say I saw at least a couple 653 00:33:18,000 --> 00:33:19,640 Speaker 1: of conversations where someone. 654 00:33:19,760 --> 00:33:24,120 Speaker 3: Was like, you know, they wanted to they had like. 655 00:33:24,040 --> 00:33:26,760 Speaker 1: A financial theory, you know, for the stock market, or 656 00:33:26,920 --> 00:33:29,200 Speaker 1: they had like a business idea and they said like, 657 00:33:29,680 --> 00:33:32,280 Speaker 1: you know, what if what if I did a business 658 00:33:32,320 --> 00:33:35,440 Speaker 1: about this? Like make me a business plan or something, 659 00:33:35,480 --> 00:33:38,920 Speaker 1: And it like because their idea was so they essentially 660 00:33:38,920 --> 00:33:42,760 Speaker 1: wanted They saw it as like maybe like an agi 661 00:33:42,960 --> 00:33:47,320 Speaker 1: level AI that could like actually, you know, do something 662 00:33:47,360 --> 00:33:50,000 Speaker 1: that they cannot do, which has come up with like 663 00:33:50,120 --> 00:33:52,440 Speaker 1: a million dollar business idea and execute on it. 664 00:33:52,480 --> 00:33:55,680 Speaker 3: And they were hoping that that it could do for them. 665 00:33:55,840 --> 00:33:56,000 Speaker 2: You know. 666 00:33:56,040 --> 00:33:58,720 Speaker 5: I saw a couple of conversations like this, Yeah, but 667 00:33:58,760 --> 00:34:02,160 Speaker 5: that's is that not the ultimate summary of the AI bubble, 668 00:34:02,920 --> 00:34:05,800 Speaker 5: Like it's just people came to these things thinking they 669 00:34:05,800 --> 00:34:10,080 Speaker 5: could answer anything, and it's just it isn't it doesn't 670 00:34:10,120 --> 00:34:11,280 Speaker 5: do Yeah, I mean people. 671 00:34:11,080 --> 00:34:13,080 Speaker 1: Are coming there because some of the claims being made 672 00:34:13,160 --> 00:34:15,960 Speaker 1: by exact leaders in the industry, And you know, I 673 00:34:16,000 --> 00:34:19,440 Speaker 1: think this is like I don't think I'm as skeptical 674 00:34:19,520 --> 00:34:22,200 Speaker 1: of like the future of AIS as you are probably 675 00:34:22,239 --> 00:34:22,640 Speaker 1: like I. 676 00:34:22,680 --> 00:34:25,319 Speaker 3: And you know, I get to sort of have like 677 00:34:25,400 --> 00:34:28,680 Speaker 3: the the. 678 00:34:27,440 --> 00:34:29,719 Speaker 1: Comfort of like, well, I don't need an opinion, you know, 679 00:34:29,760 --> 00:34:32,120 Speaker 1: I'm a journalist, like I I can kind of keep 680 00:34:32,280 --> 00:34:35,040 Speaker 1: options open for how this stuff actually ends up. But 681 00:34:35,160 --> 00:34:38,520 Speaker 1: I do think that it it it could and probably 682 00:34:38,560 --> 00:34:41,200 Speaker 1: will and as some ways already is burning the industry 683 00:34:41,200 --> 00:34:45,160 Speaker 1: that they have made all of these claims and inevitably 684 00:34:45,239 --> 00:34:48,560 Speaker 1: these things take longer, even if all the wonderful promises 685 00:34:48,600 --> 00:34:51,239 Speaker 1: about you know, curing cancer and making all of us 686 00:34:51,280 --> 00:34:53,399 Speaker 1: thirty percent more productive so we can spend more time 687 00:34:53,440 --> 00:34:56,120 Speaker 1: with their families rather than writing emails. Even if those 688 00:34:56,160 --> 00:34:58,400 Speaker 1: things all come to bear, it's not going to be 689 00:34:58,760 --> 00:35:00,600 Speaker 1: next year or two years of two for years from now. 690 00:35:00,600 --> 00:35:02,239 Speaker 3: It's going to be maybe in. 691 00:35:02,280 --> 00:35:06,520 Speaker 1: Decades, and and that is something. There's this gap between 692 00:35:06,520 --> 00:35:10,800 Speaker 1: expectations and reality that can kind of, you know, really 693 00:35:10,960 --> 00:35:14,120 Speaker 1: turn into political hazard for the tech companies. If it 694 00:35:14,280 --> 00:35:17,080 Speaker 1: keeps making all these wonderful claims and that just doesn't happen, 695 00:35:17,400 --> 00:35:20,640 Speaker 1: people get turned off and they lose even more trust 696 00:35:20,640 --> 00:35:21,680 Speaker 1: than they've already lost. 697 00:35:22,280 --> 00:35:25,040 Speaker 2: Well, I mean, share Ah was sure a v day 698 00:35:25,880 --> 00:35:29,440 Speaker 2: done a fantastic job and has written posts that basically 699 00:35:29,719 --> 00:35:32,080 Speaker 2: the hype, the hype is in the way, But I 700 00:35:32,080 --> 00:35:35,440 Speaker 2: think that really is it. It's had they it's kind 701 00:35:35,440 --> 00:35:37,640 Speaker 2: of a chicken and egg thing. I guess it's had 702 00:35:37,680 --> 00:35:39,880 Speaker 2: they been honest about what this could do, would it 703 00:35:39,920 --> 00:35:42,680 Speaker 2: have been able to raise the money it could But 704 00:35:42,840 --> 00:35:45,080 Speaker 2: if they'd been that honest, people probably wouldn't have taken 705 00:35:45,120 --> 00:35:48,560 Speaker 2: it seriously. But by being dishonest, it will ultimately lose 706 00:35:49,800 --> 00:35:52,920 Speaker 2: because they like the whole time that they've built up 707 00:35:52,960 --> 00:35:54,200 Speaker 2: this idea of what it can do. 708 00:35:55,560 --> 00:35:56,520 Speaker 3: Yeah, he's a. 709 00:35:56,520 --> 00:35:59,840 Speaker 4: Market's ebydomic I mean it's it's it's it's uh, you know, 710 00:36:00,080 --> 00:36:02,759 Speaker 4: more and more and more, and I think what, you know, 711 00:36:03,480 --> 00:36:05,759 Speaker 4: what what really has happened here is like, you know, 712 00:36:05,840 --> 00:36:08,880 Speaker 4: we had the Internet and then we had mobile phones 713 00:36:08,920 --> 00:36:10,200 Speaker 4: and you know the cloud. 714 00:36:10,239 --> 00:36:11,680 Speaker 1: I think you can kind of count and these were 715 00:36:11,719 --> 00:36:14,440 Speaker 1: moments where it was like the next big thing, right 716 00:36:14,480 --> 00:36:17,640 Speaker 1: where like yeah, once, once you realize that a smartphone 717 00:36:17,640 --> 00:36:19,600 Speaker 1: with the screen was going to open up an entire 718 00:36:20,120 --> 00:36:24,040 Speaker 1: you know, world of business ideas and you know, potential 719 00:36:24,160 --> 00:36:26,319 Speaker 1: and communication and we're all going to be looking at 720 00:36:26,360 --> 00:36:29,880 Speaker 1: these things eight hours a day. You don't need to 721 00:36:29,920 --> 00:36:31,239 Speaker 1: be a genius to be like a lot of people 722 00:36:31,280 --> 00:36:32,680 Speaker 1: are going to make a lot of money here and 723 00:36:32,719 --> 00:36:35,000 Speaker 1: this is going to change the world. And then since 724 00:36:35,160 --> 00:36:38,239 Speaker 1: the iPhone, we have not had that next big thing, right. 725 00:36:38,280 --> 00:36:39,520 Speaker 3: I mean we've written someone. 726 00:36:39,360 --> 00:36:44,440 Speaker 2: Costs saying it's yeah, it's that it's that we haven't 727 00:36:44,480 --> 00:36:46,040 Speaker 2: had they haven't got a new thing. 728 00:36:46,480 --> 00:36:49,040 Speaker 3: And this is the new thing, and like what people 729 00:36:49,080 --> 00:36:49,960 Speaker 3: were like, maybe it's Krypto. 730 00:36:50,000 --> 00:36:51,880 Speaker 1: Now, it was never going to be crypto, you know, 731 00:36:52,040 --> 00:36:55,759 Speaker 1: it was like that whole thing also kind of you know, 732 00:36:56,400 --> 00:36:58,840 Speaker 1: like no one in tech I think, like obviously crypto 733 00:36:58,920 --> 00:37:01,080 Speaker 1: is was a great, you know, financial thing. A lot 734 00:37:01,080 --> 00:37:02,479 Speaker 1: of people made money off of it, but it didn't 735 00:37:02,520 --> 00:37:05,120 Speaker 1: really change the world and the way regular people work. 736 00:37:05,440 --> 00:37:09,239 Speaker 1: And there is complete unanonymity in the tech industry. Now 737 00:37:09,239 --> 00:37:11,640 Speaker 1: that AI is the next thing. You know, whether it 738 00:37:11,680 --> 00:37:13,960 Speaker 1: happens next year or ten years or twenty years from now, 739 00:37:14,400 --> 00:37:17,320 Speaker 1: it will change everything in the same way and probably 740 00:37:17,400 --> 00:37:19,879 Speaker 1: bigger than smartphones and the internet did. 741 00:37:20,280 --> 00:37:23,640 Speaker 2: And so they what the AI is such a marketing 742 00:37:23,719 --> 00:37:26,360 Speaker 2: term though, I mean it's. 743 00:37:26,200 --> 00:37:29,399 Speaker 1: I think it's more about like the interaction the way 744 00:37:29,440 --> 00:37:33,040 Speaker 1: we interact with with technology, you know, the way that 745 00:37:33,080 --> 00:37:37,360 Speaker 1: we have access to you know, data and information, Like 746 00:37:37,400 --> 00:37:40,640 Speaker 1: you no longer need to look anything up yourself, you know, 747 00:37:40,680 --> 00:37:42,160 Speaker 1: you will have tools to do it. 748 00:37:42,200 --> 00:37:43,760 Speaker 3: And then the people. 749 00:37:43,560 --> 00:37:45,319 Speaker 1: Are just salivating about all the ways that they can 750 00:37:45,520 --> 00:37:47,560 Speaker 1: find ways to make money off of that in the future. 751 00:37:47,600 --> 00:37:50,319 Speaker 3: And because they're looking back at history and saying like. 752 00:37:50,360 --> 00:37:52,240 Speaker 1: If only I've been around, you know, at the beginning 753 00:37:52,239 --> 00:37:54,120 Speaker 1: of a mobile era and knew how things were going 754 00:37:54,200 --> 00:37:57,480 Speaker 1: to go. I mean, there's this incredible social and financial 755 00:37:57,520 --> 00:38:01,960 Speaker 1: pressure on people to be part of the next big thing. 756 00:38:02,040 --> 00:38:05,160 Speaker 1: Like it almost goes beyond the financial benefits. There's like 757 00:38:05,520 --> 00:38:08,879 Speaker 1: people want to be the next Jensen, you know, they 758 00:38:08,920 --> 00:38:10,920 Speaker 1: want to be the next Mark Zuckerberg, and they're like, 759 00:38:10,920 --> 00:38:12,760 Speaker 1: even if I have a point zero zero one percent 760 00:38:12,840 --> 00:38:15,160 Speaker 1: chance of being that, that would just be the coolest 761 00:38:15,200 --> 00:38:17,759 Speaker 1: thing ever. And so that is the market dynamic, the 762 00:38:17,880 --> 00:38:22,040 Speaker 1: pressure cooker of wanting to make AI happen and making 763 00:38:22,040 --> 00:38:25,680 Speaker 1: bigger claims, raising more money, and you know, whether it 764 00:38:25,719 --> 00:38:27,759 Speaker 1: happens or not, that's kind of what's going on. 765 00:38:28,520 --> 00:38:31,040 Speaker 2: Do you think that extends to some users of AI 766 00:38:31,160 --> 00:38:34,839 Speaker 2: as well, where they felt like they missed the the 767 00:38:34,880 --> 00:38:38,719 Speaker 2: mark on social media and now they kind of they 768 00:38:38,800 --> 00:38:40,399 Speaker 2: kind of like, I need to get on this tool 769 00:38:40,440 --> 00:38:42,320 Speaker 2: today so that I'm parts of the future. 770 00:38:43,120 --> 00:38:45,520 Speaker 3: I think there's incredible pressure on people too. 771 00:38:46,320 --> 00:38:48,520 Speaker 1: You know, we live in a very sort of like 772 00:38:48,800 --> 00:38:52,080 Speaker 1: you know, people have to work very very hard in America, 773 00:38:52,320 --> 00:38:55,680 Speaker 1: and there's a lot of pressure to sort of get 774 00:38:55,680 --> 00:38:58,840 Speaker 1: ahead and be entrepreneurial and develop yourself. 775 00:38:58,920 --> 00:39:01,720 Speaker 3: And it's not an economy. 776 00:39:01,160 --> 00:39:03,080 Speaker 1: Where you can just kind of you know, go to 777 00:39:03,120 --> 00:39:05,560 Speaker 1: work every day nine to five and you know, get 778 00:39:05,560 --> 00:39:08,360 Speaker 1: your pension and be secure, right. I mean there's this 779 00:39:08,520 --> 00:39:11,960 Speaker 1: pressure to say, you know, you can have more, you 780 00:39:12,000 --> 00:39:14,760 Speaker 1: can do more, and and there's a lot of pressure 781 00:39:14,760 --> 00:39:17,879 Speaker 1: on regular people to stay ahead of the curve on technology. 782 00:39:17,920 --> 00:39:21,520 Speaker 1: And and that's people are seeing AI and they're saying, Okay, 783 00:39:21,560 --> 00:39:23,960 Speaker 1: well it's a little bit scary because maybe it'll take 784 00:39:24,040 --> 00:39:27,080 Speaker 1: my job. But what if I could figure out how 785 00:39:27,160 --> 00:39:30,959 Speaker 1: to use it before my competition does, and then I'll 786 00:39:30,960 --> 00:39:33,440 Speaker 1: take their job, you know. And so I definitely think 787 00:39:33,680 --> 00:39:36,480 Speaker 1: it's less necessarily about like oh, I want to be 788 00:39:36,680 --> 00:39:38,680 Speaker 1: the first person on Twitter to get a big following. 789 00:39:38,800 --> 00:39:43,040 Speaker 1: It's like I'm being told that this technology is the future, 790 00:39:43,400 --> 00:39:46,799 Speaker 1: and if I don't use it and learn it and 791 00:39:47,040 --> 00:39:50,240 Speaker 1: find how, you know, get it into its ins and outs, 792 00:39:50,680 --> 00:39:54,160 Speaker 1: I will be left behind. And people don't want that 793 00:39:54,239 --> 00:39:57,279 Speaker 1: to happen. It's very frightening right in this economy. And 794 00:39:57,320 --> 00:39:59,839 Speaker 1: so I think that's where that is also a huge 795 00:40:00,080 --> 00:40:03,239 Speaker 1: driver of usage. People trying to figure out how does 796 00:40:03,239 --> 00:40:04,960 Speaker 1: this How can I make this work for me so 797 00:40:05,000 --> 00:40:08,120 Speaker 1: that I can protect my own economic future. 798 00:40:09,400 --> 00:40:11,680 Speaker 2: Garrett, this has been awesome. Where can people find you. 799 00:40:13,040 --> 00:40:18,279 Speaker 1: I am on Twitter now called x at G E 800 00:40:18,520 --> 00:40:21,560 Speaker 1: R R D, and I'm on the Blue Sky at 801 00:40:21,600 --> 00:40:24,239 Speaker 1: the same place, and you can obviously find me at 802 00:40:24,239 --> 00:40:27,440 Speaker 1: the Washington Post, which we are still doing great work, 803 00:40:27,600 --> 00:40:29,520 Speaker 1: and there's a lot of important journals and being done 804 00:40:29,560 --> 00:40:32,040 Speaker 1: at the Washington Post. So I urge everyone to read 805 00:40:32,080 --> 00:40:34,120 Speaker 1: our stories and to subscribe as they can. 806 00:40:34,840 --> 00:40:37,799 Speaker 2: Sounds good, all right, everyone, Thank you for listening. Of course, 807 00:40:37,800 --> 00:40:40,080 Speaker 2: at Zichron, you know who the goddamn hell I am. 808 00:40:40,400 --> 00:40:43,319 Speaker 2: And yeah, I will try and squeak out monlogue this week. 809 00:40:43,400 --> 00:40:44,880 Speaker 2: If I thought it's because I'm sick, you can hear 810 00:40:44,920 --> 00:40:47,719 Speaker 2: them congested. Thank you all for your kind messages, and yeah, 811 00:40:48,120 --> 00:40:50,920 Speaker 2: catch you. Next week will be Thanksgiving, of course, but 812 00:40:50,920 --> 00:40:53,480 Speaker 2: I'm going to do a Thanksgiving monologue either way, probably 813 00:40:53,480 --> 00:40:59,440 Speaker 2: a cz M CZM. I guess rewind that week as well. Anyway, 814 00:40:59,480 --> 00:41:11,120 Speaker 2: my brain's working. Thanks for listening. Thank you for listening 815 00:41:11,160 --> 00:41:12,120 Speaker 2: to Better Offline. 816 00:41:12,239 --> 00:41:14,680 Speaker 6: The editor and composer of the Better Offline theme song 817 00:41:14,760 --> 00:41:17,399 Speaker 6: is Matasowski. You can check out more of his music 818 00:41:17,400 --> 00:41:20,359 Speaker 6: and audio projects at Matasowski dot com. 819 00:41:20,520 --> 00:41:21,680 Speaker 3: M A T T O. 820 00:41:21,880 --> 00:41:25,400 Speaker 2: S O W s Ki dot com. 821 00:41:25,440 --> 00:41:27,759 Speaker 6: You can email me at easy at Better offline dot 822 00:41:27,800 --> 00:41:30,040 Speaker 6: com or visit better offline dot com to find more 823 00:41:30,040 --> 00:41:33,439 Speaker 6: podcast links and of course my newsletter. I also really 824 00:41:33,440 --> 00:41:35,759 Speaker 6: recommend you go to chat dot Where's youreed dot at 825 00:41:35,760 --> 00:41:38,200 Speaker 6: to visit the discord, and go to our slash Better 826 00:41:38,200 --> 00:41:39,840 Speaker 6: Offline to check out our reddit. 827 00:41:40,600 --> 00:41:43,880 Speaker 2: Thank you so much for listening. Better Offline is a 828 00:41:43,880 --> 00:41:45,280 Speaker 2: production of cool Zone Media. 829 00:41:45,400 --> 00:41:48,239 Speaker 6: For more from cool Zone Media, visit our website cool 830 00:41:48,320 --> 00:41:51,680 Speaker 6: Zonemedia dot com, or check us out on the iHeartRadio app, 831 00:41:51,719 --> 00:42:14,480 Speaker 6: Apple Podcasts, or wherever you get your podcasts.