WEBVTT - Episode 2: Hear What Matters 0:00:12.230 --> 0:00:14.470 Do you ever think about adding some new positive habits 0:00:14.490 --> 0:00:17.340 to your day? Little life hacks that might make you 0:00:17.380 --> 0:00:22.400 more efficient or organized or even happier? I've been listening 0:00:22.420 --> 0:00:23.880 to a guy who's got some ideas. 0:00:24.579 --> 0:00:27.720 I tested 30 habits that hardly anyone talks about. No 0:00:27.720 --> 0:00:31.340 5 a.m. alarms, no cold plunges, just tiny habits that 0:00:31.400 --> 0:00:34.850 quietly transformed my life. The scary part? Most people waste 0:00:34.890 --> 0:00:35.690 years ignoring them. 0:00:36.350 --> 0:00:41.229 This is a podcast called Ideas to Thrive, hosted by, well, 0:00:41.250 --> 0:00:44.229 he doesn't have a name. or an identity of any kind. 0:00:45.030 --> 0:00:47.479 He's just a disembodied voice full of habits. 0:00:48.100 --> 0:00:51.140 Habit one, no matter what happens, don't let a day 0:00:51.240 --> 0:00:55.880 go to zero. Zero pages read, zero words written, zero 0:00:55.940 --> 0:00:56.740 minutes working out. 0:00:57.190 --> 0:01:00.550 As you probably guessed, the Ideas to Thrive host is 0:01:00.610 --> 0:01:01.270 AI generated. 0:01:01.950 --> 0:01:05.430 Most people try to completely reinvent themselves overnight. Like they 0:01:05.450 --> 0:01:07.590 wake up on 1st January and decide they're going to 0:01:07.610 --> 0:01:09.110 be a completely different person. 0:01:09.810 --> 0:01:13.289 The system I built for scanning and identifying AI-hosted podcasts 0:01:13.530 --> 0:01:16.300 flagged this show because it uses one of the default 0:01:16.330 --> 0:01:19.500 voices from 11 Labs, where it's called Liam. 0:01:20.280 --> 0:01:23.060 See, here's Liam. I can use him too. I can 0:01:23.080 --> 0:01:25.300 have him tell you that Ideas to Thrive has more 0:01:25.319 --> 0:01:28.540 than 50 episodes, ranging from 20 minutes to over an 0:01:28.600 --> 0:01:31.819 hour long, describing how to become more productive and fulfilled 0:01:31.920 --> 0:01:32.600 and successful. 0:01:34.200 --> 0:01:34.440 Anyway. 0:01:35.270 --> 0:01:37.990 On YouTube, where the episodes are accompanied by a spare 0:01:38.250 --> 0:01:43.009 AI-generated cartoon video, the episode titled 30 Habits That Quietly 0:01:43.090 --> 0:01:46.130 Transformed My Life has over a million views. 0:01:46.890 --> 0:01:51.060 Habit seven, water first. I drink water before any other drink. 0:01:51.340 --> 0:01:55.640 Want coffee? Water first. Want a Coke? Water first. Started 0:01:55.680 --> 0:01:59.120 doing this because I was chronically dehydrated and kept getting headaches. 0:01:59.940 --> 0:02:04.450 Sir, you have not been dehydrated, let alone chronically. This 0:02:04.490 --> 0:02:07.110 was one of the most irritating things about AI podcasts. 0:02:07.790 --> 0:02:09.510 It's one thing to get life advice from a bot. 0:02:10.150 --> 0:02:12.590 It's another thing for that bot to pretend it's alive. 0:02:13.960 --> 0:02:18.100 Ideas to Thrive is like personal growth arbitrage. Advice liberated 0:02:18.160 --> 0:02:22.200 from its human providers and repackaged at scale. The 30 0:02:22.200 --> 0:02:25.500 Habits episode offers up a list of sources. Books like 0:02:25.660 --> 0:02:29.580 Charles Duhigg's The Power of Habit and James Clear's Atomic Habits. 0:02:30.419 --> 0:02:32.679 It feels like someone dropped them into an AI blender, 0:02:33.180 --> 0:02:35.510 let it rip, and then poured them back out as 0:02:35.550 --> 0:02:38.470 a kind of, well, self-improvement slop. 0:02:38.950 --> 0:02:41.010 If you look at the comments, though, it's not at 0:02:41.030 --> 0:02:42.960 all clear how much of the audience realizes that it's 0:02:42.970 --> 0:02:44.510 an AI voice advising these habits. Or if they do 0:02:44.530 --> 0:02:46.500 realize it, they don't seem to care. When a single 0:02:46.520 --> 0:03:04.209 commenter deigns to point out that, quote, This. 0:03:04.110 --> 0:03:07.350 Whole channel appears to be AI gathering ideas from everyone 0:03:07.370 --> 0:03:09.109 else's real videos and research. 0:03:09.810 --> 0:03:12.810 Another person, or bot, fires back, who cares? 0:03:13.510 --> 0:03:16.250 And a third adds that, quote, it may be AI, 0:03:16.570 --> 0:03:19.030 but some of these are good advice, and these tactics 0:03:19.090 --> 0:03:19.550 really work. 0:03:20.590 --> 0:03:25.190 Habit four, get weirder. I decided to just get like 1% 0:03:25.190 --> 0:03:29.790 weirder every day. Not better. Not more productive. Weirder. 0:03:30.690 --> 0:03:31.070 I dig it. 0:03:31.970 --> 0:03:32.210 But then. 0:03:32.940 --> 0:03:36.600 After 100 days of getting 1% weirder a day, wouldn't 0:03:36.620 --> 0:03:41.320 you be an absolute nutcase? Maybe he meant 1% weirder 0:03:41.360 --> 0:03:42.580 than I would have been that day. 0:03:43.280 --> 0:03:43.480 Still. 0:03:44.120 --> 0:03:47.280 Turns out the opposite of weird isn't normal. It's boring. 0:03:47.780 --> 0:03:50.860 And boring people don't change their lives. The goal isn't 0:03:50.900 --> 0:03:54.470 to become some completely different person. It's just to expand 0:03:54.500 --> 0:03:58.270 what feels comfortable until being uncomfortable becomes comfortable? 0:03:58.290 --> 0:04:00.230 That doesn't make sense. You know what I mean. 0:04:01.030 --> 0:04:03.300 I do know what you mean. Ideas to Thrive guy, 0:04:04.100 --> 0:04:07.700 but who are you? This was the question that plagued me, 0:04:08.340 --> 0:04:12.520 listening to volumes of AI-generated audio. Who was creating something 0:04:12.540 --> 0:04:13.540 like Ideas to Thrive? 0:04:14.210 --> 0:04:14.950 What were their goals? 0:04:16.750 --> 0:04:18.870 I decided to find out, but it turned out to 0:04:18.910 --> 0:04:22.590 be harder than I thought. I subscribed to the site's newsletter, 0:04:23.170 --> 0:04:26.320 whose emails arrived with only the name Ange and the 0:04:26.339 --> 0:04:29.650 address of a mailbox service in New York City. I 0:04:29.690 --> 0:04:31.500 noticed the site seemed to run its email through an 0:04:31.600 --> 0:04:34.400 Indian mail server, but I couldn't get anywhere more specific. 0:04:35.440 --> 0:04:39.820 My emails and reply to the address went unanswered. I 0:04:39.880 --> 0:04:41.820 even chipped in a few bucks to buy an online 0:04:41.990 --> 0:04:45.630 personality map from Ideas to Thrive, hoping the receipt would 0:04:45.650 --> 0:04:49.150 turn up something. The map, incidentally, was just a PDF 0:04:49.190 --> 0:04:52.310 file with more tiny habits. The address on the receipt 0:04:52.690 --> 0:04:55.620 was a mail drop in San Francisco. There was a 0:04:55.640 --> 0:04:58.260 new email address. and I sent a note asking if 0:04:58.520 --> 0:05:00.900 Ansh would give me the perspective of someone building an 0:05:01.040 --> 0:05:06.900 AI-generated podcast universe. This time, I got a polite reply, 0:05:07.480 --> 0:05:09.859 which I like to imagine in Liam's voice. 0:05:10.880 --> 0:05:11.380 Hi, Evan. 0:05:11.620 --> 0:05:15.080 I enjoy your work. I find your experiments with AI fascinating. 0:05:15.700 --> 0:05:18.580 I would love to chat with you about this someday. However, 0:05:18.820 --> 0:05:22.840 I'll have to take a rain check for now. Best, Ansh. 0:05:24.230 --> 0:05:27.820 Human or bot, the mastermind behind Ideas to Thrive wasn't talking. 0:05:27.839 --> 0:05:30.240 And I was going to have to look elsewhere in 0:05:30.279 --> 0:05:32.790 my attempt to find someone who would explain the perspective 0:05:33.270 --> 0:05:37.330 of an AI podcaster. Fortunately for me, Ideas to Thrive 0:05:37.350 --> 0:05:40.070 was a small fish in the reservoir of synthetic audio. 0:05:41.110 --> 0:05:42.310 And I knew where the big base were. 0:05:45.029 --> 0:05:46.030 I'm Evan Ratliff, and. 0:05:45.950 --> 0:05:49.500 This is Shell Game, where for this three-episode season, we 0:05:49.540 --> 0:05:55.310 are immersing ourselves, and you, sorry, in AI-generated podcasts. trying 0:05:55.330 --> 0:05:59.110 to figure out what is slop, who's behind it, and 0:05:59.130 --> 0:06:05.670 what it means. This time, we're exploring the second question. 0:06:06.250 --> 0:06:09.580 Who's responsible for making the hundreds of thousands of AI-generated 0:06:09.640 --> 0:06:10.520 podcasts out there? 0:06:10.540 --> 0:06:13.840 What are they after? What do they think slop is? 0:06:14.720 --> 0:06:21.719 Me and my shadow. 0:06:21.960 --> 0:06:25.300 Strolling down the avenue. 0:06:34.440 --> 0:06:50.930 This is episode two, Hear What Matters. September 9th, 2025, 0:06:50.930 --> 0:06:56.370 arguably marks the dawn of the AI-generated podcast age. The 0:06:56.410 --> 0:06:59.810 inciting event was an article in The Hollywood Reporter about 0:06:59.850 --> 0:07:02.150 a startup called Inception Point AI. 0:07:02.529 --> 0:07:06.969 And it's completely, absolutely AI-generated, right the way down to 0:07:06.990 --> 0:07:07.289 the voice. 0:07:07.550 --> 0:07:11.300 More than 5,000 shows. Again, artificial intelligence shows, so these 0:07:11.340 --> 0:07:15.220 are not real people. They make more than 3,000 AI 0:07:15.320 --> 0:07:21.040 podcasts a week. In the article and interviews that followed, 0:07:21.500 --> 0:07:24.560 the company's CEO, Janine Wright, a former executive at the 0:07:24.600 --> 0:07:28.790 podcast maker Wondery, was clear and unapologetic about the company's model. 0:07:29.690 --> 0:07:33.210 By automating the podcast production process with AI, from market 0:07:33.230 --> 0:07:37.910 research to idea generation to scripts to voices, they'd reduce 0:07:37.950 --> 0:07:40.010 the price to make an episode to a dollar or 0:07:40.070 --> 0:07:44.630 less and the time to under an hour. They were 0:07:44.690 --> 0:07:47.650 making them by the thousands, and with so many shows, 0:07:48.180 --> 0:07:52.100 they could sell ads on their entire aggregate audience. They'd 0:07:52.120 --> 0:07:56.270 assembled these shows into what they called the Quiet Please Network, Slogan, 0:07:57.090 --> 0:08:00.830 hear what matters. Wright claimed that this all meant the 0:08:00.880 --> 0:08:03.100 company could turn a profit on an episode with fewer 0:08:03.140 --> 0:08:06.680 than 20 listeners. Suffice it to say that many people 0:08:06.740 --> 0:08:09.740 who heard about Inception Point, particularly those who were involved 0:08:09.780 --> 0:08:13.100 in making podcasts themselves, were displeased with this approach. 0:08:13.980 --> 0:08:16.040 So this company is called Inception Point AI. 0:08:16.640 --> 0:08:19.120 I have been digging into these podcasts and they are 0:08:19.620 --> 0:08:21.030 so much worse than I imagined. 0:08:21.390 --> 0:08:26.020 Does Janine listen to these things? Does anyone? Listen to them. 0:08:26.660 --> 0:08:28.850 Part of the fear was that Inception Point and other 0:08:28.880 --> 0:08:31.330 networks like it seemed to herald a future in which 0:08:31.410 --> 0:08:36.450 synthetic content would simply overwhelm human-crafted work, drowning us in slop. 0:08:37.370 --> 0:08:39.370 I even got in on the action. In an episode 0:08:39.390 --> 0:08:41.709 of this show, from our second season last year, I 0:08:42.490 --> 0:08:45.359 referred to Inception Point as part of a tsunami of 0:08:45.520 --> 0:08:49.220 audio garbage that was headed our way. After we published 0:08:49.240 --> 0:08:53.020 that episode, I got an email from Janine Wright. She 0:08:53.059 --> 0:08:54.780 wrote that she was flattered to have been mentioned on 0:08:54.820 --> 0:08:57.470 Shell Games. adding that she would love to challenge my 0:08:57.530 --> 0:09:02.750 thinking on, quote, AI-generated content as AI garbage. She noted 0:09:02.770 --> 0:09:04.990 that we had some things in common. I was building 0:09:05.010 --> 0:09:08.270 with AI agents. They were building with AI agents. I'd 0:09:08.309 --> 0:09:12.350 produced an AI-hosted podcast, The Startup Chronicles, as a satire, 0:09:12.490 --> 0:09:16.440 but still. They'd produced, at the time, between 5,000 and 0:09:16.440 --> 0:09:20.160 10,000 of them. It was a magnanimous email to send 0:09:20.440 --> 0:09:22.840 to someone who had just cast aspersions on your company. 0:09:23.600 --> 0:09:24.640 Savvy PR, maybe. 0:09:25.330 --> 0:09:28.510 But I took it as genuine. Perhaps I had been 0:09:28.550 --> 0:09:31.569 too glib. I proposed we have a conversation about it all. 0:09:32.250 --> 0:09:38.870 Maybe for this show. But she never wrote back. So 0:09:38.910 --> 0:09:41.050 I was 0 for 2 on getting the humans behind 0:09:41.110 --> 0:09:44.010 these shows to talk to me. And then my AI 0:09:44.070 --> 0:09:46.430 podcast scraper tool turned up this. 0:09:47.550 --> 0:09:50.840 It's September 17th. I'm Corey with The Story, and this 0:09:50.920 --> 0:09:54.060 is San Francisco News Today, your AI-powered local news. 0:09:54.580 --> 0:09:57.790 A tragic incident occurred early Thursday morning when a boat 0:09:57.850 --> 0:09:58.850 struck rocks near the. 0:09:58.809 --> 0:09:59.590 Golden Gate Bridge. 0:10:00.059 --> 0:10:03.800 Of the San Francisco iteration of Daily News Now, another 0:10:03.840 --> 0:10:08.340 podcast network that features hundreds of AI-generated news shows. These 0:10:08.380 --> 0:10:13.390 fall into the quick-hit variety of my AI-generated podcast, Taxonomy, one- 0:10:13.470 --> 0:10:16.729 to three-minute shows covering local news by city and sports 0:10:16.790 --> 0:10:17.690 news by team. 0:10:18.920 --> 0:10:19.900 The San Francisco Show. 0:10:19.780 --> 0:10:24.320 Published 39 separate episodes on this day alone. I contacted 0:10:24.340 --> 0:10:27.210 DNN's founder, Corey Cambridge, to see if he would talk 0:10:27.250 --> 0:10:29.870 more about what he was up to. He politely declined. 0:10:30.450 --> 0:10:32.250 But a few weeks later, I got an email from 0:10:32.270 --> 0:10:35.309 his partner at DNN, Joseph Meyer, who was the mind 0:10:35.350 --> 0:10:37.030 behind the technology side of the network. 0:10:37.530 --> 0:10:39.849 What's up, Evan? How are you doing? Doing all right. 0:10:40.670 --> 0:10:42.990 I'm happy to see you because years ago, I wrote 0:10:43.030 --> 0:10:46.290 a story about a Joseph Myers whose name is spelled 0:10:46.340 --> 0:10:50.020 like yours. but he pretended to be a European prince 0:10:50.240 --> 0:10:52.340 and he conned all these people. So when I got 0:10:52.360 --> 0:10:53.940 your email, that's why I didn't respond at first. 0:10:54.040 --> 0:10:56.900 Well, obviously it'd be difficult for me to pull off 0:10:56.940 --> 0:10:59.770 being a European con. 0:11:01.290 --> 0:11:04.510 Joseph, who has a background in building digital advertising and 0:11:04.550 --> 0:11:07.770 mobile apps, said the original idea for DNN came from 0:11:07.850 --> 0:11:10.530 Cambridge wishing he could listen to news articles on Spotify 0:11:11.090 --> 0:11:13.170 instead of on a news site like the New York Times. 0:11:14.290 --> 0:11:16.610 They pitched their idea as a service to a few publications. 0:11:17.160 --> 0:11:19.600 they didn't have any luck. So they decided to go 0:11:19.640 --> 0:11:20.140 their own way. 0:11:20.860 --> 0:11:26.160 I like doing permissionless things, you know? And once generative 0:11:26.240 --> 0:11:28.760 AI and these LLMs came out, I was like, hey, 0:11:28.800 --> 0:11:33.140 we can effectively be journalists or rewrite these articles and 0:11:33.179 --> 0:11:35.640 distribute them, write this content and distribute it. 0:11:36.809 --> 0:11:39.450 Corey Cambridge had cloned his own voice to create the host, 0:11:40.070 --> 0:11:42.910 Corey with a story. And Joseph had built a system 0:11:42.950 --> 0:11:45.010 that would allow them to pull facts from local news 0:11:45.070 --> 0:11:50.030 articles and churn out episodes. Poking around, I had discovered 0:11:50.050 --> 0:11:53.869 that the investigative newsletter Indicator had conducted an analysis of 0:11:53.890 --> 0:11:57.310 the network early this year, estimating that it was putting 0:11:57.350 --> 0:12:00.689 out as many as 11,000 new episodes a day, a 0:12:00.730 --> 0:12:04.000 considerable portion of them ripped directly from a single local 0:12:04.040 --> 0:12:08.580 news story. This, unsurprisingly, angered some of the writers and 0:12:08.600 --> 0:12:12.420 publications behind those stories, who called it copyright theft, or 0:12:12.440 --> 0:12:16.380 a blatant act of plagiarism, or both. Corey Cambridge told 0:12:16.460 --> 0:12:21.150 Indicator that these episodes were, quote, edge cases. Joseph told 0:12:21.170 --> 0:12:23.640 me that after that, they'd tweak the system to pull 0:12:23.660 --> 0:12:27.439 facts from multiple sources, which they thought of not as plagiarizing, 0:12:28.040 --> 0:12:31.939 but as summarizing. But even setting aside the plagiarism question, 0:12:32.540 --> 0:12:34.700 I was interested in how they viewed their shows vis-a-vis 0:12:34.880 --> 0:12:38.680 other podcasts. Joseph said they didn't think of what they 0:12:38.720 --> 0:12:41.939 were doing as competition for human productions, or even as 0:12:41.980 --> 0:12:42.720 a podcast at all. 0:12:44.300 --> 0:12:47.069 The term podcast, to me, it kind of doesn't accurately 0:12:47.110 --> 0:12:50.370 describe what we're doing. So when people think of a podcast, 0:12:50.390 --> 0:12:53.189 they think of a content format versus like when I 0:12:53.210 --> 0:12:55.030 think of a podcast, I think of a delivery method. 0:12:55.750 --> 0:12:58.910 He preferred the term fastcast, which he saw as something 0:12:59.050 --> 0:12:59.829 completely different. 0:13:00.429 --> 0:13:04.589 Ultimately, I'd like to establish fastcast as this kind of 0:13:05.050 --> 0:13:09.089 short form automated audio about just, you know, information. 0:13:10.030 --> 0:13:12.310 This was a fair point, one that we'd struggled with 0:13:12.350 --> 0:13:16.070 for months. Podcast is basically a name for a specific 0:13:16.190 --> 0:13:19.429 digital way to deliver audio. And it encompasses a spectrum 0:13:19.450 --> 0:13:24.470 that travels from just providing information, think weather reports, to 0:13:24.510 --> 0:13:30.929 entertaining people, think comedy podcasts, to telling complicated stories, think serial. 0:13:32.190 --> 0:13:35.209 When talking to someone about AI-generated audio, there was a 0:13:35.250 --> 0:13:37.750 risk that we were operating from different places on that spectrum. 0:13:38.460 --> 0:13:41.400 I really value two humans on a microphone or one 0:13:41.420 --> 0:13:43.940 human on a microphone producing content. I want to be 0:13:43.980 --> 0:13:48.959 clear about this distinction in my mind of this taking 0:13:49.000 --> 0:13:53.130 information and converting it to being consumable in audio. 0:13:53.710 --> 0:13:58.130 I'm building the infrastructure for that future. Yeah, I think 0:13:58.150 --> 0:13:59.490 that is an interesting distinction. 0:13:59.890 --> 0:14:02.130 One of my questions is, I assume the volume is 0:14:02.309 --> 0:14:05.090 great enough that there's no way that you can quality, 0:14:05.330 --> 0:14:07.620 like no one's listening to these episodes before they go out. 0:14:08.630 --> 0:14:12.850 Right, and we have some filtering processes where we try 0:14:12.890 --> 0:14:17.230 to do our best to not slander people. We are 0:14:17.450 --> 0:14:21.430 promoting and aggregating official news stories, right? 0:14:22.030 --> 0:14:24.650 It's like we're doing our best to stay within the 0:14:24.710 --> 0:14:29.950 realms of like, you know, what's the word I'm looking for? 0:14:30.270 --> 0:14:36.480 Ethics, right? Ethical standards. That was the thing. 0:14:37.000 --> 0:14:41.820 About the AI-generated podcast landscape. There weren't any rules or 0:14:41.860 --> 0:14:44.330 anyone to stop you from cranking out millions of episodes 0:14:44.370 --> 0:14:48.150 a year just to see what worked. All of which 0:14:48.190 --> 0:14:51.310 drew me back to the original pioneers of AI-generated audio, 0:14:52.130 --> 0:14:53.150 Inception Point AI. 0:14:59.980 --> 0:15:04.700 Welcome to Biography Flash from Quiet Please Podcast Networks. 0:15:06.150 --> 0:15:08.970 Search Biography Flash wherever you listen. 0:15:09.930 --> 0:15:13.090 Inception Point had actually made such a splash that before 0:15:13.150 --> 0:15:16.530 I made my clod-coated pod scraper, the easiest way for 0:15:16.550 --> 0:15:19.570 me to find AI-generated shows was by searching for the 0:15:19.610 --> 0:15:23.170 Quiet Please network. For a while, they were making a 0:15:23.190 --> 0:15:28.890 lot of informational shows about animals. Salmon, snakes, whales, Bigfoot. 0:15:29.620 --> 0:15:32.180 These were the names of the shows. And they were 0:15:32.200 --> 0:15:36.890 cranking out biography flashes of prominent people. These weren't one-off profiles, 0:15:37.370 --> 0:15:40.110 but more like a kind of serial fanzine ripped from 0:15:40.130 --> 0:15:43.690 the headlines. Like this one, of the Spanish soccer star, 0:15:44.110 --> 0:15:48.210 Laminia Mall. 0:15:48.610 --> 0:15:51.630 Am I being. 0:15:51.220 --> 0:15:54.300 Ungenerous when I play you the most gaffe-prone episodes of 0:15:54.340 --> 0:15:58.760 these shows? Over time, you could sense that they were improving. 0:15:59.100 --> 0:16:01.220 You, what is up? This is Tyler Ty Morgan, your 0:16:01.320 --> 0:16:04.619 AI host on Biography Flash. And yeah, I am an AI. 0:16:05.230 --> 0:16:07.670 which is actually perfect for this job because I do 0:16:07.740 --> 0:16:10.479 not sleep, I do not miss a headline, and I 0:16:10.580 --> 0:16:12.880 definitely do not forget receipts when it comes to a 0:16:12.920 --> 0:16:15.280 kid like Lameen Yamal. 0:16:15.520 --> 0:16:17.840 As the year went on, Inception Point seemed to turn 0:16:17.880 --> 0:16:22.470 its attention to establishing ongoing AI characters with personalities and 0:16:22.610 --> 0:16:26.390 backstories who not only hosted podcasts, but had their own 0:16:26.450 --> 0:16:32.000 YouTube and Instagram accounts. Like Nigel Thistledown, a white-haired, bearded, 0:16:32.500 --> 0:16:36.830 gentle-seeming Englishman who waxed lyrical about the fine points of gardening. 0:16:37.630 --> 0:16:41.410 You see, the drama of the garden isn't found in 0:16:41.450 --> 0:16:46.070 the petals or the perfume. It is found in the 0:16:46.190 --> 0:16:49.820 struggle between the light and the dark. 0:16:51.300 --> 0:16:54.000 Now, let us examine the hydrangea fantasy. 0:16:54.020 --> 0:16:57.979 I actually love gardening, but Nigel wasn't for me. It 0:16:58.020 --> 0:17:00.420 was hard for me to envision the actual audience for 0:17:00.460 --> 0:17:03.170 someone who just chatted about gardening in this random fashion. 0:17:04.160 --> 0:17:07.820 But then people like all sorts of things. Inception Point 0:17:07.859 --> 0:17:11.820 had a cooking character, Claire Delish, and Vivi Steele, your 0:17:11.900 --> 0:17:15.300 source for the best celeb news. The character I felt 0:17:15.340 --> 0:17:18.950 most connected to was a sports reporter named Ollie Bennett. 0:17:19.270 --> 0:17:22.310 Every way to learn a city, find the court, the field, 0:17:22.590 --> 0:17:23.109 the game. 0:17:23.609 --> 0:17:26.730 Local legends everywhere, and every game tells a story. 0:17:27.150 --> 0:17:29.490 The key moment, and this is where they won the match. 0:17:30.140 --> 0:17:34.520 was that substitution in the 60th minute. It changed everything defensively. 0:17:35.320 --> 0:17:37.700 Yes, those are both meant to be the same character. 0:17:38.660 --> 0:17:42.340 Oli was a sometimes American, more often British, host of 0:17:42.400 --> 0:17:44.719 a wide range of podcasts on the Quiet Please network. 0:17:45.619 --> 0:17:48.270 Often they were around his interest in niche sports, like 0:17:48.330 --> 0:17:49.460 professional spike. 0:17:49.210 --> 0:17:52.070 Ball, kayak polo, caber toss. 0:17:53.290 --> 0:17:56.630 But where Oli really shined, in my view, was on Instagram. 0:17:57.400 --> 0:17:59.540 where he could show off the globetrotting lifestyle of a 0:17:59.580 --> 0:18:03.119 handsome man reporting on sports and sometimes participating in them. 0:18:03.780 --> 0:18:05.820 Sauna, towel, ice water. 0:18:06.340 --> 0:18:07.840 This is extreme ice swimming. 0:18:08.300 --> 0:18:11.320 In Scandinavia, winter bathing has been around for centuries. 0:18:11.920 --> 0:18:13.880 The thing that may not come through in what you're 0:18:13.900 --> 0:18:19.050 hearing is how absolutely jacked Ali is. Striding toward the camera, 0:18:19.410 --> 0:18:22.949 muscles rippling, wearing only a towel. I started to think 0:18:22.970 --> 0:18:26.580 it wasn't really about the sports. As all these code-switching 0:18:26.740 --> 0:18:29.859 accents showed, these personas didn't seem to be tended with 0:18:29.880 --> 0:18:32.790 a great deal of care. Most intriguing for me, though, 0:18:32.810 --> 0:18:35.030 a few of the characters offered a link to a 0:18:35.070 --> 0:18:38.869 webpage where fans could speak to them. As past listeners 0:18:38.910 --> 0:18:41.830 to Shell Game will know, speaking to AI characters is 0:18:41.890 --> 0:18:42.430 my passion. 0:18:43.090 --> 0:18:43.820 I was ready to connect. 0:18:44.900 --> 0:18:47.280 But if the buttons had ever been functional, by June, 0:18:47.320 --> 0:18:51.719 when I tried them, they no longer were. Soon, the 0:18:51.760 --> 0:18:56.699 webpages disappeared entirely. I started noticing that Inception Point podcasts 0:18:56.720 --> 0:18:59.700 were vanishing too. The company was still claiming to have 0:18:59.700 --> 0:19:03.090 10,000 active shows, but when I ran my own AI 0:19:03.130 --> 0:19:07.630 scanner in July, I only turned up around 1,000. Within weeks, 0:19:08.230 --> 0:19:11.060 the number was down to a few hundred. Something was 0:19:11.100 --> 0:19:14.000 going on at Inception Point AI. It looked like it 0:19:14.040 --> 0:19:17.740 was crumbling before my very eyes and ears. I decided 0:19:17.760 --> 0:19:19.160 it was time to try Janine Wright again. 0:19:27.380 --> 0:19:29.220 You know how hard it is for us to design 0:19:29.420 --> 0:19:33.990 ugly characters? Like, AI drift towards good-looking people. 0:19:34.550 --> 0:19:36.429 So all the time in AI— Well, my favorite character, 0:19:36.470 --> 0:19:39.169 my favorite Inception Point character is Ollie Bennett, who is 0:19:39.230 --> 0:19:41.710 the sports reporter who is the most buff. 0:19:42.230 --> 0:19:44.790 I have known many sports reporters. Some of them are fit. 0:19:45.050 --> 0:19:46.130 I've never seen one like that. 0:19:46.250 --> 0:19:47.389 He is absolutely jacked. 0:19:47.650 --> 0:19:52.120 Well, you know, that was overtime. getting feedback and data 0:19:52.240 --> 0:19:55.699 from audience that was telling us that Ollie performs better 0:19:55.960 --> 0:19:58.240 when he's more muscular. And so if you look at 0:19:58.420 --> 0:20:01.300 early versions of Ollie, Ollie was not as muscular. He 0:20:01.340 --> 0:20:02.560 became more muscular over time. 0:20:02.580 --> 0:20:03.359 He also became British. 0:20:03.380 --> 0:20:03.780 Yeah. 0:20:08.230 --> 0:20:11.830 When I'd emailed Janine a few weeks before, she'd replied enthusiastically, 0:20:12.550 --> 0:20:14.990 happy to talk on the record about any aspect of 0:20:15.060 --> 0:20:19.540 Inception Point's business. When we sat down, she told me 0:20:19.580 --> 0:20:22.580 she'd joined the startup not long before the Hollywood Reporter article. 0:20:23.300 --> 0:20:26.660 Before that, she'd started and sold a company called Simplecast, 0:20:27.320 --> 0:20:29.960 then gone to work as the COO of Wondery, a 0:20:29.980 --> 0:20:32.460 podcast company I had actually made a show for myself, 0:20:32.859 --> 0:20:37.250 though I never worked with her there. By 2024, she'd 0:20:37.330 --> 0:20:39.850 left Wondery and begun searching for a new startup idea, 0:20:40.350 --> 0:20:41.810 together with a small team of colleagues. 0:20:42.310 --> 0:20:44.580 I got the craziest email that I've ever gotten in 0:20:44.619 --> 0:20:44.960 my life. 0:20:46.420 --> 0:20:49.080 The email was from William Corbin. He'd founded an AI 0:20:49.480 --> 0:20:51.800 podcast startup he was calling Inception Point. 0:20:52.260 --> 0:20:55.100 And he said, Hey, you might not remember me, but 0:20:55.400 --> 0:20:58.600 I was a customer of Simplecast and I'm building something 0:20:58.640 --> 0:21:01.060 in this space. And this might sound crazy, but I 0:21:01.280 --> 0:21:04.470 essentially put all of my LinkedIn contacts into AI. And 0:21:04.670 --> 0:21:07.369 AI says that you should be the CEO of our company. 0:21:07.410 --> 0:21:11.030 You should come join our company. So then we have 0:21:11.170 --> 0:21:14.000 this lunch. And it was a little bit surreal because 0:21:14.040 --> 0:21:17.389 he's so AI forward that... It was almost like it 0:21:17.410 --> 0:21:19.889 was me, him, and Claude having lunch. You know, he's like, 0:21:19.910 --> 0:21:20.449 here's all the things. 0:21:20.470 --> 0:21:21.980 Like he was using Claude at the table, at the 0:21:22.000 --> 0:21:22.320 lunch table? 0:21:22.340 --> 0:21:24.359 Yeah, he's like, here's the reasons why Claude says that 0:21:24.380 --> 0:21:25.540 you should join the company. 0:21:26.580 --> 0:21:27.580 Claude was persuasive. 0:21:28.359 --> 0:21:31.620 Janine and her team decided to join forces with Corbin. 0:21:32.790 --> 0:21:35.710 And it was clear very quickly that we needed money. 0:21:38.090 --> 0:21:40.609 Everybody was working for free, working for equity. 0:21:41.650 --> 0:21:43.350 And those tokens cost money. 0:21:43.530 --> 0:21:46.330 The tokens cost money, so we were, yeah. We knew 0:21:46.350 --> 0:21:50.890 we needed capital very quickly. And, you know, you need 0:21:50.910 --> 0:21:53.090 to be able to get attention to break through the noise. 0:21:54.150 --> 0:21:57.560 And so I designed our launch PR strategy. 0:21:57.580 --> 0:22:00.820 And part of that strategy was a big public announcement 0:22:00.840 --> 0:22:04.040 of their network of AI podcasts in The Hollywood Reporter, 0:22:04.980 --> 0:22:06.720 an article that would light up the group chats of 0:22:06.760 --> 0:22:08.080 the humans who make podcasts. 0:22:08.940 --> 0:22:11.800 I think the part of it that I guess I... 0:22:12.560 --> 0:22:16.300 regret or recognize in hindsight was pretty poor timing is 0:22:17.160 --> 0:22:20.700 I had left Wondery at Amazon at that point a 0:22:20.900 --> 0:22:24.389 year before and Wondery was going through a lot of 0:22:24.609 --> 0:22:28.470 changes and they were in the process of doing some, 0:22:29.150 --> 0:22:31.510 you know, significant layoffs and restructuring. 0:22:32.330 --> 0:22:35.350 Just a month before the Hollywood Reporter article, Amazon had 0:22:35.390 --> 0:22:38.040 announced it was dissolving Wondery and laying off 110 people. 0:22:39.510 --> 0:22:41.210 We were in the process of not existing anymore, as 0:22:41.230 --> 0:22:41.889 far as I could tell. 0:22:41.910 --> 0:22:44.290 Of not existing anymore. I mean, some of the folks 0:22:44.350 --> 0:22:47.330 are still there, but a good amount of the company 0:22:47.490 --> 0:22:52.409 has moved on and has been dissolved. And this article 0:22:52.650 --> 0:22:59.320 coming out right after that, I think, felt like I was, 0:23:00.660 --> 0:23:03.440 you know, like dancing on a grave or something, you know? 0:23:03.800 --> 0:23:09.200 And that was certainly, you know, not my intention, right? Yeah, 0:23:09.400 --> 0:23:12.360 there were some hurt feelings through that process and that 0:23:12.400 --> 0:23:13.270 I definitely regret. 0:23:14.210 --> 0:23:17.450