Panic Responsibly: What Social Media Should Teach Us About AI의 전체 녹취록

if they got an earned media piece, put out a press release, put up a TV ad, you'd have to do very expensive polling to kind of know, or you'd have a general sense of the readership of that new news publication, whatever it might be. 00:05:13.199 --> 00:05:15.839 But now we had real-time analytics. 00:05:15.839 --> 00:05:19.680 And we thought that, right, like more clicks were better. 00:05:19.680 --> 00:05:23.279 So let's keep doing whatever gets us more clicks, whatever gets us more engagement. 00:05:23.360 --> 00:05:23.600 Yeah. 00:05:23.759 --> 00:05:27.439 And kind of like the proverbial frog being boiled, like in a campaign, right? 00:05:27.439 --> 00:05:29.600 Like you're just doing whatever you need to do to win. 00:05:29.600 --> 00:05:33.040 But I think it was very much over time. 00:05:33.040 --> 00:05:43.439 It wasn't an overnight thing, but over time, because we're measuring all that, you move more and more to the extremes to what will ever get you clicks, because then that's where the algorithms come in. 00:05:43.439 --> 00:05:47.839 The algorithms, human nature, whatever gets covered in the press, right? 00:05:47.839 --> 00:05:52.399 Like we're all responding to incentives that were being given. 00:05:52.399 --> 00:06:00.480 And those incentives caused us to get more and more extreme the the longer we kept doing online campaigning. 00:06:00.800 --> 00:06:01.199 Yeah. 00:06:01.199 --> 00:06:03.040 I I I think that's right. 00:06:03.040 --> 00:06:19.120 I I hadn't included that on my list, but the the idea that we could measure everything, there was never a question of should we, uh, but we could, because it was it was now on computers and we could look at what what's hitting the server, uh, really changed the the incentives. 00:06:19.120 --> 00:06:27.279 And and this is something that uh I I have to sort of help candidates and campaigners unlearn in all of my trainings. 00:06:27.279 --> 00:06:33.519 Is they're they're the making the graph go up and to the right is not our goal in politics. 00:06:33.519 --> 00:06:39.120 The winner is not the person with the most likes or the most follows or the most views, but the candidate who gets the most votes. 00:06:39.120 --> 00:06:40.319 And we did lose sight of that. 00:06:40.639 --> 00:06:57.759 Well, I was just gonna say, you mentioned fundraising too, and I do think fundraising is a big part of this as well, because particularly saw this in the 2010 cycle, right, with the NRSC, where you had the Tea Party candidates um, you know, winning primaries over incumbents, people that were not backed by the national party. 00:06:57.759 --> 00:07:08.319 And now that you had candidates who could start to fundraise all across the United States, they no longer needed the national party infrastructure to help them. 00:07:08.319 --> 00:07:18.240 So they had no incentive to also then necessarily work with the broader Republican or Democratic apparatus for everything. 00:07:18.240 --> 00:07:24.879 And so I think that that also um I think that may have also been something that that harmed this as well. 00:07:24.879 --> 00:07:30.480 Rather, so they were now playing for a national audience, not just the constituents that they were trying to serve. 00:07:30.800 --> 00:07:31.040 Yeah. 00:07:31.040 --> 00:07:36.399 And and you bring up a another confluence of events that had nothing to do with social media. 00:07:36.399 --> 00:07:39.120 And that was campaign finance reform. 00:07:39.120 --> 00:07:58.800 So when McCain Feingold gets passed, uh, which sort of weakened the party structure, they had no idea what was coming with with social media that that sort of these insurgent type candidates were were then going to be able to have the infrastructure that didn't exist when they were writing this. 00:07:58.800 --> 00:08:12.959 So it's a really interesting kind of confluence of of events where you say, okay, well, the parties are weaker and the tools for being on the outside are are better and and sort of just comes together at a at this interesting time. 00:08:12.959 --> 00:08:16.079 I think we're now starting to swing back to stronger parties. 00:08:16.079 --> 00:08:18.720 There's some you know recent Supreme Court decisions. 00:08:18.720 --> 00:08:26.560 Um but but the the interesting thing on the the analytics is that that colored so much of what we did in digital campaigning. 00:08:26.560 --> 00:08:31.600 Because we could say, look, well, you know, you run TV ads and it's gonna be wasted. 00:08:31.600 --> 00:08:41.440 I can show you who who got every view of this digital uh And I can test your messages before you actually spend the money behind it, right? 00:08:41.679 --> 00:08:41.919 Exactly. 00:08:41.919 --> 00:08:43.679 Because you can do all that A B testing, yeah. 00:08:44.000 --> 00:08:50.480 And and and we we we we're gonna guarantee you that there's no re waste and it's reaching only the voters you want to persuade. 00:08:50.480 --> 00:08:54.240 And and the the tail started wagging the dog a lot. 00:08:54.960 --> 00:09:01.600 Yeah, it really did because you were and it was how you were at least I'll speak for myself, it was how I was trying to justify more budget. 00:09:01.600 --> 00:09:02.240 Exactly. 00:09:02.240 --> 00:09:14.960 Like I remember too, in both Giuliani and NRSC, I oftentimes had to like, I basically had to pay my own lunch, meaning I was having to fundraise for whatever I wanted to do on digital. 00:09:15.120 --> 00:09:15.279 Yep. 00:09:15.519 --> 00:09:22.000 And so then you're trying to get creative around it because you're still in that time period, we're still trying to prove this, right? 00:09:22.000 --> 00:09:25.919 Like it's still pretty early days of digital campaigning. 00:09:25.919 --> 00:09:28.240 The iPhone is still relatively new. 00:09:28.240 --> 00:09:32.320 Mobile was just starting to take off in that 2010 cycle. 00:09:32.320 --> 00:09:37.360 And so there was still a lot of convincing we had to do of those TV,

right? 00:09:37.360 --> 00:09:39.360 Of the consultants and everyone who's wanting. 00:09:39.360 --> 00:09:42.399 I mean, heck, even still today, we see so much going up on broadcast. 00:09:42.399 --> 00:09:45.759 But um, but yeah, those analytics were the way that you did it. 00:09:46.080 --> 00:09:48.480 Yeah, and and and we finally caught the car. 00:09:48.480 --> 00:09:50.159 It's like the dog that catches the car. 00:09:50.159 --> 00:09:52.799 And and now people say, okay, yeah, digital's a thing. 00:09:52.799 --> 00:10:02.399 And you you see sort of this overcorrection where we're, you know, we're doing things like uh targeting ads too narrowly. 00:10:02.399 --> 00:10:07.200 Um so I think I think that's probably the the so so would you say uh I don't know. 00:10:07.200 --> 00:10:13.840 Well, would could we say that one of the mistakes might have been putting those analytics forward-facing, right? 00:10:13.840 --> 00:10:29.440 When you can see someone else's YouTube views going up or someone's likes on a post going up, that I wait if we took all of those off, I wonder if we might get back to the that open social web we used to have, or or it could be lost forever. 00:10:29.840 --> 00:10:33.840 I'm sure we would still try to fight find a way to have competition and get those metrics, right? 00:10:33.840 --> 00:10:38.559 Like I think that there's when you're in politics, you can't help it with the competition. 00:10:38.559 --> 00:10:39.360 But I do agree. 00:10:39.360 --> 00:10:42.080 I remember actually just jogged a memory in my brain. 00:10:42.080 --> 00:10:56.000 I remember back in when I first started at Facebook, I asked Mark if we would ever remove likes from the like count from pages because my theory was that like that didn't really indicate how many people were seeing your content on there. 00:10:56.000 --> 00:11:00.000 So, like, is it what does that vanity metric do? 00:11:00.000 --> 00:11:02.559 And I pretty much got laughed out of the room. 00:11:02.559 --> 00:11:14.000 It may not have been my best, my best first question to Mark, but it was something I was super curious about because I think that's another thing too about the analytics is there's what you can see forward facing. 00:11:14.000 --> 00:11:31.200 There's what you can see on your own analytics, but then there's also what the companies themselves know about who they're showing your content to, et cetera, um, that we're pretty much blind to because they're guessing who might be interested in our content that isn't already following us. 00:11:31.519 --> 00:11:32.000 Yeah. 00:11:32.000 --> 00:11:36.000 It really uh created a lot of headaches. 00:11:36.000 --> 00:11:52.399 I I mean, I I think um this idea that we could measure things, um, I I a big frustration I have, I would say it with with sort of digital campaigning writ large is sort of dashboard culture. 00:11:52.399 --> 00:11:56.240 Uh this idea that, well, you know, what is a dashboard? 00:11:56.240 --> 00:11:59.600

It's the thing in my car that I look at to decide, do I pull over and get gas? 00:11:59.600 --> 00:12:00.639 Am I going too fast? 00:12:00.639 --> 00:12:01.279 That kind of thing. 00:12:01.279 --> 00:12:08.559 But if if if we had dashboards on our car that look like what these social media platforms provide us in terms of analytics, we would crash. 00:12:08.559 --> 00:12:16.480 Um and so it it does seem like we we're losing the forest for the trees in many ways. 00:12:16.879 --> 00:12:22.879 Well, and I think too, the dashboards that we have don't like on online to continue that metaphor. 00:12:22.879 --> 00:12:30.960 I think in an ideal world, there would be something where you can say, please bring this the check engine light on if this kind of thing's happening, right? 00:12:30.960 --> 00:12:40.879 Like if I'm if I like to be able to set, like I don't think we well, I know I'll speak for myself again, didn't set values when we started to do this. 00:12:40.879 --> 00:12:43.200 Because again, the goal was just to win. 00:12:43.200 --> 00:12:47.919 It was not thinking about how is it do I want to approach this? 00:12:47.919 --> 00:12:50.720 What are my values and what I want to be putting out there? 00:12:50.720 --> 00:13:02.720 What do I think is going to be best for both, you know, me as a user, but also society writ large, but also for, you know, this campaign professionals profession profession, sorry, bad words there, but you know what I mean. 00:13:02.720 --> 00:13:11.919 Um, around it, uh in a way that I wish we still have time to do right now as we go into the AI era, but that window is closing as we go into the presidential cycle. 00:13:12.320 --> 00:13:12.559 All right. 00:13:12.559 --> 00:13:30.320 Before we dig into AI, I uh uh I the I I see a lot of similarities be between what's what's going on, uh, but but let's park on on social media for for just a little while longer, which is um I I I think you're right. 00:13:30.320 --> 00:13:38.240 We we were focused on how how do we win and and for campaigns, uh we're really focused on quarter by quarter. 00:13:38.240 --> 00:13:42.159 You know, what's our what's our fundraising, where where are we? 00:13:42.159 --> 00:13:54.000 Uh and and there is no, and this is by design, by the way, there's no one thinking about the long term of campaigns and the political industry and careers. 00:13:54.000 --> 00:13:59.519 I'm trying to change some of that now with the Center for Campaign Innovation, but there's just no money for it, right? 00:13:59.519 --> 00:14:04.399 When you're you're limited on what you can raise, there's a hard deadline on when the election happens. 00:14:04.399 --> 00:14:09.120 And so there's this interesting thing where we were focused on our incentives. 00:14:09.120 --> 00:14:17.919 The there were companies out there that that sort of saw what was going on and maybe didn't think about the implications. 00:14:17.919 --> 00:14:25.919 Uh, and then you've got regulators who weren't

really competent enough to to sort of see around the corner. 00:14:25.919 --> 00:14:27.759 So it's a again, setting that up. 00:14:27.759 --> 00:14:29.759 That's the mirror that we're we're looking back at. 00:14:30.000 --> 00:14:36.000 Well, and so you think back, so both on the quarter by quarter, I'd argue the companies are the same way with their product roadmaps. 00:14:36.000 --> 00:14:49.360 Like it was very hard to get Facebook to think long term about what because they're also in a quarterly financial like reporting type things that they're worried about and trying to think what products they need to get out there, et cetera. 00:14:49.360 --> 00:14:53.679 And I'd argue, again, I know we're gonna talk about AI, but like that feels to get compressed a lot more. 00:14:53.679 --> 00:15:10.720 And then on the regulator side, I think you went through this too when the FEC decided not to regulate bloggers and basically and didn't really kind of regulate online ads, but like the last time they did that advisory opinion, ads were still on the side of the Facebook desktop. 00:15:10.720 --> 00:15:13.759 They weren't even in, they weren't even in the news feed. 00:15:13.759 --> 00:15:21.519 And so, and now of course our FEC is pretty much defunct um um with how it's operated for how long? 00:15:21.519 --> 00:15:22.960 Quite quite a bit of time. 00:15:22.960 --> 00:15:30.159 And so, yeah, it's it's from the very beginning, it's been as we keep using the word the wild, wild west, but it really was. 00:15:30.480 --> 00:15:32.399 You're listening to the campaign trend podcast. 00:15:32.399 --> 00:15:41.759 I'm speaking with Katie Harbath about her new book, Disrupting Politics, a front row seat to the collision of technology and democracy out now. 00:15:41.759 --> 00:15:50.480 And Katie, during several of those years, I think it's fair to say that you and I were on opposite sides of the divide. 00:15:50.480 --> 00:15:54.639 You were at Facebook, I was on these campaigns trying to go right up to the the line. 00:15:54.639 --> 00:15:57.279 Uh thankfully, I hope we can still call each other friends. 00:15:57.279 --> 00:15:58.080 Uh oh yeah. 00:15:58.320 --> 00:15:59.919 And and you were doing your job. 00:15:59.919 --> 00:16:00.960 I was doing my job. 00:16:01.039 --> 00:16:02.000 Like, yeah. 00:16:02.000 --> 00:16:04.720 So, what was it like being on the inside? 00:16:04.720 --> 00:16:09.440 So you had this really cool big job at a huge company going all over the world. 00:16:09.440 --> 00:16:16.480 And meanwhile, the the political team that you had been rooting for was, I think, at odds from time to time. 00:16:16.480 --> 00:16:18.080 Um, what was that like? 00:16:18.399 --> 00:16:19.440 Um, oh my gosh. 00:16:19.440 --> 00:16:24.080 I mean, it was it got more challenging as the years went on, for sure. 00:16:24.080 --> 00:16:31.440 I think, you know, when I initially started at Facebook, um, there were only, you know, six or seven people in the DC office.

00:16:31.440 --> 00:16:36.000 The only person they had hired before that was Adam Connor, who came from Democratic politics, right? 00:16:36.000 --> 00:16:39.440 And so I was purposely hired for my Republican background. 00:16:39.440 --> 00:16:45.039 And I remember going to things like CPAC and other things, and they're like, wait, Facebook hired a Republican? 00:16:45.039 --> 00:16:54.159 Like they were kind of shocked that, because Silicon Valley was seen as very, very liberal because of how many people had gone and helped Obama over the years and stuff like that. 00:16:54.159 --> 00:16:57.039 And I talk about that a bit in the in the book. 00:16:57.039 --> 00:17:14.319 And so initially it was sort of like trying to, trying to just be that point person so that when you and others had problems, like I could help to escalate those throughout the company versus you all just, you know, complaining on a listserv that you can't find somebody on Facebook to help you to do that. 00:17:14.319 --> 00:17:33.759 Um, and it was, it was, and from day one, I was playing this bridging person between folks like yourself and the company and trying to explain to the company how politics worked, but then also trying to explain to you all why we as Facebook were making the decisions we were. 00:17:33.759 --> 00:17:38.480 Um, and then that got trickier as, and then, like I actually would say earlier on, right? 00:17:38.480 --> 00:17:44.240 You had things like the young guns who came in for a live stream, which was Paul Ryan, Kevin McCarthy, and Eric Canner. 00:17:44.240 --> 00:17:57.519 And coming to headquarters and coming to Silicon Valley was a thing that a lot of politicians wanted to do because they wanted to look cool and techy, and, you know, and they wanted that picture at the Googleplex or at Facebook campus and stuff like that. 00:17:57.519 --> 00:18:05.920 So I was spending a lot of time bringing people to campus out in California or having them over to the DC campus to talk about that. 00:18:05.920 --> 00:18:20.400 And, you know, then we had a big presence at the conventions, which was so much fun, um, where we actually were able to do live streams and have events and be able to have, you know, help showcase the company both on the left and the right. 00:18:20.400 --> 00:18:31.920 Um, but then we also continued to have challenges like the trending topics controversy, which happened in May of 2016, where a conservative um contractor uh claimed that. 00:18:31.920 --> 00:18:33.759 So we had like just like Twitter had, right? 00:18:33.759 --> 00:18:36.319 Like, what are people talking about the most on the platform? 00:18:36.319 --> 00:18:40.720 Um made the accusation that we were suppressing conservative content in there. 00:18:40.720 --> 00:18:46.880 And that undercurrent on the right, that Silicon Valley and Facebook's liberal just kind of broke out in the open. 00:18:46.880 --> 00:18:53.920 And we had to do a big meeting at headquarters with folks like Tucker Carlson, Jim Demint, Glenn Beck, et cetera, about all of that. 00:18:53.920 --> 00:19:00.559 But one of the things that came out of that meeting was then the company decided to get rid of the editors that were deciding that. 00:19:00.559 --> 00:19:07.519 That was kind of the start of some of the issues around fact-checking and some of the stories and stuff that were coming up around all of that. 00:19:07.519 --> 00:19:13.200 And then, and I'll pause after this point because but to because there's a whole nother chapter after the 2016 election. 00:19:13.200 --> 00:19:23.279 But like then the 2016 election happens and we start to institute a lot more hurdles for people like yourself to have to go through to use the platform. 00:19:23.680 --> 00:19:24.160 Yeah. 00:19:24.160 --> 00:19:26.079 It it it it is interesting. 00:19:26.079 --> 00:19:33.920 The arc was very much hey, we gotta work really hard to get people to start using this, and we're gonna we're gonna get them to use it. 00:19:33.920 --> 00:19:35.680 Uh oh, but don't use it like that. 00:19:35.680 --> 00:19:37.839 I mean, that that was sort of what it felt like. 00:19:37.839 --> 00:19:55.599 Um, so I'll just give a very specific example with um in 2012, the Obama campaign had this thing where you could connect your Facebook profile and see your friends that might be registered to vote, and their analytics team had modeled one way or another. 00:19:55.599 --> 00:20:00.240 On the so that's 2012, 2014, I'm working for Ed Gillespie's campaign for Senate. 00:20:00.240 --> 00:20:12.400 We're looking at, okay, well, we've we we were given a target by Facebook of, you know, because a little bit of history here, you weren't able to target ads to someone who didn't have a friend who didn't like your page. 00:20:12.400 --> 00:20:14.079 Uh oh, I forgot that. 00:20:14.079 --> 00:20:22.640 And so we had we invested a lot in making sure that we got about 1% of the electorate to like our page on Facebook. 00:20:22.640 --> 00:20:25.519 So we would have a good chance of being able to advertise them. 00:20:25.519 --> 00:20:36.079 But then we also implemented this thing where we said, hey, reach out to your friends, ask them to vote for uh vote for Ed if you know we think that they're a likely target. 00:20:36.079 --> 00:20:41.920 Well, by the time 2016 comes around, so again, we're in the we're looking at the span of four years. 00:20:41.920 --> 00:20:47.599 That was Cambridge Analytica, that was the end of democracy, that data shouldn't have ever left Facebook. 00:20:47.599 --> 00:20:55.599 And so it's just kind of a whiplash on our end of like, okay, it was brilliant when this team did it, yeah, and then sinister when we did it. 00:20:55.599 --> 00:20:59.920 Uh, so there's a little bit of whiplash there that was kind of frustrating. 00:21:00.319 --> 00:21:01.039 I I hear you. 00:21:01.039 --> 00:21:02.160 And it was for me too. 00:21:02.160 --> 00:21:22.319 And in fact, I opened the book with me

driving down Constitution Avenue the day after the 2016 election, not knowing what future I'm driving into, because the thing I spent 15 years building that everybody loved and was praising and was showcasing and loving turned villainous overnight. 00:21:22.319 --> 00:21:33.279 And now all of a sudden, it is the most horrible thing I could have done in all the land was to help, you know, candidates learn how to learn how to use this. 00:21:33.279 --> 00:21:41.440 And so it was a big whiplash for for me and my team internally too, because to try to navigate that. 00:21:41.440 --> 00:21:48.400 And we were navigating that even I talk about in the book in 20, it was probably 2018 after Cambridge Analytica. 00:21:48.400 --> 00:21:53.200 Um, I had a meeting where Cheryl was deciding if my team was gonna exist anymore. 00:21:53.200 --> 00:21:58.640 And I remember talking to the head of the sales team, and I was like, we're either digging our grave or building our next mountain. 00:21:58.640 --> 00:21:59.759 I don't know which one we're doing. 00:21:59.759 --> 00:22:06.240 And what we're walking into because there were so many conversations about maybe we just shouldn't do this at all. 00:22:06.240 --> 00:22:09.759 And internally we were trying to fight that this was still important to do. 00:22:10.079 --> 00:22:10.319 Yeah. 00:22:10.319 --> 00:22:33.920 My read on it, and here's what I I think that the the social media platforms, not just Facebook, I I don't think they understood the power and influence they were wielding when it came to politics, or they lacked the courage to accept that responsibility until it was probably too late and the cat was out of the bag, the the harm had been done. 00:22:33.920 --> 00:22:40.400 But looking at AI today, I don't think anybody learned the right lessons. 00:22:40.400 --> 00:22:55.920 And and so I I think, and I'm curious if you agree or where you might critique this, that we're overreacting in the wrong direction when it comes to AI, in insofar as, and I'll give one very concrete example. 00:22:55.920 --> 00:23:00.960 So Facebook, you know, didn't have any political policies for years and years. 00:23:00.960 --> 00:23:04.000 Open AI from the very outset has this policy. 00:23:04.000 --> 00:23:06.400 You can't use Chat GPT for politics. 00:23:06.400 --> 00:23:09.440 And we're now working on trying to claw back. 00:23:09.440 --> 00:23:10.319 Well, can we use this? 00:23:10.319 --> 00:23:10.880 Can we use this? 00:23:10.880 --> 00:23:11.519 Can we, you know? 00:23:11.519 --> 00:23:16.480 Um, and so they now have this policy that I think is is is fairly nonsensical. 00:23:16.480 --> 00:23:25.200 You know, you could uh research a policy about education and use chat GPT to brainstorm that, but you can't use it to draft a press release announcing that. 00:23:25.200 --> 00:23:29.200 And so it's like, all right, we we overcorrected the other way. 00:23:29.200 --> 00:23:32.240 How's that read from your perspective? 00:23:32.480 --> 00:23:32.640 Yeah. 00:23:32.640 --> 00:23:34.160 Oh, a hundred percent. 00:23:34.160 --> 00:23:39.279 And they've twisted themselves up into pretzels, trying to be like. 00:23:39.279 --> 00:23:46.559 So I usually, when I talk to these companies and talk to the folks inside these companies, they'll be like, what do we need to do? 00:23:46.559 --> 00:23:48.160 So so nobody gets mad at us. 00:23:48.160 --> 00:23:50.079 And I'm like, that is the wrong question. 00:23:50.079 --> 00:23:55.920 The question you need to ask yourself, yeah, is what risk tolerance do you have? 00:23:55.920 --> 00:23:58.960 And where are you willing to withstand pain? 00:23:58.960 --> 00:24:04.000 Because no matter what you decide, you will have people angry at you. 00:24:04.000 --> 00:24:20.000 And what you're seeing is, and and to be clear, like just like at Facebook, inside each of these companies are people who do understand how politics works and they are trying to push back on some of this or trying to find a way through. 00:24:20.000 --> 00:24:27.680 But then there's a lot of others who end up getting involved and you get this pretzel where you on the campaign can't figure out what it is. 00:24:27.680 --> 00:24:30.160 I keep asking them, like, what are the clear rules? 00:24:30.160 --> 00:24:42.319 Like, just tell the campaigns what they can do, but they can't get people internally focused on it until, like, from from an engineer standpoint, they think election day is just November 3rd. 00:24:42.319 --> 00:24:42.640 Right. 00:24:42.640 --> 00:24:48.000 But as we all know, or it didn't early, I mean, early voting started like a couple of weeks ago. 00:24:48.160 --> 00:24:48.480 Yeah. 00:24:48.720 --> 00:24:51.119 So you need to have it a lot sooner. 00:24:51.119 --> 00:25:00.960 And that is they are, they have not learned some of those lessons or they're not prioritizing it because they're also like, eh, it's a midterms, like it's not the presidential. 00:25:00.960 --> 00:25:04.400 And so it's just harder to get, it's harder to get resources. 00:25:04.400 --> 00:25:07.759 And it's really frustrating because I can just see the same things playing out again. 00:25:07.759 --> 00:25:20.400 But to your point, from the opposite end where it's starting super restrictive, and now they're trying to figure out what they what they open up versus it being super open and then slowly like putting guardrails around it. 00:25:20.720 --> 00:25:27.359 It really is uh it's it's sort of the pendulum swings, you know, um uh from one direction to the other. 00:25:27.359 --> 00:25:36.559 And and I I I think one of the I the the thing I would just say to people is like this time is not like last time, right? 00:25:36.559 --> 00:25:47.039 I I think they're and and applying the learning the exact well lear yes, learn the lessons from what happened with social media and politics. 00:25:47.039 --> 00:25:50.079 Um but this is different. 00:25:50.079 --> 00:25:52.960 This time is

different, and it's a it's a new ball of wax. 00:25:52.960 --> 00:26:08.480 I I think and again, I I totally take the perspective of uh, and I wonder if this isn't why, you know, we're we're both kind of solopreneurs now, because we don't have to deal with other people um uh trying to squelch what we're we're talking about. 00:26:08.480 --> 00:26:14.799 But um this idea that you know politics and Facebook would that was my life, you know, of every day. 00:26:14.799 --> 00:26:22.880 That because that's where the voters were, that's where the grassroots is, and and that was the kind of, you know, when you're on a campaign, that's the center of your universe, nothing else matters. 00:26:22.880 --> 00:26:30.160 Uh if you're Facebook, you're a multinational, multi-billion dollar company, you've got so many other things going on. 00:26:30.160 --> 00:26:44.480 So uh I I that that's a key perspective of, you know, there were smart people inside trying to make these arguments, but there might not have been the uh the bandwidth or decision-making capacity to to handle that in the world. 00:26:44.799 --> 00:26:54.160 We also were really we also were really, and I use the Royal We as Facebook there, very much responding to what was happening in the press. 00:26:54.160 --> 00:27:10.160 And so I remember, you know, multiple times when campaign professionals like yourselves or the parties really like for political ads, for instance, the number of people who came up to me and were like, please, please, please, Facebook, do not get rid of political ads. 00:27:10.160 --> 00:27:12.640 Like, please, please, please, do not. 00:27:12.640 --> 00:27:27.359 And I'm like, okay, then I need you all to give me some cover externally on this because I need, or I need something that I can take to leadership to show this, because they feel they felt like they were just getting hammered. 00:27:27.359 --> 00:27:30.880 And they're like, this in the grand scheme of things, this isn't that much money. 00:27:30.880 --> 00:27:31.279 Right. 00:27:31.519 --> 00:27:32.799 So not worth the headache. 00:27:32.799 --> 00:27:33.200 Yeah. 00:27:33.519 --> 00:27:36.799 Why is it worth the headache to even do? 00:27:36.799 --> 00:27:52.880 And that was a challenge as well, because you know, for I understood it, like if you're a campaign, on one hand, your your person might be a senator who's trying to push for, you know, something like the Honest Ads Act, but then they want to use the same stuff on their campaign. 00:27:52.880 --> 00:27:54.960 And it's like, all right, which one? 00:27:54.960 --> 00:27:58.640 Like, you guys gotta decide like what this is gonna look like. 00:27:58.640 --> 00:28:02.640 And I think that's part of the reason we haven't seen regulation is we're asking them to regulate themselves. 00:28:02.960 --> 00:28:03.200 Yeah. 00:28:03.200 --> 00:28:16.000 And and that's always hard to do, especially when, I mean, you and I, I think are are digital natives and we're we're fairly senior digital natives, and and people like us are just now starting to be in in Congress. 00:28:16.000 --> 00:28:19.680 And so I think that that is, you know, one of the challenges. 00:28:19.680 --> 00:28:22.000 Again, uh, there are AI natives. 00:28:22.000 --> 00:28:23.119 That's not me. 00:28:23.119 --> 00:28:28.720 Um and so I'm sure people are cringing at the way we talk about AI regulation and things like that. 00:28:28.720 --> 00:28:32.799 So um, well, I I I'm so excited to dig into this book. 00:28:32.799 --> 00:28:37.920 And just I know it's been a number of years uh laboring to get this here. 00:28:37.920 --> 00:28:42.720 Uh, and and it should be a fun uh walk down memory lane. 00:28:42.720 --> 00:28:59.839 Um any any closing thoughts for for folks about what what's the what's the lesson that that you learned um that we should not make the same mistake again in the future, whether that's AI, whether that's another social media question. 00:29:00.720 --> 00:29:04.799 Well, there's there's two things that I I end the book on. 00:29:04.799 --> 00:29:10.880 One is especially for for for folks working in AI, but I think that's very broad, right? 00:29:10.880 --> 00:29:13.440 Like I think we're all kind of working in this AI space. 00:29:13.440 --> 00:29:18.400 Yeah but I really think early on we like good intentions are not safeguards. 00:29:18.400 --> 00:29:21.599 Just because you mean well, it's not a good safeguard. 00:29:21.599 --> 00:29:23.440 And scale is not a strategy. 00:29:23.440 --> 00:29:32.720 And I feel with AI, we are in that same mode again, where it's just like, well, I mean well, maybe people will believe that. 00:29:32.720 --> 00:29:37.039 And it's like they're gonna go off your actions, not what just what you say. 00:29:37.039 --> 00:29:47.279 But I also talk about and I end with my phrase panic responsibly, because I think that it's important for all of us to remember as individuals, we still have agency. 00:29:47.279 --> 00:29:58.160 We still have agency over the tools we choose to use, we still have agency over what we choose to scroll, click on, consume, like put into these AI tools. 00:29:58.160 --> 00:30:05.920 And I think that that's important to remember that us as a voter and a citizen have a role in this conversation too. 00:30:05.920 --> 00:30:13.920 And we can't get paralyzed between, at least at the moment, we're being told if you use AI, it's cheating, but if you don't learn how to use it, you're gonna lose your job. 00:30:13.920 --> 00:30:34.000 And so it's we're in this uncomfortable spot where we're going to be in for a while, of this messy middle where you can only, you got to figure out what's the best step for you today, but understanding that you may have to pivot in a week or two of what that looks like, and that there really isn't a playbook for us right now. 00:30:34.000 --> 00:30:36.960 And it just reminds me a lot of the post-2016 world. 00:30:36.960 --> 00:30:39.519 And even before that, we didn't have playbooks. 00:30:39.519 --> 00:30:47.599 And maybe we don't, like you said, we can't copy that playbook from before, but we can learn the lessons of what we wish we knew when we were creating a new one. 00:30:47.920 --> 00:30:48.240 All right. 00:30:48.240 --> 00:30:50.720 Well, thanks to Katie for a great conversation. 00:30:50.720 --> 00:30:53.119 Her new book, Disrupting Politics, is out now. 00:30:53.119 --> 00:30:54.240 Go order it. 00:30:54.240 --> 00:30:56.079 Link is in the show notes. 00:30:56.079 --> 00:31:02.079 If this episode made you a little bit smarter or gave you something to think about, you know that all we ask is that you share it with a friend or colleague. 00:31:02.079 --> 00:31:04.880 You look smart in the process, more people will learn about the show. 00:31:04.880 --> 00:31:11.920 It's a win-win all around, and you can visit campaigntrend.com to check out our newsletters and past podcasts. 00:31:11.920 --> 00:31:13.599 With that, I'll say thanks for listening. 00:31:13.599 --> 00:31:15.200 We'll see you next time. 00:31:15.200 --> 00:31:22.720 The Campaign Trend Podcast is produced by Advocacy Cut, a media production company.