WEBVTT - Surveillance: This AI Cycle Is Different, NYU's Dhar Says 0:00:09.840 --> 0:00:13.840 Welcome to the Bloomberg Surveillance Podcast. I'm term Keene Jay Leye. 0:00:13.960 --> 0:00:17.560 We bring you insight from the best in economics, finance, investment, 0:00:18.000 --> 0:00:23.520 and international relations. Find Bloomberg Surveillance on Apple Podcasts, SoundCloud, 0:00:23.600 --> 0:00:30.159 Bloomberg dot Com, and of course on the Bloomberg. In 0:00:30.160 --> 0:00:33.159 the bond market, we shape up as follows treasuries tens 0:00:33.520 --> 0:00:36.599 yields unchanged two point six three two on a two 0:00:36.680 --> 0:00:40.080 year note two point four six three percent. That's where 0:00:40.120 --> 0:00:42.400 I want to start this morning, at the front end 0:00:42.479 --> 0:00:44.080 of the yield curve, and we have a great gust 0:00:44.120 --> 0:00:46.760 to do it with Jerum Schneider, PIMCO, Head of short 0:00:46.920 --> 0:00:50.320 term bond Portfolios. Good morning to Jerome, Good morning, Good morning. 0:00:50.400 --> 0:00:52.519 Let's just start with two lines that I picked up 0:00:52.560 --> 0:00:54.880 from the notes I saw in your most recent research. 0:00:54.960 --> 0:00:58.200 Short term bond strategy to manage two things, potential defense 0:00:58.240 --> 0:01:01.760 in volatile markets and the optionality to move to higher 0:01:01.800 --> 0:01:05.600 risk allocations opportunistically. Let's just walk through those two issues 0:01:05.680 --> 0:01:07.800 right now. Why do I need the potential defense in 0:01:07.880 --> 0:01:10.839 volatile markets with VOLTSI love so. The question that remains 0:01:10.840 --> 0:01:13.280 for investors is how long is this economic cycle going 0:01:13.280 --> 0:01:16.440 to persist for and admittedly the financial conditions have eased 0:01:16.480 --> 0:01:18.320 quite a bit since the concerns of the fourth quarter, 0:01:18.400 --> 0:01:22.080 but monetary policy is only one ingredient to to the 0:01:22.120 --> 0:01:26.520 market conditions that helped perpetuate positive returns going forward. Ultimately, 0:01:26.560 --> 0:01:29.800 what we're gonna be dealing with is continued bouts of illiquidity, 0:01:29.800 --> 0:01:34.319 continued increases of volatility within the marketplace. And what's really 0:01:34.360 --> 0:01:37.800 happened is investors, whether your retail or investors or institutional 0:01:38.120 --> 0:01:41.679 have to be thinking about returns or expected returns from 0:01:41.680 --> 0:01:44.400 the construct of not only thinking about, hey, I could 0:01:44.400 --> 0:01:47.840 earn five seven percent or even more obviously historically in 0:01:47.840 --> 0:01:51.600 equities or emerging markets those higher higher beta allocations, but 0:01:51.760 --> 0:01:55.240 doing so with higher volatility expectations. In other words, we 0:01:55.320 --> 0:01:58.200 need to calibrate expectations for returns on a risk adjusted 0:01:58.200 --> 0:02:02.040 basis taking into account more volatility due to the uncertainty 0:02:02.080 --> 0:02:04.520 within the marketplace. So at the moment, volatility is really 0:02:04.560 --> 0:02:08.160 loud cross asset exactly what binds effects, et cetera. But 0:02:08.200 --> 0:02:10.440 think about how quick that healing process happened, and think 0:02:10.480 --> 0:02:13.120 how quickly how many bouts of illiquity we've had simply 0:02:13.120 --> 0:02:16.400 over the past three years and illiquidity. I'm using as 0:02:16.440 --> 0:02:19.240 a proxy here for volatility. But the volatility could emanate 0:02:19.280 --> 0:02:22.000 from politics. It could it could emanate from left tail 0:02:22.080 --> 0:02:25.120 risks such as China trade. And while these risks have 0:02:25.240 --> 0:02:28.400 actually been elongated, the runways of these risks, whether from 0:02:28.440 --> 0:02:33.000 monetary policy, trade policy, or even Brexit, they've elongated. The 0:02:33.040 --> 0:02:36.640 uncertainty could actually persist for quite some time over the 0:02:36.760 --> 0:02:38.920 over the next few years. So while we have a 0:02:39.000 --> 0:02:42.240 relative period of warmth as as spring approaches, don't be 0:02:42.480 --> 0:02:44.040 don't be fooled, and we should be prepared for that 0:02:44.120 --> 0:02:46.720 higher volatility. I just figured this out. Are you here 0:02:46.720 --> 0:02:48.200 to do a double? Is he gonna be on the 0:02:48.240 --> 0:02:52.360 real yield today? One television at least? But you're just 0:02:53.000 --> 0:02:55.480 this is your power, This is the power of feraoh 0:02:55.800 --> 0:02:59.360 booking the brightest guy in the world. Short term paper. Yeah, 0:03:00.280 --> 0:03:02.800 what the commercial paper used to be the thing we 0:03:02.800 --> 0:03:06.520 looked at the pulse. What's the new commercial paper? Guys 0:03:06.520 --> 0:03:08.760 like you look at every day. I'll tell your story, Tom, 0:03:08.800 --> 0:03:10.760 it's really interesting. So two years ago, when I was 0:03:10.760 --> 0:03:13.240 sitting in these seats, we're talking about the evolution of 0:03:13.280 --> 0:03:15.840 money market funds. Money market fund reform and putting people 0:03:15.880 --> 0:03:17.600 to sleep, telling him that the end of the money 0:03:17.600 --> 0:03:20.120 market fund world was coming soon, and we saw a 0:03:20.160 --> 0:03:22.760 trillion dollars worth of prime money market fund assets. Those 0:03:22.800 --> 0:03:25.799 that buy commercial paper go and buy government only money 0:03:25.840 --> 0:03:28.400 organt funds, those buy T bills de risking. So the 0:03:28.440 --> 0:03:30.920 story now has actually changed over the past few months. 0:03:31.120 --> 0:03:33.200 People have been lulled to sleep that they want to 0:03:33.280 --> 0:03:35.880 de risk. They're de risking out of these higher volatility 0:03:36.160 --> 0:03:39.560 classes equities, and they're putting their money, ironically back into 0:03:39.560 --> 0:03:42.320 prime money market funds. There's been a hundred billion dollars 0:03:42.320 --> 0:03:44.760 of new assets put into prime money market funds for 0:03:44.800 --> 0:03:47.760 an incremental return of just a few basis points over 0:03:47.800 --> 0:03:50.320 the safety of T bills. So why would you do that? 0:03:50.440 --> 0:03:53.160 And I think ultimately the world of commercial paper has 0:03:53.240 --> 0:03:56.120 once again become rather complacent to the to the notion 0:03:56.240 --> 0:03:58.200 of any type of credit risk, and we have to 0:03:58.240 --> 0:04:01.680 be more discerning in terms of credit risk. Clearly, you 0:04:01.760 --> 0:04:06.160 can steal that for the real appreciating seeds. So let's 0:04:06.200 --> 0:04:08.440 go to the next line to to have the optionality 0:04:08.640 --> 0:04:13.200 to move to higher risk allocations opportunistically. Are you looking 0:04:13.240 --> 0:04:16.040 for another December kind of moment, not an end of 0:04:16.120 --> 0:04:20.200 cycle moment, but a moment where that risk aversion builds, 0:04:20.680 --> 0:04:23.719 the faith in the cycle fades rapidly, and you get 0:04:23.760 --> 0:04:26.360 to deploy some capital into risk assets. So the starting 0:04:26.360 --> 0:04:30.320 point of analysis and portfolio management is diversification, but more importantly, 0:04:30.360 --> 0:04:33.880 managing downside risks. And we've really viewsed that managing downside 0:04:33.920 --> 0:04:36.800 risk is beginning in two thousand eighteen at Pimcoe, where 0:04:36.839 --> 0:04:39.960 we were really thinking about focusing on corporate credit and 0:04:40.000 --> 0:04:44.240 differentiating corporate credit from the data the index, picking specific 0:04:44.279 --> 0:04:46.679 stories that we liked, under accentuating those that we didn't, 0:04:46.800 --> 0:04:49.240 and diversizing. Now, granted, this was a process that began 0:04:49.279 --> 0:04:52.159 in late two thousand seventeen early two eighteen, and we 0:04:52.160 --> 0:04:54.880 didn't know when when actually the opportunity would would persist, 0:04:54.880 --> 0:04:56.520 but we thought, you know, there would be some some 0:04:56.560 --> 0:04:59.400 tail risks along the way, maybe some rude awakenings, but 0:04:59.560 --> 0:05:01.839 ultimately the growth, the slowing growth that we were seeing 0:05:01.880 --> 0:05:04.120 in the globe economy would translate into opportunity. When it 0:05:04.200 --> 0:05:06.800 happened in the fourth quarter, we had access liquidity. We 0:05:06.880 --> 0:05:10.039 saw some inclinations that liquidity was being dampened back in 0:05:10.080 --> 0:05:12.840 October off of our short term team at PIMCO, and 0:05:12.839 --> 0:05:15.160 that there was. As a result, we became more cautious 0:05:15.160 --> 0:05:19.440 from a liquidity standpoint when those opportunities pent became opportunistic. 0:05:19.640 --> 0:05:23.560 We found that the recalibration in in in UH spreads 0:05:23.600 --> 0:05:26.840 wider actually made things pretty good on a risk adjusta basis. 0:05:26.920 --> 0:05:28.960 To take some of our underweights back to market weight. 0:05:29.120 --> 0:05:31.320 In other words, some of the richness in the market 0:05:31.760 --> 0:05:34.280 was not really attractive. We sold it, and then as 0:05:34.279 --> 0:05:37.400 an active manager, we found an opportunity to actually redeploy 0:05:37.520 --> 0:05:40.520 capital when it became cheap, and more importantly, earned a 0:05:40.560 --> 0:05:44.200 liquidity premium or excess return by providing the market liquidity 0:05:44.279 --> 0:05:47.159 when traditional equity providers like banks, etcetera. Weren't there. So 0:05:47.240 --> 0:05:49.560 you've played it really well the last few months. You 0:05:49.640 --> 0:05:52.520 bought the weakness of December. I just wanted to what 0:05:52.600 --> 0:05:56.400 degree you're fighting the strength of the first quarter. So 0:05:56.800 --> 0:05:59.719 we've seen credits move in dramatic fashion. Some credits that 0:05:59.720 --> 0:06:02.520 were used at Libor plus forty six to nine months 0:06:02.560 --> 0:06:04.760 ago and blew out to Lybra plus three hundred or 0:06:04.760 --> 0:06:07.360 more are back to Libra plus fifty. So there's been 0:06:07.360 --> 0:06:09.520 almost a round trip. So you're right. This is a 0:06:09.680 --> 0:06:11.440 this is a market where you want to be trading 0:06:11.960 --> 0:06:14.760 simply focusing on credit work from the bottom up, and 0:06:14.760 --> 0:06:16.520 and that's what we're doing on our corporate credit side. 0:06:16.560 --> 0:06:19.159 So there is a there's basically a momentum now to 0:06:19.200 --> 0:06:22.080 take a breath. Let's find some opportunities to potentially be 0:06:22.160 --> 0:06:24.400 more defensive. At this point in time, in two thousand 0:06:24.360 --> 0:06:27.560 and six, people were screaming for seven basis points, ten 0:06:27.600 --> 0:06:30.400 basis points. Are we back to that silliness? No, where 0:06:30.680 --> 0:06:33.800 people are just trying to find the next ten basis points. 0:06:33.800 --> 0:06:36.360 It's hundreds of a percentage point. It's more fundamental now. 0:06:36.520 --> 0:06:39.200 So that was more structured, the leverage in the system. 0:06:39.200 --> 0:06:42.000 With structured now it's more fun fundamental, whereby we have 0:06:42.040 --> 0:06:44.320 to differgiate, you know, the you know, the cream from 0:06:44.200 --> 0:06:46.000 the from the rest of from the rest of the product. 0:06:46.040 --> 0:06:48.760 And as a result, that just takes homework, it takes resources. 0:06:48.800 --> 0:06:52.480 And as a result, having diversification, finding other opportunities to 0:06:52.520 --> 0:06:55.080 earn spread an income as we're doing, and mortgages some 0:06:55.160 --> 0:06:58.560 high quality ASSETBAC securities that actually offers your plenty of 0:06:58.680 --> 0:07:01.400 plenty of avenues. More importantly, I think this is The 0:07:01.400 --> 0:07:03.960 thing time is there's no real reason to reach for 0:07:04.040 --> 0:07:06.640 yield when you have portfolios they can yield between three 0:07:06.680 --> 0:07:08.919 and three and a half percent. Those are actually pretty 0:07:08.960 --> 0:07:11.720 attractive in a world where dividends are less than two percent, 0:07:11.960 --> 0:07:14.320 So why take the why take equity risk of this 0:07:15.240 --> 0:07:35.040 than you Pimco as well? What a joy for John 0:07:35.080 --> 0:07:39.200 Farrell and me always have Brian Weezer in our studios 0:07:39.200 --> 0:07:41.840 with pivotal research and of course brilliant work on the 0:07:41.880 --> 0:07:45.559 linkage of advertising into social We learned a few weeks 0:07:45.560 --> 0:07:48.840 ago that he had gone to the dark side and 0:07:49.000 --> 0:07:52.040 returned to the death start the sound side of group 0:07:52.080 --> 0:07:55.360 EM where he is helping them not so much with 0:07:55.520 --> 0:07:59.560 messaging but just thinking about what's out there. And the 0:07:59.600 --> 0:08:03.680 first we've got to talk about is absolute brilliance. Is 0:08:03.800 --> 0:08:07.640 there's not one office dog a group M is there? 0:08:07.640 --> 0:08:10.640 How many are there? Well? It? I work out of 0:08:10.640 --> 0:08:14.040 the Portland, Oregon office of one of our agency's mind 0:08:14.080 --> 0:08:18.240 Share and we have three dogs. You have a three 0:08:18.280 --> 0:08:23.200 dog night. We need a dog in the office. And 0:08:23.200 --> 0:08:25.480 and to be clear, the one of the dogs wears 0:08:25.520 --> 0:08:28.360 a bow tie every day. Tell me a code him 0:08:28.360 --> 0:08:32.360 Tom No, his name is Cooper. I think Cooper, that's cute, 0:08:32.920 --> 0:08:35.800 but it's a wonderful environment. It really is. What told 0:08:35.800 --> 0:08:37.360 you about how different it is. You used to cover 0:08:37.440 --> 0:08:40.040 these companies on the South Side, thinking about a sad 0:08:40.120 --> 0:08:42.800 on w p P. Once upon a time I had cells, 0:08:42.840 --> 0:08:46.920 I had buys. My position on the durability the agency 0:08:46.920 --> 0:08:49.760 model never changed, I'd argue, even when I had cell 0:08:49.840 --> 0:08:53.080 ratings from every agency holding company, I argued I was 0:08:53.120 --> 0:08:56.240 probably the greatest long term champion in the sense that 0:08:56.280 --> 0:08:59.000 I believed in the durability in a way that I 0:08:59.040 --> 0:09:02.599 think very few people, uh in equity markets. Yeah, I 0:09:03.240 --> 0:09:06.640 heard that. Yeah, absolutely. What's the state of advertising right now? 0:09:06.679 --> 0:09:09.400 I mean it is Facebook, Google, Facebook, Google, Facebook Google. 0:09:09.640 --> 0:09:11.560 Do you go in and try to change that with 0:09:11.640 --> 0:09:14.880 a new w PP after sir Martin or do you 0:09:15.000 --> 0:09:17.400 live with Facebook? No? Well, I mean here's the thing, 0:09:17.679 --> 0:09:21.400 And I think when I joined I worked for another 0:09:21.440 --> 0:09:24.200 holding company in two thousand three ten, and when I 0:09:24.200 --> 0:09:26.920 showed up, they're also coming from all street. The weird 0:09:27.000 --> 0:09:29.679 thing was how everyone's complaining about how dependent they were 0:09:29.720 --> 0:09:34.480 on television, how bad it was alternatives. Well, they have 0:09:34.520 --> 0:09:36.960 alternaate exactly. And but the point is that even then, 0:09:37.240 --> 0:09:40.240 every advertiser needed to think what was our alternative. In cinema, 0:09:40.280 --> 0:09:43.200 for example, was something that was held out as a hope, right, 0:09:43.240 --> 0:09:45.680 the idea that you could put it put money into 0:09:45.720 --> 0:09:48.000 on screen advertising in different minds, and then of course 0:09:48.000 --> 0:09:51.599 the Internet was considered an alternative. There will be alternatives, 0:09:51.640 --> 0:09:54.480 and every advertiser needs to think of the best alternative 0:09:54.480 --> 0:09:57.040 negotiating agreement. You have been more than anyone I know, 0:09:57.480 --> 0:10:00.720 strongest on you know what TV will survive, etcetera. Now 0:10:00.720 --> 0:10:04.160 there's certain networks I'll lead with NBC Universal who are 0:10:04.160 --> 0:10:08.000 saying we need fewer ads in prime time or on YouTube. 0:10:08.000 --> 0:10:10.280 They have those silly ads where you hit skip ad 0:10:10.360 --> 0:10:13.320 to get through them. That's a guy like you're it's 0:10:13.360 --> 0:10:16.680 your fault we're seeing those ads. What is the new Brian? 0:10:16.760 --> 0:10:20.640 We'ser attention span? Well for seconds? To be clear, I 0:10:20.679 --> 0:10:24.079 think that clutter is a significant problem, and I think 0:10:24.120 --> 0:10:26.959 that individuals. It doesn't take thought leadership to know that 0:10:27.080 --> 0:10:28.960 John and I are living in it it. I know, but 0:10:29.000 --> 0:10:32.480 I think every advertiser has to think differently about whether 0:10:32.559 --> 0:10:34.360 or not they should be allocating as much money to 0:10:35.040 --> 0:10:37.920 UH spots and dots for lack of a better do 0:10:38.040 --> 0:10:41.440 you guys know how much we hate that garbage cluttering 0:10:41.559 --> 0:10:44.160 up everything out there. We're not going to a commercial 0:10:44.200 --> 0:10:48.960 break now, are we, No, I know, But it's people. 0:10:49.200 --> 0:10:51.880 I mean, can you put any more ads on the uniforms? 0:10:51.920 --> 0:10:55.560 And English question? To what degree? Halfway established a bit 0:10:55.600 --> 0:10:58.760 of a false economy web by the advertiser believes that 0:10:58.800 --> 0:11:02.280 they're getting clicks and each month, but actually there's nothing 0:11:02.360 --> 0:11:05.760 behind it. So this is arguably the most important thing 0:11:06.000 --> 0:11:08.920 when it comes to UH. Well, a lot of the 0:11:08.920 --> 0:11:10.959 people I work with things that are focusing on it's 0:11:10.960 --> 0:11:14.760 things like viewability of ads. It's things like, uh, knowing 0:11:14.840 --> 0:11:19.280 that ads were actually engaging with consumers. Unfortunately, the industry 0:11:19.360 --> 0:11:21.480 is really good at optimizing trees and not so good 0:11:21.480 --> 0:11:24.800 at optimizing forests. And what that means practically is, uh, 0:11:24.840 --> 0:11:27.520 there's a focus on keeping costs down, which means quality 0:11:27.679 --> 0:11:30.760 may also go down. It's really hard to justify that. 0:11:30.840 --> 0:11:33.720 Let's just say you're you could own the entirety of 0:11:33.720 --> 0:11:35.600 an AD block, and maybe that AD block maybe a 0:11:35.600 --> 0:11:38.720 thirty second block instead of two minutes. That might be 0:11:38.840 --> 0:11:42.440 worth five times or ten times as much. The problem 0:11:42.480 --> 0:11:44.840 is so much of the industry so focused on the 0:11:45.040 --> 0:11:48.640 cost per unit, and that focus on the individual unit 0:11:48.720 --> 0:11:50.439 is a big problem and we actually need to in 0:11:50.760 --> 0:11:53.760 my position, we need to help marketers think more holistically 0:11:54.080 --> 0:11:56.360 about how to optimize the entirety of what they're doing. 0:11:56.960 --> 0:11:59.160 So it is a problem, and I think people might 0:11:59.559 --> 0:12:04.400 try to helps is that the Pluck go the distance 0:12:06.760 --> 0:12:10.679 radio is a really strong media. Actually all seriousness, I mean, 0:12:10.720 --> 0:12:14.600 but that's a good example that radio and audio more 0:12:14.679 --> 0:12:17.280 generally speaking, is something that more markers should be thinking of. 0:12:18.200 --> 0:12:20.839 Stop they're doing. I mean, we have been humbled by 0:12:20.880 --> 0:12:24.400 the success of our podcast. John and I collectively fell 0:12:24.440 --> 0:12:27.400 off the chair when Bank of America first stepped up 0:12:27.440 --> 0:12:30.920 and said we believe in your project, Apple Music, Spotify 0:12:31.000 --> 0:12:34.720 and the whole thing. There's Brian, They're doing audio because 0:12:34.720 --> 0:12:38.320 they don't have tough time to score around with video. Right. Well, 0:12:38.400 --> 0:12:41.600 I think that frankly that because television and video based 0:12:41.600 --> 0:12:44.960 advertising it is expensive at a per unit basis, is 0:12:45.000 --> 0:12:47.920 perceived as being cluttered. Uh. There is a real problem, 0:12:48.000 --> 0:12:50.160 especially with younger audiences, in terms of reaching them on 0:12:50.240 --> 0:12:53.240 traditional TV. You're reaching frequency curves look all out of whack. 0:12:53.559 --> 0:12:55.720 There's a lot of interest in audio. Now that said, 0:12:56.320 --> 0:12:59.200 I could have said that five years ago, and radio 0:12:59.440 --> 0:13:03.240 broadly to find early grew right. There's a shift to 0:13:03.280 --> 0:13:07.080 spend to Spotify, Pandora to some degree, but the industry 0:13:07.120 --> 0:13:11.960 collectively didn't grow. That said, we can still advocate for Hey, Marketer, 0:13:12.120 --> 0:13:14.240 if you haven't thought of this, you really need to. 0:13:15.080 --> 0:13:17.559 Let's talk about the privacy scandal that has engulfed the 0:13:17.600 --> 0:13:20.240 likes of Facebook over the last twelve months. How does 0:13:20.280 --> 0:13:25.560 that change things for your clients? So every brand has 0:13:25.600 --> 0:13:28.240 to think whether they think or not they should be 0:13:28.280 --> 0:13:31.360 thinking about whether or not risks and rewards of working 0:13:31.360 --> 0:13:33.920 with any given media partner are worthwhile, right, and so 0:13:34.120 --> 0:13:38.440 when it comes to the privacy issues with Facebook in particular, 0:13:38.520 --> 0:13:42.560 but more generally, we call it brand safety, right, Uh, 0:13:42.720 --> 0:13:45.240 brands have to consider if their association with a media 0:13:45.280 --> 0:13:48.840 owner has a negative effect or if because the consumer 0:13:48.960 --> 0:13:51.880 might see that they might be self loathing essentially, and 0:13:51.880 --> 0:13:54.480 while they're consuming the medium, think I hate being on 0:13:54.520 --> 0:13:57.120 this medium and I hate the advertisers who are making 0:13:57.120 --> 0:14:03.079 this possible. I mean, that can happen, but more generally, UH, 0:14:03.120 --> 0:14:07.600 there's a risk of association with media owners too, in 0:14:07.600 --> 0:14:09.600 the mind of people who are not using the platform right, 0:14:09.720 --> 0:14:15.200 enabling bad behaviors, enabling societal problems, whatever. Every brand has 0:14:15.240 --> 0:14:18.440 to consider whether or not their brand is tarnished by 0:14:18.480 --> 0:14:22.680 any association. The vast majority of advertisers have not considered 0:14:22.800 --> 0:14:25.480 that they are negatively impacted at this point in time. 0:14:26.000 --> 0:14:29.080 But that is the way to think about it. Again, 0:14:29.120 --> 0:14:31.200 even in my prior role, I still would have said 0:14:31.840 --> 0:14:35.280 Facebook and the lights are going to keep growing, because 0:14:35.280 --> 0:14:37.920 that's not a connection. That's let's think about private data 0:14:38.360 --> 0:14:42.160 as a currency. At some point, private datoris a currency 0:14:42.200 --> 0:14:44.400 is going to become so expensive that the advertiser and 0:14:44.440 --> 0:14:47.400 advertising incrementally is going to have to get more expensive. 0:14:47.440 --> 0:14:49.200 Let's say, yes, Okay, now I see what you're saying. 0:14:49.320 --> 0:14:51.760 So I would argue that every brand needs to make 0:14:51.760 --> 0:14:54.960 sure that if they can't persuade a consumer to supply 0:14:55.200 --> 0:14:58.600 willingly their data to the brand for use in targeting, 0:14:58.640 --> 0:15:01.640 they probably don't deserve it. And that goes back to 0:15:01.680 --> 0:15:03.080 the whole idea. What is a brand? Of brand is 0:15:03.080 --> 0:15:04.920 supposed to be a mark of trust. Of brand is 0:15:04.960 --> 0:15:07.080 supposed to be a mark of why a consumer should 0:15:07.080 --> 0:15:10.280 want to engage. You think, Feral thinks this is an interview. 0:15:10.800 --> 0:15:13.720 You're here to consult. What does radio need to do 0:15:13.760 --> 0:15:18.320 to stay red. It's fiber right now. I mean, come on, 0:15:18.800 --> 0:15:21.640 give me some thought leadership on how to drive Bloomberg 0:15:21.680 --> 0:15:24.440 surveillance forward over the next five years. Oh yeah, I know, 0:15:24.520 --> 0:15:27.160 i'd focus. Well, I mean, I'm going to bridge these 0:15:27.160 --> 0:15:30.800 two comments here because to the extent that if you 0:15:30.840 --> 0:15:36.360 can imagine a world where concerns around privacy are enhanced, 0:15:37.080 --> 0:15:40.800 where laws and regulations are more pronounced with respect how 0:15:40.880 --> 0:15:45.200 data is shared, and the relative advantage of using different 0:15:45.200 --> 0:15:50.160 media will evolve. Radio doesn't quite have that issue, at 0:15:50.240 --> 0:15:52.840 least traditional audio. Now that said, I think that every 0:15:52.840 --> 0:15:55.280 advertiser wants to find ways to use data that they have. 0:15:55.400 --> 0:15:59.080 I would argue an advertiser's best position to do gather 0:15:59.120 --> 0:16:02.000 a ton of data on the insights of consumers using 0:16:02.040 --> 0:16:04.760 a given media, use a ton of data on the 0:16:04.800 --> 0:16:07.200 output of an impact, and then let the art take 0:16:07.240 --> 0:16:09.360 the work in the middle. Yeah, we did all that 0:16:09.480 --> 0:16:12.160 and we come up with John's TV platform. The real yield. 0:16:12.360 --> 0:16:15.040 I mean, they saw a pure fixed income. That's all 0:16:15.360 --> 0:16:17.360 that one of the Brian Weezer. This has been painful. 0:16:17.440 --> 0:16:27.400 Thank you so much as well. Congratulations. Do you know 0:16:27.480 --> 0:16:34.800 what I'm you have no idea. Yeah, Field of Dreams, okay, 0:16:36.480 --> 0:16:42.200 Radiota out of the corn field. No, no idea, Okay, 0:16:42.520 --> 0:16:44.520 thank you, I mean it works. Brian Weezer, thank you 0:16:44.560 --> 0:16:48.080 so much. I'm sorry, but I think this. Yeah, that's 0:16:48.080 --> 0:16:51.560 a baseball movie. You have seen that movie. Okay, But anyways, 0:16:51.840 --> 0:16:58.600 just as really the association is is you just you 0:16:58.640 --> 0:17:01.360 work into it. It's like you don't it's heaven. No, 0:17:01.520 --> 0:17:04.800 it's Bloomberg surveillance. And we almost went with that phrase. 0:17:18.720 --> 0:17:21.720 I have been so so so waiting for this conversation. 0:17:21.840 --> 0:17:24.520 We're gonna go for you two hours NonStop. I'm kidding, 0:17:25.040 --> 0:17:27.440 of course, but we're gonna carve out on a Friday, 0:17:28.440 --> 0:17:31.680 clear language on what's all the rage. And it may 0:17:31.680 --> 0:17:35.120 be one you're in nanotechnology and this, that and the other, 0:17:35.960 --> 0:17:39.639 but once again it is the use and abuse of 0:17:39.680 --> 0:17:43.760 the phrase artificial intelligence. All of this goes back to 0:17:43.840 --> 0:17:48.760 a meeting in nineteen at Dartmouth College where the the 0:17:48.840 --> 0:17:52.840 brains of Pittsburgh showed up from the University of Pittsburgh 0:17:52.880 --> 0:17:56.560 in a small school called Carnegie Mellon University and Allen 0:17:56.640 --> 0:18:00.119 Newell and her Herbert Simon and other giants. Actually, he 0:18:00.200 --> 0:18:04.040 tried to figure out how we would handle computers and machines. 0:18:04.119 --> 0:18:06.840 Flash forward and out of Pittsburgh at n y U 0:18:06.880 --> 0:18:09.760 as vacant dar and he joins us this morning. We 0:18:09.800 --> 0:18:12.360 are honored to have you here, professor. There's so much 0:18:12.480 --> 0:18:17.120 to straighten out here. What is most wrong in the 0:18:17.160 --> 0:18:21.480 new vogue of artificial intelligence is it's covered by the media. 0:18:21.960 --> 0:18:27.800 What drives you nuts the most? Um? Really happy to 0:18:27.840 --> 0:18:30.679 be here, Tom, And incidentally, Hope Simon was one of 0:18:30.680 --> 0:18:33.720 my mentors and graduate school and on my thesis committee, 0:18:33.760 --> 0:18:36.920 so I was honored to have him um and get 0:18:36.960 --> 0:18:40.960 to know him. I think AI has seen several hype 0:18:40.960 --> 0:18:44.679 cycles lost four decades full disclosure, my brother was in 0:18:44.720 --> 0:18:47.119 one of them. Continued, So you know we've so so 0:18:47.400 --> 0:18:50.000 I've seen several of these hype cycles during my career, 0:18:50.200 --> 0:18:53.119 and h we're seeing another one at the moment. What 0:18:53.160 --> 0:18:56.160 do we get wrong? What's the hype that drives you nuts? Well, 0:18:56.200 --> 0:18:58.680 you know the by the way, I think this one 0:18:58.760 --> 0:19:00.800 is different, right, So as just want to qualify that. 0:19:00.880 --> 0:19:04.359 But I think the there's a lot of misunderstanding that 0:19:04.440 --> 0:19:09.560 AI will solve everything that machines have, you know, will 0:19:09.760 --> 0:19:13.679 have become incredibly intelligent and they'll replace us. And you 0:19:13.720 --> 0:19:16.359 know that may well happen in the very long term, 0:19:16.480 --> 0:19:20.080 but I think that we ought to have some somewhat 0:19:20.200 --> 0:19:23.160 muted expectations. I mean, we've certainly come a long way, 0:19:23.800 --> 0:19:26.760 but I think we have a really long way to 0:19:26.800 --> 0:19:29.720 go go. Jack Clark, writing for Bloomberg four years ago, says, 0:19:29.760 --> 0:19:32.560 this time is different. You just alluded to that what's 0:19:32.600 --> 0:19:38.119 different now within artificial intelligence or these machines running our lives. 0:19:38.520 --> 0:19:43.920 So I think what's what's different now is that we've 0:19:44.000 --> 0:19:49.000 made a dent in solving perception kinds of problems. So 0:19:49.160 --> 0:19:53.840 machines have gotten a lot better at seeing, hearing, getting 0:19:53.840 --> 0:19:57.200 better at reading um And I think what that does 0:19:57.560 --> 0:20:01.720 is it changes the nature of interaction with the machine. 0:20:01.760 --> 0:20:06.000 The machine can now ingest inputs directly from the environment 0:20:06.119 --> 0:20:09.120 instead of them being formatted in a way and structured 0:20:09.119 --> 0:20:10.720 in the way that it can understand. Right, So it's 0:20:10.720 --> 0:20:13.560 able to deal with more instructured kind of input like 0:20:13.640 --> 0:20:17.480 humans are. And and that's interesting in many domains, including healthcare. 0:20:17.560 --> 0:20:19.520 But you can look at images and stuff like that. 0:20:19.560 --> 0:20:21.520 But in your high school math in India, I'm sure 0:20:21.520 --> 0:20:23.840 you got to this quickly. Linear regressions and you've got 0:20:23.920 --> 0:20:25.880 a signal on the back end, which is that risk 0:20:26.040 --> 0:20:29.280 that's out there, that unknowable that's out there. We're all 0:20:29.320 --> 0:20:31.600 living this right now with the emotion of two plane 0:20:31.600 --> 0:20:34.920 crashes over software in a cockpit that just flat out 0:20:35.160 --> 0:20:38.479 didn't work. That's certainly what the FAA is getting too quickly. 0:20:38.960 --> 0:20:42.280 Is there a new reliance on machines and on computer 0:20:42.400 --> 0:20:47.880 technology where that risk, that sigma it becomes too unknowable. Well, 0:20:47.920 --> 0:20:51.919 you know, you're you're raising an incredibly important question. And 0:20:52.160 --> 0:20:54.040 you know, one of my questions really is when do 0:20:54.040 --> 0:20:56.280 we trust machines? And the way I look at trust 0:20:56.359 --> 0:20:59.280 is in terms of risk. In my mind, it boils 0:20:59.280 --> 0:21:01.399 down to how often will the machine be wrong? And 0:21:01.440 --> 0:21:04.800 what are the consequences when the machine is wrong? Things 0:21:04.840 --> 0:21:08.800 like drive a less cars, autopilot. The consequences of era 0:21:08.920 --> 0:21:13.240 are severe. Uh. And so those kinds of decisions tend 0:21:13.280 --> 0:21:16.200 to be particularly sensitive. When they talk about handing control. 0:21:16.960 --> 0:21:19.240 You're in class throwing chalk at somebody at n WANT. 0:21:19.280 --> 0:21:22.040 You're the dumbest kid in engineering n y U. And 0:21:22.080 --> 0:21:25.159 if you look at the risk distribution, we all understand 0:21:25.200 --> 0:21:28.000 it's not gaussy and it's not a normal bell curve 0:21:28.400 --> 0:21:30.639 like the height of our high school class. It can 0:21:30.680 --> 0:21:34.080 be a goofy distribution or whatever, which is like if 0:21:34.119 --> 0:21:37.719 I paint ten thou BMW's one of them will be wrong. 0:21:38.240 --> 0:21:42.600 That's a lot different than airplanes crash exactly. What's what's 0:21:42.680 --> 0:21:45.200 different about airplanes and driver a less cars is you're 0:21:45.240 --> 0:21:48.600 not really interested in their distribution. You're not exactly gonna 0:21:48.640 --> 0:21:50.919 really interested in the average case. You're interested in the 0:21:50.960 --> 0:21:56.239 worst case. Right, it's that worst case Rumsfeldian unknown that 0:21:56.280 --> 0:21:58.919 you're really concerned about. That you may not have counter 0:22:00.040 --> 0:22:02.440 guy like you with the privileges studying with the CMU 0:22:02.520 --> 0:22:05.560 guys at Pittsburgh a million years ago. If if you 0:22:05.680 --> 0:22:08.240 look at these tech boys out and still these children 0:22:08.280 --> 0:22:11.520 out in Silicon Valley saying we're gonna do driverless cars, 0:22:12.000 --> 0:22:14.320 you've got in your head all those risks out there, 0:22:14.640 --> 0:22:18.760 When do we actually get driverless cars? Given how each 0:22:18.880 --> 0:22:24.200 of us perceives the next traffic accigen accident very slowly, 0:22:25.280 --> 0:22:27.680 is my prediction. I don't think this is going to 0:22:27.800 --> 0:22:30.760