WEBVTT - Could Robots Take Away This Classic Wall Street Job? 0:00:01.000 --> 0:00:07.560 Stuff. This is the sound of a busy trading floor 0:00:07.640 --> 0:00:11.520 in Wells Fargo skyscraper in New York City. It's filled 0:00:11.520 --> 0:00:16.239 with stockbrokers who are researching companies, writing reports, chatting with investors, 0:00:16.239 --> 0:00:20.040 and processing their trades. These are classic and highly paid 0:00:20.120 --> 0:00:22.639 jobs that have been around for more than a century. 0:00:24.440 --> 0:00:36.599 Imagine if one day those jokes went away. Yeah, this 0:00:36.720 --> 0:00:40.160 is a scenario that Ken Center, a veteran Internet analyst, 0:00:40.360 --> 0:00:43.840 one of those classic brokers, is forcing himself to think about. 0:00:45.240 --> 0:00:49.600 Canada developer sidekick named Brian Healy have spent months building 0:00:49.640 --> 0:00:53.320 an artificially intelligent robo analysts that performs a lot of 0:00:53.360 --> 0:00:56.640 what Ken currently does for a living. Ken calls the 0:00:56.640 --> 0:01:01.000 system ERA, which stands for Artificially Intelligent Equity Research Analyst. 0:01:01.360 --> 0:01:04.600 It's an elaborate warning to investments, firms and banks that 0:01:04.640 --> 0:01:07.039 they should learn and embrace AI to catch up with 0:01:07.120 --> 0:01:10.959 Internet giants like Google, even if it means automation destroys 0:01:11.000 --> 0:01:13.760 some lucrative all street jobs. And it's a broader wake 0:01:13.840 --> 0:01:17.240 up call for all of us. Artificial intelligence is here 0:01:17.480 --> 0:01:21.000 and it's changing complex, high paying jobs for good. We 0:01:21.200 --> 0:01:35.640 better be ready. M Hi, I'm brad Stone, I'm Julie 0:01:35.720 --> 0:01:39.040 Verge and I'm Alice de Ba, and this week Undercrypted, 0:01:39.240 --> 0:01:42.160 we're taking you inside Ken's quest to build an ever 0:01:42.280 --> 0:01:46.720 evolving and always learning software program. This program could eventually 0:01:46.840 --> 0:01:49.280 put Ken out of a job, or at least do 0:01:49.400 --> 0:01:52.720 several parts of his job a lot better than him. 0:01:52.760 --> 0:02:01.040 Stay with us, So, Alian, Julie, you guys recently went 0:02:01.120 --> 0:02:03.880 to go meet the brains behind this project. We did. 0:02:04.160 --> 0:02:06.400 It was a cool October morning. We picked up an 0:02:06.440 --> 0:02:09.320 audio recorder and headed down a few blocks to Wells 0:02:09.320 --> 0:02:16.120 Fargo's office on Park Avenue. And Ali, You've known Ken 0:02:16.120 --> 0:02:18.760 the stock analysts for a while, Yeah, all the way 0:02:18.760 --> 0:02:20.760 back to the days when I was a reporter voters 0:02:20.800 --> 0:02:22.919 in the Wall Street Journal. I'd often call him for 0:02:22.960 --> 0:02:26.200 advice on tech companies and news. He's tall, with wavy, 0:02:26.240 --> 0:02:28.760 sweep back hair, and when we met him, he's wearing 0:02:28.760 --> 0:02:31.240 a crisp suit and tied like he was born in 0:02:31.240 --> 0:02:34.160 the outfit. Brian, on the other hand, comes from a 0:02:34.200 --> 0:02:37.720 tech background. He helped build Alexa, the digital at home 0:02:37.760 --> 0:02:41.560 assistant whose voice you hear on Amazon's Echo speaker. Brian 0:02:41.600 --> 0:02:45.919 has a goatee and no nonsense, close cropped hairstyle. And 0:02:45.960 --> 0:02:48.080 I have to admit he didn't look so comfortable in 0:02:48.120 --> 0:02:51.040 that suit that he was wearing. It sounds like an 0:02:51.080 --> 0:02:53.480 odd couple. How did they meet? It actually kind of 0:02:53.480 --> 0:02:56.440 started out by a chance. Ken was organizing an AI 0:02:56.520 --> 0:03:00.200 conference in two and had reached out to Brian's US 0:03:00.280 --> 0:03:02.920 to be a speaker. He was busy, so he recommended 0:03:02.919 --> 0:03:05.400 Brian go in his place. They hit it off, and 0:03:05.480 --> 0:03:08.600 soon after the conference they started working together on other projects. 0:03:08.960 --> 0:03:10.560 You know, I would come up on the weekends and 0:03:10.600 --> 0:03:13.720 stuff and um talk about the technology and just sort 0:03:13.720 --> 0:03:17.840 of be a resources available when he's talking to clients. UM. 0:03:17.880 --> 0:03:19.520 And then it just kind of kept evolving from you know, 0:03:19.560 --> 0:03:22.760 we kept working together. While spending all this time researching 0:03:22.760 --> 0:03:25.639 how Internet giants were using AI to create better products, 0:03:25.960 --> 0:03:29.960 they came upon a scary but important question. My goal 0:03:30.240 --> 0:03:32.800 was to try and get Bryan to help maybe just 0:03:33.280 --> 0:03:35.560 sort of deepen my understanding so that I could help 0:03:35.920 --> 0:03:39.240 our clients to understand this. And I guess we sort 0:03:39.240 --> 0:03:41.120 of put it out there is almost a question kind 0:03:41.120 --> 0:03:43.880 of like what what can't be automated? At this point, 0:03:44.720 --> 0:03:47.560 that's when they agreed to create a software version of 0:03:47.640 --> 0:04:01.160 Ken but better era. This is to Brian built reads 0:04:01.200 --> 0:04:04.040 news stories on companies and distills that into a sentiment 0:04:04.120 --> 0:04:07.200 score from one to zero, with one being wonderful and 0:04:07.320 --> 0:04:12.000 zero for absolutely terrible. Then she'd monitor the stock market 0:04:12.040 --> 0:04:15.000 to see if these positive or negative articles would move 0:04:15.080 --> 0:04:19.320 share prices. If she spotted a correlation, she'd remember that 0:04:19.440 --> 0:04:23.119 and use it to make predictions in the future, including buy, sell, 0:04:23.240 --> 0:04:27.279 and hold ratings. Finally, or would sure not short rand 0:04:27.360 --> 0:04:32.359 summaries explaining the predictions she was making. The project started 0:04:32.360 --> 0:04:35.680 out as an experiment not for public consumption. Could Brian 0:04:35.839 --> 0:04:38.719 use machine learning, a hot type of AI to make 0:04:38.760 --> 0:04:43.119 a robot analyst that was actually useful? And Brian said, well, 0:04:43.279 --> 0:04:45.240 you have to explain to me what you do. So 0:04:45.600 --> 0:04:47.680 I said, okay, well let's let's see if we can 0:04:47.800 --> 0:04:50.800 we can do this. Um, I'll take you through what 0:04:50.920 --> 0:04:53.800 I do as an equity research journealist, and you know 0:04:53.839 --> 0:04:56.760 to the best you know the best I can, and 0:04:56.960 --> 0:04:59.160 you take me through what would be sort of the 0:04:59.160 --> 0:05:03.200 tools that would could be used to replace or enhance 0:05:03.240 --> 0:05:06.960 what I do. He used techniques like natural language processing 0:05:07.000 --> 0:05:09.520 to build a system this summer in his spare time. 0:05:09.920 --> 0:05:12.279 Because we have to remember, Brian has a day job 0:05:12.400 --> 0:05:15.240 as head of AI at a company called Lola. You know, 0:05:15.360 --> 0:05:17.400 this would probably be a good time to define our 0:05:17.520 --> 0:05:22.359 terms for the uninitiated. What is artificial intelligence? He's Brian 0:05:22.440 --> 0:05:26.840 with an explanation. The most basic answer, Machine learning is 0:05:27.279 --> 0:05:31.880 any engineering technique that means the software is not discriminately programmed. 0:05:32.560 --> 0:05:36.400 So I didn't write code that had specific discrete branders 0:05:36.400 --> 0:05:40.120 and comportent things. It was a system of statistically learning. 0:05:40.560 --> 0:05:43.479 In general, the industry term for machine learning means learning 0:05:43.560 --> 0:05:47.440 specifically from large volumes of data. It gets very complicated. 0:05:47.480 --> 0:05:49.960 Beyond this, we won't take you down the rabbit hole, 0:05:50.160 --> 0:05:52.440 but if you only take away one thing from this 0:05:52.480 --> 0:05:57.719 podcast is this these computer programs update themselves without humans 0:05:57.720 --> 0:05:59.719 having to do very much work. You know a model 0:05:59.720 --> 0:06:01.840 that's it takes data and passes it to a data 0:06:01.960 --> 0:06:04.520 analytics team, and then they pass it to a product 0:06:04.520 --> 0:06:06.680 development team and their product development teams who works with 0:06:06.720 --> 0:06:10.520 the engineers to bring it back into the model. Um, 0:06:10.560 --> 0:06:14.360 you're you're allowing your essentially seeing where data can drive 0:06:14.400 --> 0:06:17.640 the changes within a model's performance. What you're talking about 0:06:17.839 --> 0:06:20.760 is removing certain bottlenecks in terms of how these companies 0:06:20.760 --> 0:06:24.600 innovative ERA makes stock calls, and it also tracks how 0:06:24.720 --> 0:06:28.320 these recommendations end up panning out. If their prediction turns 0:06:28.320 --> 0:06:30.480 out to be wrong, the system remembers it and it's 0:06:30.600 --> 0:06:41.040 less likely to make that decision again in the future. Okay, 0:06:41.120 --> 0:06:43.360 so let's get back to the story. So here Ken 0:06:43.360 --> 0:06:47.200 and Brian trying to build this robo analyst. Initially, eras 0:06:47.240 --> 0:06:50.320 summaries were tough to eat, they were disjointed, and the 0:06:50.400 --> 0:06:52.599 grammar was off. You could really tell it was written 0:06:52.640 --> 0:06:55.440 by a computer. But after a couple of months her 0:06:55.480 --> 0:06:59.440 reports started making sense. I think I sort of realized 0:06:59.440 --> 0:07:02.440 it could be a thing when it started actually producing language. 0:07:02.839 --> 0:07:05.280 So when that got generated, it was kind of like, Okay, 0:07:05.279 --> 0:07:07.600 this is actually kind of meat like in saying what 0:07:07.640 --> 0:07:09.960 it's doing. It sort of explaining itself, and it just 0:07:10.000 --> 0:07:13.440 sort of highlighted that as we keep consuming more data 0:07:13.640 --> 0:07:15.680 over time, these are only going to get better, and 0:07:15.720 --> 0:07:18.040 so we should just keep doing it. So this is 0:07:18.080 --> 0:07:21.240 when the project starts feeling real. Yeah, and Ken decided 0:07:21.280 --> 0:07:24.040 it was time to share ERA with the outside world, 0:07:24.480 --> 0:07:26.760 but first he needed to get permission from his company, 0:07:26.800 --> 0:07:29.520 which also happens to be one of the biggest banks 0:07:29.520 --> 0:07:32.800 in America Wells Fargo. This must have been a pretty 0:07:32.800 --> 0:07:35.680 big risk for Ken. Wells Fargo is a pretty traditional 0:07:35.760 --> 0:07:38.480 bank and it's come under a lot of regulatory scrutiny 0:07:38.480 --> 0:07:41.520 over the past few years for creating accounts that weren't 0:07:41.560 --> 0:07:44.840 opened by actual customers, and the bureaucracy that comes with 0:07:44.960 --> 0:07:47.960 doing anything new and a large company is significant, and 0:07:48.080 --> 0:07:51.800 bread on top of that, Ken had just joined the company. 0:07:51.920 --> 0:07:54.440 When he told the compliance department about what he was 0:07:54.560 --> 0:07:57.640 up to, they weren't very pleased. There was sort of 0:07:57.840 --> 0:08:01.000 a disbelief at first that we we actually wanted to 0:08:01.000 --> 0:08:04.720 put an analyst out there that we weekly was artificially intelligent, 0:08:05.280 --> 0:08:08.400 and that we wanted to provide predictions around the stocks, 0:08:08.400 --> 0:08:11.440 and that actually, you know, could write its own research, 0:08:11.720 --> 0:08:14.239 and not only that, but it could bold the sections 0:08:14.280 --> 0:08:17.040 of text that it felt really it wanted to underscores 0:08:17.080 --> 0:08:21.280 being important to its specific stock thesis. Right, So that 0:08:21.400 --> 0:08:24.280 was sort of I think people just sort of kind 0:08:24.280 --> 0:08:26.080 of through their head back and we're in a bit 0:08:26.120 --> 0:08:33.760 of disbelief. Why were they so shocked? So equity research 0:08:33.800 --> 0:08:36.640 analysts spend all day thinking about the future of the 0:08:36.679 --> 0:08:39.840 companies they cover and how that will impact stock prices. 0:08:40.320 --> 0:08:44.040 They send these reports to clients, making different recommendations on 0:08:44.080 --> 0:08:47.720 whether to buy, sell, or hold, along with a price target, 0:08:48.080 --> 0:08:50.160 and that's usually an estimate of where they think the 0:08:50.200 --> 0:08:52.480 stock is going to be in about twelve months. And 0:08:52.520 --> 0:08:55.320 if they're wrong. If they're wrong, investors can get mad 0:08:55.360 --> 0:08:58.400 and go somewhere else for advice and trading services, and 0:08:58.440 --> 0:09:00.600 these are things clients pay a lot lot of money 0:09:00.640 --> 0:09:04.400 for the analyst whole reputation, and by extension, the bank's 0:09:04.440 --> 0:09:08.200 reputation rests on the accuracy of their stock ratings, so 0:09:08.280 --> 0:09:17.840 there was a lot that could go wrong. Okay, so 0:09:17.960 --> 0:09:21.360 so far, the analyst Ken has formed an unlikely friendship 0:09:21.400 --> 0:09:24.800 with the developer Brian. They built this bot to recommend 0:09:24.840 --> 0:09:28.240 when to buy and sell certain companies. Their ambition shocks 0:09:28.280 --> 0:09:33.360 Wells Fargo's compliance department. What happens next well Can eventually 0:09:33.400 --> 0:09:36.840 succeeds in convincing the compliance department to let him proceed, 0:09:37.280 --> 0:09:39.680 and it comes time to unveil Era to the public. 0:09:40.040 --> 0:09:44.920 This happens on September two thousand seventeen. She started making 0:09:44.920 --> 0:09:49.360 stock recommendations using this complex new software that only a 0:09:49.400 --> 0:09:52.240 few people on Wall Street even understood. What was the 0:09:52.320 --> 0:09:57.800 response like, is your AI technological work with Era? Is 0:09:57.800 --> 0:10:01.599 it gonna put securities research a business? No, and we 0:10:01.920 --> 0:10:04.760 did it more as a study. I think for clients 0:10:04.760 --> 0:10:07.559 who tend to be a little longer term in duration, 0:10:08.000 --> 0:10:11.280 they're interested in how do you build it right? What 0:10:11.400 --> 0:10:14.599 is you know? How do you think about the application 0:10:15.240 --> 0:10:18.240 of this technology? For clients were a little bit more 0:10:18.280 --> 0:10:21.280 short term and focus, you know what are Era's predictions 0:10:21.280 --> 0:10:23.920 this week? Right? Can EARRA help me manage news flow 0:10:24.559 --> 0:10:27.680 and help me synthesize? What can means here is that 0:10:27.760 --> 0:10:30.760 the summaries error rights can help clients pick through the 0:10:30.840 --> 0:10:34.720 daily avalanche of online news to find real developments that 0:10:34.760 --> 0:10:37.920 will actually move the stock. We recently asked Brian to 0:10:37.960 --> 0:10:42.600 hook up ERA to an Amazon Echo speaker. Alexa. Ask 0:10:42.760 --> 0:10:47.199 my AI analyst what she thinks of Google as of yesterday. 0:10:47.440 --> 0:10:51.640 I think Google looks like a hold and that forecast 0:10:51.679 --> 0:10:57.160 is good until October. Here's why from Nashable dot com. 0:10:57.240 --> 0:11:00.360 Alphabet just took an important step towards becoming a major 0:11:00.400 --> 0:11:03.880 force in the online payments world. As tech Crunch reports, 0:11:04.600 --> 0:11:06.920 Let's say they mark AI is an important topic to 0:11:06.960 --> 0:11:10.000 cover for a number of names. Well, they're you know inbox? 0:11:10.080 --> 0:11:12.760 Could you know swell with all these AI articles, but 0:11:12.840 --> 0:11:14.520 they may or may not actually be relevant to the 0:11:14.520 --> 0:11:22.600 stock's performance. So this program launches in September and it 0:11:22.679 --> 0:11:26.320 starts making all kinds of calls. The first week went fine. 0:11:26.760 --> 0:11:29.960 The second week the Era through a curve bool downgrading 0:11:30.000 --> 0:11:33.120 shares of Facebook, recommending the investor to sell the shares 0:11:33.160 --> 0:11:35.480 of the company. And now we should add here that 0:11:35.520 --> 0:11:38.760 as of this taping, Facebook stock is of about fifty 0:11:39.280 --> 0:11:41.960 this year alone, and forty six analysts that we have 0:11:42.040 --> 0:11:45.040 listed on the Bloomberg terminal, only two of them have 0:11:45.160 --> 0:11:48.920 cell ratings on the stock. So Era made a contrarian call. 0:11:49.559 --> 0:11:53.000 And the funny thing is Era's downgrade contradicted Ken's own 0:11:53.000 --> 0:11:55.160 opinion because he has a by rating on the stock. 0:11:55.679 --> 0:11:58.800 Turns out Era had read thousands of stories about Russian 0:11:58.840 --> 0:12:01.800 ads on Facebook that were designed to divide US voters 0:12:02.000 --> 0:12:05.280 out of last year's US presidential election. Russia probe is 0:12:05.280 --> 0:12:09.400 focused on people, including three former Trump aids, charged with crimes. 0:12:09.400 --> 0:12:14.320 Another phase focuses on corporations like Twitter, Google, and especially Facebook. Today, 0:12:14.320 --> 0:12:16.920 politicians who are up in arms. Congress had called for 0:12:17.000 --> 0:12:20.679 hearings all these negative articles drowned out other positive pieces. 0:12:20.920 --> 0:12:23.319 Why is it so important that we all see these 0:12:23.360 --> 0:12:26.080 ads right now? Well, when you look at them, they 0:12:26.080 --> 0:12:30.160 are I have seen them, and this means eras AI 0:12:30.240 --> 0:12:33.200 algorithms picked up on this in advice selling to avoid 0:12:33.240 --> 0:12:35.480 a drop in the share price, which she had predicted 0:12:35.640 --> 0:12:39.240 over the next week, and what ended up happening, Erra 0:12:39.400 --> 0:12:42.280 was wrong. Facebook start ended up rising a bit over 0:12:42.280 --> 0:12:46.240 the next week. Some clients weren't happy. A few saw 0:12:46.280 --> 0:12:48.480 it as proof that what Ken and Brian were doing 0:12:48.520 --> 0:12:52.080 didn't matter. For others, it confirmed their view that the 0:12:52.120 --> 0:12:57.640 way Ken was using AI was fundamentally flawed. That was 0:12:57.679 --> 0:13:00.160 a disappointment because I think that you know, people are 0:13:00.000 --> 0:13:02.760 are their zero in on things that they're kind of 0:13:02.760 --> 0:13:06.280 missing the bigger picture here. To Richard Johnson, a vice 0:13:06.320 --> 0:13:09.960 president at research firm Greenwich Associates, it was a sign 0:13:10.000 --> 0:13:13.120 of how far AI and finance still has to go. 0:13:13.960 --> 0:13:18.160 The robot downgrading Facebook, I think that's probably where this 0:13:18.200 --> 0:13:20.040 type of analysis is going to struggle right now. I 0:13:20.040 --> 0:13:21.680 think we're very much in the early days of it. 0:13:22.200 --> 0:13:25.280 But you know, in that example, perhaps you know the 0:13:25.640 --> 0:13:28.560 album Nettle bit of fine tuning and to kind of 0:13:29.280 --> 0:13:33.280 you know, give less weight to the Russian fake news 0:13:33.360 --> 0:13:35.560 type stories that are in the media because I think, 0:13:36.320 --> 0:13:38.880 you know, we all kind of thing that your Facebook 0:13:38.880 --> 0:13:43.160 will survive, will will survive this. When Ken first explained 0:13:43.280 --> 0:13:46.280 error to Wells Fargo's compliance team, he said the only 0:13:46.280 --> 0:13:50.480 relevant rule they could find was recent guidance on robo advisors. Now, 0:13:50.520 --> 0:13:54.040 these programs automatically decide how much of your investment portfolio 0:13:54.040 --> 0:13:57.560 should be in stocks, bonds, and other assets. One recommendation 0:13:57.600 --> 0:14:00.520 that they had was to tell clients about quote changes 0:14:00.559 --> 0:14:04.840 to algorithmic code that may materially affect their portfolios. But 0:14:04.920 --> 0:14:07.959 this doesn't apply to era's self learning approach. What I 0:14:08.000 --> 0:14:10.320 think it's very interesting about this and why there's such 0:14:10.360 --> 0:14:13.160 a great learning opportunity as that robo advisors tend to 0:14:13.520 --> 0:14:16.920 tend to be programmed. What can means to say here 0:14:17.000 --> 0:14:21.040 is this, robo advisors are mostly programmed in the old 0:14:21.080 --> 0:14:24.720 fashioned way humans right, software code that gives step by 0:14:24.760 --> 0:14:28.280 step explicit constructions on what to do in certain situations. 0:14:28.840 --> 0:14:32.200 With a product like ERA, you don't program her, right, 0:14:32.240 --> 0:14:34.880 the data programs her so her, so her algorithms are 0:14:34.920 --> 0:14:37.720 constantly changing as a result of that, and so we 0:14:37.880 --> 0:14:39.800 to some extent we moved out, I think a little 0:14:39.800 --> 0:14:42.240 further than what you know, finn Ra and kind of 0:14:42.280 --> 0:14:45.800 these agencies that regulate, you know, our industry are are 0:14:45.920 --> 0:14:49.920 used to. The solution to this was to include a 0:14:49.960 --> 0:14:53.960 bold faced disclaimer and every report stressing the era's stock 0:14:54.080 --> 0:14:57.760 ratings were not investment advice and should only be read 0:14:57.800 --> 0:15:01.560 to gain a greater understanding of artificial intelligence. And when 0:15:01.600 --> 0:15:03.960 we met Ken in New York, we mentioned to him 0:15:04.000 --> 0:15:06.520 that he'd been putting the disclaimer near the top of 0:15:06.520 --> 0:15:10.040 each note he publishes about ever, maybe I did just 0:15:10.080 --> 0:15:12.120 to kind of, you know, continue to get it as 0:15:12.160 --> 0:15:13.920 soon as I break from in terms of what I'm 0:15:13.960 --> 0:15:17.040 writing about to what when Aaron starts writing, I'm just 0:15:17.120 --> 0:15:20.120 hitting with the disclaimer. I don't want to control what 0:15:20.200 --> 0:15:22.560 people trade on one way or the other. If they 0:15:22.600 --> 0:15:25.160 take her advice and they feel that, you know, they 0:15:25.160 --> 0:15:28.280 agree with her sun reads and they like her points great, 0:15:28.520 --> 0:15:31.000 you know, I don't want to tell them, don't you know, 0:15:31.200 --> 0:15:38.640 don't take an information that she has. So where's Eric today? 0:15:39.040 --> 0:15:42.000 She's now covering more than five hundred stocks and reading 0:15:42.040 --> 0:15:44.880 about half a million news stories a day. That's half 0:15:44.880 --> 0:15:47.560 a million a day, a few weeks ago, the system 0:15:47.600 --> 0:15:50.520 also analyzed tech company earnings for the first time and 0:15:50.560 --> 0:15:54.600 issued nine recommendations on stocks like Amazon, Google, and Netflix. 0:15:55.000 --> 0:15:57.760 And this time did I get it right? This time? 0:15:57.760 --> 0:16:00.320 They all proved to be accurate after a week. Well, 0:16:00.360 --> 0:16:02.640 recently these stocks have performed so well my dog can 0:16:02.680 --> 0:16:06.080 probably recommend it. But seriously, it does beg the question. 0:16:06.360 --> 0:16:10.160 Has Ken succeeded in creating his own autonomous replacement? Here's 0:16:10.160 --> 0:16:12.040 what he had to say about that. But I think 0:16:12.080 --> 0:16:13.960 in the end it's sort of evolved over time. Is 0:16:14.000 --> 0:16:17.720 one more of an enhancement. It certainly took took out 0:16:17.800 --> 0:16:21.080 maybe or it showed how a certain amount of the 0:16:21.080 --> 0:16:23.360 work that we do in terms of handling news volume 0:16:24.000 --> 0:16:26.800 and and trying to sort of be specific in terms 0:16:26.840 --> 0:16:30.280 of waitings of that news volume in terms of stock prediction, 0:16:30.400 --> 0:16:34.160 how it could be I think improved. Other analysts agree 0:16:34.200 --> 0:16:36.840 on the point that cutting the workload is a great idea, 0:16:37.360 --> 0:16:39.880 but they're way less keen to say that an AI 0:16:39.920 --> 0:16:43.680 analyst can help them make stock recommendations alongside their own calls. 0:16:44.320 --> 0:16:48.280 As far as having an air bra our official intelligence 0:16:48.320 --> 0:16:54.320 Bank research analyst, I don't think that's happening in the 0:16:54.320 --> 0:16:57.720 the next decade um, but I'll often watch my back. 0:16:58.280 --> 0:17:01.120 That's Mike Mayo, a veteran bank analyst at Wells Fargo, 0:17:01.320 --> 0:17:04.680 who came in to chat with us about Kennon Bryan's creation. Now. 0:17:04.720 --> 0:17:06.800 I was super excited to see Mike because I remember 0:17:06.880 --> 0:17:10.040 my first day as a reporter on the financial market. 0:17:10.119 --> 0:17:12.240 Mike was one of my first calls. He actually picked 0:17:12.320 --> 0:17:14.240 up the phone, and I feel like I'm some super 0:17:14.280 --> 0:17:17.480 cool twenty two year old talking to this big guy analyst. 0:17:17.560 --> 0:17:20.200 He's a sort of a finance nerd, but you can 0:17:20.320 --> 0:17:23.680 tell instantly that he's so passionate about what he does. 0:17:24.400 --> 0:17:27.239 For him. ARA is all about enriching his role and 0:17:27.280 --> 0:17:30.040 making banks more efficient. I don't want to spend all 0:17:30.040 --> 0:17:34.280 this time looking at data feeds and checking articles and 0:17:34.800 --> 0:17:37.119 you know, checking barons over the weekend and did I 0:17:37.160 --> 0:17:40.199 catch the Wall Street Journal story? And a lot of 0:17:40.240 --> 0:17:43.760 time for looking at all this information just to see 0:17:44.520 --> 0:17:47.879 if we're missing anything. Now, Mike said that is associate 0:17:47.880 --> 0:17:52.040 analyst probably spends about seventy of his time checking news articles, 0:17:52.160 --> 0:17:56.080 gathering other relevant information, and manipulating data, which are all 0:17:56.200 --> 0:18:00.000 tasks that AI can automate. Mike himself reckons that probably 0:18:00.000 --> 0:18:02.240 about a third of his time is spent on those tasks. 0:18:03.000 --> 0:18:06.520 The search. Yeah, you can automate looking for the needle 0:18:06.560 --> 0:18:09.280 in the haystack, or in this case, a few needs. 0:18:09.800 --> 0:18:13.119 You can automate finding a few needles in the haystack. 0:18:13.720 --> 0:18:16.439 That would be fantastic. That give us more time to 0:18:16.520 --> 0:18:19.879 go kick the tires and talk to management and have 0:18:20.119 --> 0:18:24.480 much more creative research. Richard, the researcher from Greenwich Associates, 0:18:24.680 --> 0:18:28.880 estimates that of finance jobs at risk from AI automation 0:18:29.280 --> 0:18:32.760 and research jobs among the most exposed. I think there 0:18:32.800 --> 0:18:34.920 will be an impact on jobs. I think for sure, 0:18:36.119 --> 0:18:39.480 we know that there's significant productivity and efficiency gains that 0:18:39.520 --> 0:18:43.199 can come from it. You still need, uh, you know, 0:18:43.280 --> 0:18:46.399 some people, you know, some human analysts to kind of 0:18:46.400 --> 0:18:49.560 interpret a lot of the data and so forth. Of course, 0:18:50.040 --> 0:18:54.080 in this example about the world's fargo on the programmer, 0:18:55.160 --> 0:18:58.160 he obviously had to put a lot of inputs into that, 0:18:58.280 --> 0:19:01.919 telling the machine learning algorithm what to look for and 0:19:01.960 --> 0:19:04.359 so forth. So that type of skill is still going 0:19:04.359 --> 0:19:07.119 to be needed. But maybe you don't need a team 0:19:07.119 --> 0:19:17.720 of fifty junior analysts trying to so will that iconic 0:19:17.880 --> 0:19:21.800 trading floor Hubbub and chaos ever be silenced by machines. 0:19:22.520 --> 0:19:25.520 For now, I'd say that the human analysts are safe, 0:19:25.640 --> 0:19:27.960 but I'm I'm definitely not ready to put my money 0:19:28.040 --> 0:19:31.120 on them and their associates all being there in five 0:19:31.200 --> 0:19:33.639 to ten years. Ken himself is holding out hope for 0:19:33.760 --> 0:19:36.800 human stock research. He used his final moments of the 0:19:36.840 --> 0:19:40.280 interview to re emphasize that ERA is an enhancement rather 0:19:40.320 --> 0:19:44.240 than a replacement to everything that Mike said. As we 0:19:44.280 --> 0:19:46.960 look out over the next so many years, it is 0:19:47.000 --> 0:19:49.520 really an opportunity for us to see kind of an 0:19:49.680 --> 0:19:53.000 enhanced performance and an ability for us to move into 0:19:53.080 --> 0:19:57.640 sort of a creative research sphere that would be difficult 0:19:57.680 --> 0:20:08.000 to do otherwise. So, Julie, are other banks starting to 0:20:08.040 --> 0:20:10.800 do this too? Were showing signs of following the path 0:20:10.920 --> 0:20:14.240 forged by Brian and Ken. We've seen a little bit 0:20:14.359 --> 0:20:17.359 like this. Um nothing exactly what Brian and Kenn are doing, 0:20:17.560 --> 0:20:20.760 but Morgan Stanley has a sort of bought that will 0:20:20.840 --> 0:20:24.080 help research analysts dive through earning season, which is obviously 0:20:24.080 --> 0:20:26.280 one of the busiest times of years for these guys. 0:20:26.320 --> 0:20:29.320 Tons of news articles coming out each day, so we'll 0:20:29.320 --> 0:20:32.560 help automate what those earnings are and whatnot. But otherwise 0:20:32.560 --> 0:20:36.320 there's not anything exactly like ERA that's skimming through news articles, 0:20:36.359 --> 0:20:39.400 social media and whatnot on a daily basis. I heard 0:20:39.440 --> 0:20:42.200 you refer to Era as a she a couple of times. 0:20:42.280 --> 0:20:45.320 Ken did as well, why are we giving this AI 0:20:45.400 --> 0:20:49.240 analyst agender. It's a little bit of a canned response, 0:20:49.320 --> 0:20:52.840 and that it's just because there aren't enough female analysts 0:20:52.840 --> 0:20:55.000 on Wall Street. So I guess if AARA takes over, 0:20:55.080 --> 0:20:59.960 then there won't be enough male analysts on Wall Street. Okay, Alice, 0:21:00.280 --> 0:21:03.600 it's a little surprising that it took Wall Street this 0:21:03.760 --> 0:21:06.800 long to do this when algorithmic trading it already transformed 0:21:06.840 --> 0:21:10.000 finance years ago. What took the industry so long? Well, 0:21:10.119 --> 0:21:13.600 some of it is evident in the performance of error actually, 0:21:13.800 --> 0:21:16.800 So each time it makes a call on a stock, 0:21:16.920 --> 0:21:18.880 that's a data point that can be fed back into 0:21:18.920 --> 0:21:21.879 the system. Now you compare that to you know, automated 0:21:21.920 --> 0:21:24.560 trading systems that hedge funds they trade, you know, maybe 0:21:24.760 --> 0:21:26.560 you know, ten times a minute or something like that. 0:21:26.760 --> 0:21:28.399 So you just have a lot more data points to 0:21:28.400 --> 0:21:30.760 train the systems on. Oh, so that leads to the 0:21:31.200 --> 0:21:33.879 last question here of course, as journalist, we are not 0:21:33.960 --> 0:21:37.720 allowed to invest or buy individual shares in the companies 0:21:37.760 --> 0:21:40.439 we cover. But let's put aside those rules for a second. 0:21:40.760 --> 0:21:44.159 With what we learned today about Era, Julie, would you 0:21:44.200 --> 0:21:48.880 follow the bot's recommendation? Would you take advice from an Ai? 0:21:49.240 --> 0:21:52.400 Not at this point, but I with how quickly she's 0:21:52.560 --> 0:21:56.160 gotten smarter and learned from her recommendations, I would say 0:21:56.160 --> 0:21:58.400 in a year or two, I want to be surprised 0:21:58.400 --> 0:22:01.040 if I was al Stair. I'm a I'm a human 0:22:01.160 --> 0:22:03.240 all the way. Um, the only thing I probably would 0:22:03.280 --> 0:22:05.680 use it for is to scan the all the news 0:22:05.760 --> 0:22:07.840 articles to see what's most important that that I would 0:22:07.840 --> 0:22:10.960 make my hand decisions. Yeah, I think I probably agree 0:22:11.000 --> 0:22:14.720 with you. I'm gonna I'm gonna sound not as future 0:22:14.800 --> 0:22:17.040 leaning as as Julie. I guess I just think that 0:22:17.640 --> 0:22:20.480 the inputs that Air is looking at, primarily it seems 0:22:20.480 --> 0:22:23.400 like it is the news is you know, one important 0:22:23.400 --> 0:22:25.760 to mention, but it's not everything. And it just does 0:22:25.800 --> 0:22:28.439 seem like there's so much intuition and knowledge of a 0:22:28.520 --> 0:22:32.199 management team and it's past mistakes and future opportunities that 0:22:32.240 --> 0:22:34.760 goes into making these stock picks. I don't think air 0:22:34.880 --> 0:22:36.879 is there yet. But Julie, maybe I do agree with 0:22:36.920 --> 0:22:40.639 you that one day, probably within our lifetimes, perhaps we 0:22:40.760 --> 0:22:44.320 we will be taking training advice from an artificial intelligence. Yeah, 0:22:44.359 --> 0:22:46.639 and I guess maybe the most likely scenario is that 0:22:46.720 --> 0:22:49.720 it's just Ken and Mike and not the three or 0:22:49.720 --> 0:22:51.920 four people they have working for them, right, Like, they 0:22:51.960 --> 0:22:54.439 just have an era and that's all they need. The 0:22:54.440 --> 0:22:56.720 problem is if they if they give you bad advice, 0:22:56.800 --> 0:23:09.840 you know, can you really blame the robot? Tbd And 0:23:09.920 --> 0:23:13.120 that's it for this week's episode of Decrypted. Thanks for listening. 0:23:13.440 --> 0:23:16.800 Is artificial intelligence changing your workplace? Send us an email 0:23:16.800 --> 0:23:19.920 at decrypted at Bloomberg dot net, or you can reach 0:23:19.960 --> 0:23:22.840