WEBVTT - Smart Talks with IBM and Malcolm Gladwell: Using AI to Rethink the Way Work Gets Done 0:00:02.440 --> 0:00:07.280 Hello everyone. This is Smart Talks with IBM, a podcast 0:00:07.640 --> 0:00:12.239 from Pushkin Industries, I Heart Media and IBM about what 0:00:12.280 --> 0:00:15.160 it means to look at today's most challenging problems in 0:00:15.200 --> 0:00:24.439 a new way. I'm Malcolm Gladwell. Today I'm chatting with 0:00:24.520 --> 0:00:28.240 Rob Thomas, the senior vice president of IBM Cloud and Data, 0:00:28.640 --> 0:00:35.519 where his responsibility is bringing new ideas to life. But 0:00:35.600 --> 0:00:38.800 despite being on the cutting edge of these technologies, he 0:00:38.880 --> 0:00:46.880 still has an appreciation for age old problems. There's a 0:00:47.000 --> 0:00:49.400 rabbit and a beaver and they're staring at the Hoover 0:00:49.520 --> 0:00:53.480 dam and the beaver says the rabbit, No, I didn't 0:00:53.479 --> 0:00:55.440 build it, but it's based on an idea of mind. 0:00:57.120 --> 0:00:59.720 And the point of that story is there's ideas are 0:00:59.720 --> 0:01:05.280 a dye that doesn't so great story. Everybody's got a 0:01:05.319 --> 0:01:08.440 bunch of ideas. By the way, we're too quick to 0:01:08.480 --> 0:01:12.000 dismiss the beaver. He's right, but I have you seen 0:01:12.080 --> 0:01:15.720 beaver dams. I mean, he's right, it was his idea, 0:01:16.360 --> 0:01:18.319 but he had nothing to do with the giants Cement 0:01:18.360 --> 0:01:24.600 Hoover dow. In my interview with Rob will touch on 0:01:24.640 --> 0:01:27.479 the importance of the cloud during the pandemic and how 0:01:27.520 --> 0:01:38.160 IBM has been playing a part in vaccine distribution Stay tuned. I, 0:01:38.360 --> 0:01:41.160 for one, had no idea about what it means to 0:01:41.200 --> 0:01:43.360 be the senior vice president of IBM Cloud and Data, 0:01:43.720 --> 0:01:45.720 so I asked Rob to break it down for me 0:01:45.760 --> 0:01:52.160 in Layman's terms, we build software, and software is the 0:01:52.240 --> 0:01:56.480 lingua franca of our time. Anything that will get done 0:01:56.680 --> 0:02:02.160 in businesses and even interaction with consumers is going to 0:02:02.200 --> 0:02:05.680 be done with software. It's really the language of everything 0:02:05.720 --> 0:02:08.560 that's happening in the world. That's what we build. We 0:02:08.639 --> 0:02:12.480 are focused on doing that for businesses. So how how 0:02:12.520 --> 0:02:16.359 long have you been at IBAN twenty years or one? 0:02:16.400 --> 0:02:19.639 I guess to be precise. And I started in consulting, 0:02:20.600 --> 0:02:24.520 and then I moved into our semiconductor business, and I 0:02:24.560 --> 0:02:29.320 was doing consulting. And the moment that really changed my 0:02:29.400 --> 0:02:33.120 whole career was doing work with Nintendo where we were 0:02:33.160 --> 0:02:38.640 designing the microprocessor for the Nintendo We and I realized, 0:02:39.200 --> 0:02:41.320 we're going to do this one time, but then they're 0:02:41.360 --> 0:02:45.040 going to be building software that we get copied billions 0:02:45.040 --> 0:02:47.560 of times and used by people all over the world. 0:02:48.320 --> 0:02:50.720 Maybe I'm not in the right business. And that really 0:02:50.840 --> 0:02:56.360 piqued my curiosity around software, which then led me to 0:02:56.720 --> 0:02:59.840 move into the IBM software business, where I've been for 0:03:00.240 --> 0:03:02.440 most of my career at this point. So I've been 0:03:02.440 --> 0:03:06.720 in software total twelve thirteen years. But you have seen 0:03:06.760 --> 0:03:09.960 I'm guessing, so twelve years in software. Am I right 0:03:10.000 --> 0:03:13.120 in thinking that Morris happened in those twelve years of 0:03:13.160 --> 0:03:16.840 software than happened in the entire history of software before? That? 0:03:16.880 --> 0:03:20.720 Is that a fair statement? Close? It's I'd say close. 0:03:20.960 --> 0:03:24.400 It's certainly the rate and pace of innovation has increased. 0:03:25.040 --> 0:03:27.640 Now has actually something hasn't read an outcome? Maybe that's 0:03:27.680 --> 0:03:31.280 a different question. But if you think about you know, 0:03:31.440 --> 0:03:35.120 software dates way back to even the first main frame 0:03:35.160 --> 0:03:38.560 that we ever built in the fifties. So a lot 0:03:38.560 --> 0:03:40.520 of good things have been happening in software for a 0:03:40.520 --> 0:03:43.400 long time. But the rate and pace is a level 0:03:43.440 --> 0:03:46.240 that we've never seen, and that's certainly been what is 0:03:46.280 --> 0:03:50.080 accelerated in the last decade. I mean, I remember my 0:03:50.160 --> 0:03:53.680 dad was a mathematician at the University of Waterloo. I 0:03:53.760 --> 0:03:56.120 remember coming home as a kid, going into his office 0:03:56.120 --> 0:03:59.520 and seeing stacks of computer cards. So in my lifetime 0:04:00.320 --> 0:04:03.320 I have I have gone from looking at stacks of 0:04:03.360 --> 0:04:07.720 computer cards to something far more so. I mean, I 0:04:07.760 --> 0:04:11.000 am aware of just how fast this fast, this uh, 0:04:11.160 --> 0:04:13.760 this pace is gone and it will be different a 0:04:13.840 --> 0:04:17.480 year from now. Right, that's how fast this is moving. 0:04:18.480 --> 0:04:21.640 Let's zero in on that a little bit. Um, What's 0:04:22.160 --> 0:04:26.120 what's shifting right now? Imagine I'm a client and I 0:04:26.160 --> 0:04:27.920 come to you and I say, you know, I want 0:04:27.960 --> 0:04:30.919 to be prepared for next year and the year after next. 0:04:31.320 --> 0:04:35.880 What should be at the top of my mind. Let 0:04:35.880 --> 0:04:37.279 me give you a quick story, if you don't mind. 0:04:37.640 --> 0:04:40.240 There was a time in the US where you could 0:04:40.279 --> 0:04:45.240 not easily get from one city to another. And at 0:04:45.240 --> 0:04:48.080 that point, back in the nine fifties, there was a 0:04:48.160 --> 0:04:52.880 decision that said, let's actually build the infrastructure to connect 0:04:53.520 --> 0:04:58.279 every city in America. And the result was fifty plus 0:04:58.360 --> 0:05:00.919 years of work fo dollar ours and we now have 0:05:01.120 --> 0:05:05.280 forty eight thousand miles of highways that connects all these cities. 0:05:06.360 --> 0:05:09.520 But the real impact is more profound than that, because 0:05:10.240 --> 0:05:14.560 you're able to eliminate traffic at intersections by building over passes. 0:05:15.440 --> 0:05:20.200 There are all these second order businesses that were built. Hotels, 0:05:21.080 --> 0:05:24.440 gas stations, the salty snacks that you buy in a 0:05:24.480 --> 0:05:30.400 gas station, fast food rest areas, so an entire economy 0:05:31.040 --> 0:05:34.279 was built around the idea that the first step was 0:05:34.320 --> 0:05:38.679 just to connect all the cities in the US. And 0:05:39.160 --> 0:05:42.360 that's what's happening right now with software. It is connecting 0:05:42.400 --> 0:05:46.120 businesses and individuals in a way that we've never been 0:05:46.120 --> 0:05:49.360 connected before, and we are just at the beginning of 0:05:49.400 --> 0:05:52.760 all the second order effects that will come as a 0:05:52.800 --> 0:05:57.240 result of them. And the biggest problem in software it's data. 0:05:58.320 --> 0:06:01.479 Just like you had all these disparate cities and you 0:06:01.600 --> 0:06:06.279 were building highways to connect those cities, every company has 0:06:06.440 --> 0:06:10.600 all these different data sets all over the place and 0:06:10.600 --> 0:06:13.719 it's a really hard problem. But AI is not going 0:06:13.800 --> 0:06:17.960 to be a reality for businesses until the data problem 0:06:18.160 --> 0:06:22.200 is solved. That's one thing that I spent a lot 0:06:22.200 --> 0:06:26.000 of time on Right now, we're dig digging from it 0:06:26.120 --> 0:06:29.440 that into that meaning of that phrase, the data problem. 0:06:29.480 --> 0:06:34.680 I think every individual wants any company they interact with, 0:06:34.839 --> 0:06:40.120 whether it's their local bank or restaurant, or the local cleaners, 0:06:40.760 --> 0:06:45.320 whatever it may be, they want that business to know them. 0:06:45.360 --> 0:06:48.719 It's the whole idea of when you had towns where 0:06:48.720 --> 0:06:51.200 there was just one general store and the owner knew you, 0:06:51.640 --> 0:06:54.240 they knew what you wanted. I think everybody wants that 0:06:54.360 --> 0:07:01.440 level of engagement, and that is what software enables. And 0:07:01.520 --> 0:07:05.320 the basis of that is data. And the biggest problem 0:07:05.480 --> 0:07:09.760 every business faces today is how do I understand my data? 0:07:09.960 --> 0:07:12.119 What it tells me about my customers, what it tells 0:07:12.120 --> 0:07:16.440 me about my products. So this is fundamentally about how 0:07:16.480 --> 0:07:20.640 do we live in a better way. You're talking about that. 0:07:20.800 --> 0:07:24.160 I'm I'm a big company, and I have different sets 0:07:24.160 --> 0:07:26.560 of data and they're all in different places, and they 0:07:26.560 --> 0:07:29.120 don't speak to each other, and I can't combine them 0:07:29.120 --> 0:07:30.600 and make sense. Is that what you mean by the 0:07:30.680 --> 0:07:36.000 data problem? Correct? And even if I can combine them 0:07:36.000 --> 0:07:40.240 and connect them, the data is not in a usable form. 0:07:40.280 --> 0:07:43.120 You know, one one data, says m Gladwell, the other 0:07:43.120 --> 0:07:46.160 one says Malcolm g Is that the same person? Maybe? 0:07:46.200 --> 0:07:50.240 Maybe not. It's really hard because these systems have been 0:07:50.240 --> 0:07:53.880 built up over time. We do work with a company 0:07:53.920 --> 0:07:58.480 called Wonderman. Thompson story that they shared with me just 0:07:58.520 --> 0:08:04.200 this month was doing work with Peloton. So Peloton collects 0:08:04.480 --> 0:08:06.800 a lot of data what you call first party data 0:08:06.880 --> 0:08:09.640 from a bike or the tread. I think you're a 0:08:09.680 --> 0:08:14.720 runner if I recall and w P P. Wonderman Thompson 0:08:14.800 --> 0:08:17.120 has all this third party data, which is what do 0:08:17.200 --> 0:08:20.280 they know about consumers? So just to connect those two 0:08:20.360 --> 0:08:25.080 data sets, build predictive models, and then to turn that 0:08:25.120 --> 0:08:29.520 into an advertising campaign. The AI part is actually relatively easy. 0:08:29.880 --> 0:08:34.160 It's actually connecting the data, rationalizing the data, cleaning the data. 0:08:34.440 --> 0:08:36.680 That's the really hard part that nobody talks about because 0:08:36.679 --> 0:08:40.559 all we ever see is you know the outcome, yeah, 0:08:40.800 --> 0:08:43.400 which is so I understand this is super interesting. So 0:08:43.480 --> 0:08:47.040 let's imagine you Robb or a Peloton user, and so 0:08:47.160 --> 0:08:49.199 we have a data stream that comes from the bike 0:08:49.840 --> 0:08:53.040 which says that you bike. Let's just say for them, 0:08:53.040 --> 0:08:55.400 I'm gonna flatter you an hour and a half a 0:08:55.480 --> 0:08:59.240 day at some insane pace and neither of which are true. 0:08:59.240 --> 0:09:01.720 But keep going. I did do a half hour today, 0:09:01.720 --> 0:09:04.120 but it was a very slow pace. I gotta tell you. So, 0:09:04.200 --> 0:09:07.240 I'm and I'm looking at your via whatever it is 0:09:07.280 --> 0:09:10.880 I'm collecting. I'm assuming Peloton collects a lot of sort 0:09:10.880 --> 0:09:14.000 of physiological and you stat up on the bike, and 0:09:14.080 --> 0:09:17.760 from that we can generate a rough sense of who 0:09:17.800 --> 0:09:20.719 you are, how what your athletic interests are, how fit 0:09:20.800 --> 0:09:23.600 you are, all those kinds of things, and Wonderman's Old 0:09:23.640 --> 0:09:27.960 shop wants to know, how can I use that picture 0:09:28.040 --> 0:09:31.920 of the kind of athlete you are to help bring 0:09:31.920 --> 0:09:34.240 you the kinds of ad messages that you'll respond to. 0:09:34.320 --> 0:09:36.880 Is that a fair? Is that the problem? It could 0:09:36.920 --> 0:09:39.040 be bringing it to me? But it's more likely because 0:09:39.040 --> 0:09:42.680 obviously they d anonymize all this data. It's more of all, right, so, 0:09:42.720 --> 0:09:45.160 how do we find somebody else that's like rob? What 0:09:45.240 --> 0:09:47.320 are the attributes of that person? And then how do 0:09:47.360 --> 0:09:50.280 we relate to them in a way that makes it 0:09:50.320 --> 0:09:53.720 feel like we're talking to them as opposed to talking 0:09:53.760 --> 0:09:58.760 to a cohort or a group. The number one prediction 0:09:58.800 --> 0:10:01.480 that most companies want to is what's going to happen 0:10:01.520 --> 0:10:04.440 to my sales next month or the month after or 0:10:04.440 --> 0:10:08.520 the month after. And what we found is that tends 0:10:08.559 --> 0:10:11.440 to be a product of as many as fifty or 0:10:11.440 --> 0:10:14.400 a hundred different inputs. How many people are visiting the website, 0:10:14.840 --> 0:10:17.520 how many people are calling the call center, how many 0:10:17.559 --> 0:10:20.240 sales calls if I have a face to face salesforce 0:10:20.440 --> 0:10:23.400 are they making? How many marketing campaigns am I running? 0:10:24.880 --> 0:10:27.360 If you take all of these different data points, which 0:10:27.400 --> 0:10:30.400 is awful in fifty or a hundred. You feed those 0:10:30.440 --> 0:10:35.679 into a model. Then the first month you see how 0:10:35.720 --> 0:10:38.200 close with the model. Then you adjust, second month, you 0:10:38.240 --> 0:10:41.200 see how close was the model? And these models get 0:10:41.240 --> 0:10:44.880 really good over time. And we think we can help 0:10:44.920 --> 0:10:48.520 companies predict their financial performance in a month and a 0:10:48.600 --> 0:10:52.600 quarter in a year based on all these different data sources, 0:10:52.640 --> 0:10:55.640 all these different inputs. That's pretty valuable to let's say 0:10:55.640 --> 0:10:59.600 every company. So IBM, what's IBM's role in that you've 0:10:59.640 --> 0:11:01.800 described that problem to me? What is you guys come 0:11:01.840 --> 0:11:06.959 in and you say, we'll do what? A couple of 0:11:07.000 --> 0:11:08.840 years ago I started. I was trying to think about 0:11:08.920 --> 0:11:12.720 what is the right metaphor so that I can educate 0:11:14.120 --> 0:11:18.240 our customers on this and built this concept that I 0:11:18.320 --> 0:11:26.520 called the AI ladder. To think of it as steps 0:11:26.559 --> 0:11:30.160 that you take up a ladder towards AI. The bottom 0:11:30.240 --> 0:11:32.320 rung is collect data. So you have to be able 0:11:32.360 --> 0:11:36.360 to collect all your data. I'll use a library analogy. 0:11:36.480 --> 0:11:38.200 This is just you have to get books. You have 0:11:38.240 --> 0:11:42.040 to get books into the library that's collecting. Next is 0:11:42.080 --> 0:11:45.680 you have to organize that data and the now the 0:11:45.720 --> 0:11:48.120 a lot back to library analogy. That's the card catalog. 0:11:48.520 --> 0:11:51.200 So where are all the different data sets. I might 0:11:51.240 --> 0:11:53.600 have five copies of the same data. How do I 0:11:53.640 --> 0:11:56.559 know that's the same copy. Maybe one's checked out, maybe 0:11:56.559 --> 0:12:00.000 one's on microfilm. These are actually all problems that existing businesses. 0:12:00.160 --> 0:12:03.640 So you've got to collect data, you've got to organize data. 0:12:04.040 --> 0:12:06.520 Then you have to analyze the data. So you're actually 0:12:06.520 --> 0:12:11.160 starting to do data science machine learning in the library 0:12:11.160 --> 0:12:14.600 metaphor that's where you're displaying your best seller list or 0:12:14.640 --> 0:12:18.560 you're displaying, you know, popular magazine titles. And then the 0:12:18.600 --> 0:12:21.559 top of the ladder is what I call infuse. So 0:12:21.600 --> 0:12:26.120 then how do you take those models and infuse them 0:12:26.120 --> 0:12:30.679 into a business process. So it's those four steps the ladder. 0:12:30.720 --> 0:12:36.280 You have to collect, organized, analyze, and fuse. We build 0:12:36.280 --> 0:12:38.960 software that helps customers with each of the wrongs of 0:12:39.000 --> 0:12:42.400 that ladder, helps them do the collection. We actually build 0:12:42.400 --> 0:12:44.880 what we call a data catalog to help you organize 0:12:44.880 --> 0:12:47.400 your data. So we help them with all wrongs of 0:12:47.440 --> 0:12:51.720 that ladder. Because ultimately, then you've probably heard of IBM Watson, 0:12:51.880 --> 0:12:55.680 that is our AI platform. Once you've done those things, 0:12:57.080 --> 0:13:04.439 you can use AI and get really great outcomes. Imagine 0:13:04.480 --> 0:13:09.439 if someone from the White House came to you and 0:13:09.480 --> 0:13:12.559 said we're about to do something we've never done in 0:13:12.600 --> 0:13:17.040 this haven't done in this country for seventy years, which 0:13:17.080 --> 0:13:19.960 is try and vaccinate everybody in the shortest possible time. 0:13:20.600 --> 0:13:24.320 We have a multiple sets of three and eventually probably 0:13:24.320 --> 0:13:27.880 four or five different kinds of vaccines being administered by 0:13:28.040 --> 0:13:31.520 tens of thousands of local municipalities too, people who have 0:13:31.559 --> 0:13:38.199 a wide ranging set of risk factors, urgency, pre existing 0:13:38.200 --> 0:13:39.720 conditions going on, you know, on and on and on 0:13:39.760 --> 0:13:43.360 and on on um. Can you help us do this 0:13:43.600 --> 0:13:51.400 as efficiently and cost effectively and socially consciously as possible? 0:13:52.640 --> 0:13:54.760 Is that a kind of task that you're talking about 0:13:55.840 --> 0:14:00.280 now that is in part as much a logistics problem 0:14:00.320 --> 0:14:02.959 as it is a data problem. Let me describe to 0:14:02.960 --> 0:14:05.040 you one of the data problems though, that exist around 0:14:05.040 --> 0:14:09.640 this because we're doing the work with CVS on the 0:14:09.760 --> 0:14:16.600 COVID vaccine rollout. Yes, and so if you're CVS where 0:14:16.600 --> 0:14:21.320 you're actually administering, their biggest problem is everybody has a question. 0:14:22.440 --> 0:14:25.680 CVS can't hire enough people to answer the ten questions 0:14:25.720 --> 0:14:28.480 you have, the ten questions I have, the twenty questions 0:14:28.520 --> 0:14:32.800 your cousin has. They came to us and said, can 0:14:32.840 --> 0:14:37.080 we use AI too respond to all the inquiries we're 0:14:37.080 --> 0:14:41.160 getting and actually help route people to where they can 0:14:41.200 --> 0:14:43.320 figure out they can get the vaccine when they're eligible. 0:14:44.000 --> 0:14:47.400 So we built an AI agent for them that is 0:14:47.440 --> 0:14:52.600 now dealing with the the vaccine rollout every day that 0:14:52.720 --> 0:14:56.200 starts with data. They have a place that they store 0:14:56.560 --> 0:15:00.480 data about different questions. We've got models that we have 0:15:00.640 --> 0:15:05.000 trained on language, meaning we can understand different types of questions, 0:15:05.040 --> 0:15:09.400 which really inferred versus implied versus what is stated. That's 0:15:09.440 --> 0:15:14.000 a real data problem. That's where we've spent the majority 0:15:14.040 --> 0:15:17.600 of our time looking at this, this current situation. So 0:15:17.680 --> 0:15:20.480 you would that you when you say it's a data problem, 0:15:20.640 --> 0:15:26.080 meaning that you started by trying to anticipate, by looking 0:15:26.080 --> 0:15:28.160 at the data and using that to try and anticipate 0:15:28.200 --> 0:15:30.960 all the possible questions that someone might ask, Is that 0:15:30.960 --> 0:15:34.760 what you're correct? Yes, um, and then training a machine 0:15:34.840 --> 0:15:38.480 learning model based on those inputs so that when the 0:15:38.560 --> 0:15:42.160 question was asked, we had a high probability of giving 0:15:42.160 --> 0:15:44.640 the right answer. How long did it take you to 0:15:44.640 --> 0:15:50.440 build that system? Now this is the wonders of modern software. 0:15:50.600 --> 0:15:53.760 To your question on acceleration, we did this in forty 0:15:53.800 --> 0:15:58.680 five days. Are you serious. Yeah, it's insane. How how 0:15:58.680 --> 0:16:04.240 many people worked on it? Thirty somewhere in that room. 0:16:04.320 --> 0:16:09.600 It's not a huge group. Wow. Their thing about systems 0:16:09.640 --> 0:16:12.840 like this is you hope it's really good on day one, 0:16:13.440 --> 0:16:15.000 but you know for sure it's going to be better 0:16:15.360 --> 0:16:18.440 on day ten. It's gonna be better again on day twenty. 0:16:18.840 --> 0:16:21.480 These are learning systems. They do get better over time. 0:16:21.880 --> 0:16:25.480 And the thing is with with the really difficult problems. 0:16:26.040 --> 0:16:28.000 And this is this is why I like to talk 0:16:28.040 --> 0:16:31.200 about AI is giving human superpowers. Most people want to 0:16:31.200 --> 0:16:34.400 say it replaces humans. I actually think given superpowers because 0:16:34.400 --> 0:16:37.360 in these cases you start to move the harder problems 0:16:37.400 --> 0:16:41.720 to the humans, and so therefore your your customer satisfaction 0:16:41.760 --> 0:16:45.120 goes up because people are getting their problems resolved. Would 0:16:45.120 --> 0:16:47.640 you ever get to Probably not, because there's always going 0:16:47.680 --> 0:16:50.640 to be something that's too difficult for the AI to handle, 0:16:50.680 --> 0:16:52.640 But I think you can keep moving it up for sure. 0:16:53.400 --> 0:16:55.920 Or maybe given what you've just said, would it be 0:16:55.960 --> 0:16:57.720 more fair to say you don't want to get to 0:16:57.760 --> 0:17:01.480 a hundred, that you want to rise nerve a certain 0:17:01.720 --> 0:17:05.080 category of problem for a human human interaction, because that 0:17:05.160 --> 0:17:10.159 might be ultimately more satisfying to the question. We have 0:17:10.240 --> 0:17:12.600 that discussion a lot, and certainly in the ones that 0:17:12.600 --> 0:17:16.120 I've worked on, that's that's typically the case, because let's 0:17:16.119 --> 0:17:19.400 not forget these are businesses, and the goal of most 0:17:19.440 --> 0:17:22.400 businesses is to sell something. So sometimes the best way 0:17:22.440 --> 0:17:24.600 to sell something is to really help somebody with their 0:17:24.600 --> 0:17:27.960 problem and then show them how your other product can 0:17:28.000 --> 0:17:30.960 make their life even easier. When you think back in 0:17:31.040 --> 0:17:35.680 the cases that you have kind of problems that that 0:17:35.840 --> 0:17:38.159 your group has been asked to solve it, IBM of 0:17:38.240 --> 0:17:42.400 last couple of years, what was the hardest I don't 0:17:42.400 --> 0:17:45.080 know that I could name a single thing that's harder 0:17:45.119 --> 0:17:48.359 than others. The ones that are the most time consuming, 0:17:48.840 --> 0:17:53.280 things like regulatory compliance. If you're a bank, you've got 0:17:53.280 --> 0:17:55.879 a lot of different regulations that you have to to 0:17:56.119 --> 0:18:01.320 live up to. And it's easy to help build AI 0:18:01.359 --> 0:18:05.240 that can make loan decisions yes or no, good idea, 0:18:05.280 --> 0:18:09.040 bad idea, eliminate bias from that decision. That's very doable. 0:18:10.359 --> 0:18:14.000 Am I Compliance with the regulations of where that individual 0:18:14.080 --> 0:18:15.960 is based, because they're in a zip code, or they're 0:18:15.960 --> 0:18:18.640 in a state, they're in a country, those problems get 0:18:18.680 --> 0:18:23.440 really difficult because you're kind of connecting, you know, reams 0:18:23.480 --> 0:18:28.240 of legality to a day to day business process. Those 0:18:28.280 --> 0:18:32.840 get those those get really difficult. Has any customer ever 0:18:32.880 --> 0:18:34.480 come to you with a problem that you, guys said, 0:18:34.520 --> 0:18:39.639 we can't solve that. We are way too curious to 0:18:40.000 --> 0:18:44.679 ever give up that easily. It's more of, you know, 0:18:45.119 --> 0:18:48.000 it's the it's the cheap, fast, and good triangle. If 0:18:48.040 --> 0:18:50.879 you've heard that, you know you only get two of those. 0:18:51.080 --> 0:18:53.080 Do you want it cheap and fast, it's probably not good. 0:18:53.280 --> 0:18:55.280 If you want it good and fast, it's probably not cheap. 0:18:56.560 --> 0:18:58.520 If you want it cheap and good, it's probably not fast. 0:18:59.119 --> 0:19:02.600 I think all of these situations come down to that triangle. 0:19:04.320 --> 0:19:05.879 So you have a group of people who work on 0:19:05.920 --> 0:19:08.720 these kinds of problems, and I'm curious, what do you 0:19:08.840 --> 0:19:14.879 look for when you're bringing someone onto that team. Is 0:19:14.880 --> 0:19:18.520 there a set of skills associated with dealing with this 0:19:18.640 --> 0:19:22.600 area of the application of AI to these very complicated 0:19:23.600 --> 0:19:26.440 data fields. Is there a specific set of skills that 0:19:26.480 --> 0:19:32.919 are crucial and rare hard find? The skill that's easy 0:19:32.960 --> 0:19:38.119 to test for is do you have the technical ability? 0:19:38.640 --> 0:19:45.439 Do you understand Python? Do you understand machine learning? You 0:19:45.480 --> 0:19:47.679 can kind of see from somebody's body of work and 0:19:47.720 --> 0:19:51.160 what they've studied. Do they have that skill? Part where 0:19:51.160 --> 0:19:55.720 it gets harder is the empathy question. Can you actually 0:19:55.800 --> 0:20:01.280 understand a situation, understand a user, and empathize with what 0:20:01.320 --> 0:20:03.720 they're trying to do such that you're not just building 0:20:03.720 --> 0:20:06.080 a model for a robot, You're actually building this for 0:20:06.320 --> 0:20:10.119 a human on some end, that one's hard, harder to 0:20:10.160 --> 0:20:12.840 test for. And then the third one is I would 0:20:12.880 --> 0:20:17.840 just call it curiosity, how widely read as somebody do 0:20:17.920 --> 0:20:23.040 they understand business business problems because those kind of softer skills, 0:20:24.480 --> 0:20:27.600 those make a huge difference when you're solving these kind 0:20:27.600 --> 0:20:29.919 of problems. So it's easy to test for the first. 0:20:30.600 --> 0:20:32.480 The other two are a little harder to test for, 0:20:32.960 --> 0:20:36.400 and the best data scientists in the world have all 0:20:36.400 --> 0:20:43.600 three of those. Let's talk about um the cloud. I 0:20:43.680 --> 0:20:47.200 see this word hybrid cloud. I don't know what it means. 0:20:48.040 --> 0:20:50.280 So can you start by telling me what it means 0:20:50.680 --> 0:20:53.320 and then fit this into the conversation we've been having. 0:20:55.160 --> 0:20:59.439 So any company that's been around for more than three years, 0:21:00.680 --> 0:21:06.480 maybe five, they've got somewhere that they keep their data 0:21:07.720 --> 0:21:11.159 and they keep the different technology that they have, and 0:21:11.200 --> 0:21:13.840 in many cases that's in their office or that's in 0:21:13.880 --> 0:21:18.480 a data center right right near their office. They've also 0:21:18.560 --> 0:21:23.320 started over time to start to build new data sets 0:21:23.440 --> 0:21:29.639 or new software in a public cloud, something from you know, 0:21:29.760 --> 0:21:34.840 something inside of IBM Cloud or Amazon Web Services or 0:21:34.960 --> 0:21:40.000 Microsoft Azure. The minute that you have more than one environment, 0:21:41.960 --> 0:21:44.760 you have a hybrid cloud, whether whether you know it 0:21:44.840 --> 0:21:48.280 or not. So think of it as I've got multiple 0:21:48.359 --> 0:21:50.520 data sets and multiple places to kind of back to 0:21:50.560 --> 0:21:55.480 the US Highway example, or I've got software applications and 0:21:55.640 --> 0:21:59.879 multiple places. You have to get that to act like 0:22:00.080 --> 0:22:05.560 a single technology environment. That is the essence of hybrid cloud, 0:22:05.640 --> 0:22:08.160 which is I can manage that as a single environment. 0:22:08.560 --> 0:22:11.639 The average company now has five different environments cloud wise, 0:22:12.880 --> 0:22:15.400 it acts like one. I can connect the data sets 0:22:15.480 --> 0:22:19.960 the average company has. Is that by by design because 0:22:20.000 --> 0:22:23.119 they feel it's safe for Is that just because the 0:22:23.160 --> 0:22:26.879 hodgepodge nature in which we grow our I team needs 0:22:26.960 --> 0:22:28.720 means that we end up being all over the place. 0:22:31.000 --> 0:22:34.359 It's because there's a lot of people that work in 0:22:34.720 --> 0:22:37.800 every company, and everybody wants their own thing. That's how 0:22:37.800 --> 0:22:41.800 it happens. So that this department started in their own 0:22:41.840 --> 0:22:46.440 data center, This department started on IBM cloud. This department 0:22:46.520 --> 0:22:50.560 wanted a CRM system from Salesforce, this department wanted to 0:22:50.680 --> 0:22:54.800 use Azure. It's human nature. People just go do what 0:22:54.880 --> 0:22:57.400 they want to do. And you wake up one day 0:22:57.400 --> 0:22:59.560 and you realize, hey, we've got a lot of different 0:22:59.560 --> 0:23:02.920 cloud elements. And so if you're storing your customer data 0:23:03.000 --> 0:23:06.360 with Salesforce and you've got these three other environments, how 0:23:06.359 --> 0:23:10.159 do you get the customer data to inform you know 0:23:10.200 --> 0:23:12.080 what you're doing in the other parts of your business. 0:23:12.200 --> 0:23:18.680 That's a hybrid cloud problem. And how how hard of 0:23:18.720 --> 0:23:21.960 a problem is that? I mean, as that total naive 0:23:22.000 --> 0:23:24.520 outside or I would have said, oh, surely all these 0:23:24.520 --> 0:23:26.760 cloud businesses would have made it really easy to share 0:23:27.280 --> 0:23:30.160 stuff in one place with stuff you've got in other place. 0:23:30.200 --> 0:23:33.920 Is that not true? Unfortunately the opposite is true, because 0:23:34.440 --> 0:23:37.480 for the pure play public cloud providers, the incentive was 0:23:37.520 --> 0:23:40.600 actually the opposite. It's Hotel California. For them, you can 0:23:40.640 --> 0:23:42.679 bring your stuff in, but you know you can. You 0:23:42.680 --> 0:23:44.600 can check in, but you will never let you check out, 0:23:45.520 --> 0:23:47.919 and they charge actually enormous fees if you want to 0:23:48.359 --> 0:23:50.800 get your data out. So it's a bit of a 0:23:50.840 --> 0:23:54.359 strategy tax for them to make it easy. It's also 0:23:54.400 --> 0:23:58.480 a hard problem just because you're trying to connect different 0:23:58.560 --> 0:24:02.159 data sets. Do you have in card cataloged that connects 0:24:02.200 --> 0:24:05.520 all these different sources. It's actually not easy to do. 0:24:06.200 --> 0:24:08.159 And what happens when you don't do that then you 0:24:08.240 --> 0:24:10.879 end up rebuilding everything and so suddenly you're storing all 0:24:10.920 --> 0:24:14.679 the same data five times. That gets very expensive. So 0:24:14.960 --> 0:24:21.320 let's imagine what Having this conversation give me your sense 0:24:21.359 --> 0:24:23.639 of where will be what would what would we be 0:24:23.720 --> 0:24:27.840 talking about five years from now, we'll probably be having 0:24:28.119 --> 0:24:31.680 very similar discussions as possible. Technology will be more advanced, 0:24:31.720 --> 0:24:34.480 but a lot of them probably talked about. Let's be honest, 0:24:34.520 --> 0:24:38.359 these have been around for for quite a while. There's 0:24:38.400 --> 0:24:41.480 a story this this guy, Charles Towns, he was the 0:24:41.520 --> 0:24:44.800 inventor of the laser, and he tells his story. There's 0:24:44.800 --> 0:24:47.040 a rabbit and a beaver and they're staring at the 0:24:47.080 --> 0:24:51.159 Hoover Dam and the beaver says the rabbit, no, I 0:24:51.160 --> 0:24:53.400 didn't build it, but it's based on an idea of mine. 0:24:55.080 --> 0:24:57.679 And the point of that story is there's ideas. Are 0:24:57.720 --> 0:25:03.280 a dime? Doesn't it so? Great? Story? Everybody's got a 0:25:03.280 --> 0:25:06.399 bunch of ideas. By the way, we're too quick to 0:25:06.440 --> 0:25:10.280 dismiss the beaver. He's right, but have you seen beaver 0:25:10.359 --> 0:25:13.080 Dam's I mean, I know he is right. It was 0:25:13.160 --> 0:25:15.520 his idea, but he had nothing to do with the 0:25:15.560 --> 0:25:22.040