1 00:00:00,320 --> 00:00:04,360 The Better Business Analysis Institute presence, the Better 2 00:00:04,360 --> 00:00:07,680 Business Analysis podcast. With Kingsman Walsh. 3 00:00:13,040 --> 00:00:16,600 Last week I had the pleasure of talking to Pankaj Zanki who has 4 00:00:16,600 --> 00:00:20,280 over 10 years experience in data and data analytics and we had a 5 00:00:20,280 --> 00:00:23,880 conversation about big data, AI and the insurance industry. 6 00:00:24,600 --> 00:00:27,600 Punkage has worked in many different roles like engineer, 7 00:00:27,800 --> 00:00:31,000 architect, project manager, technical BA for some big 8 00:00:31,000 --> 00:00:34,760 companies in the US like United Health, Home Depot, Progressive 9 00:00:34,760 --> 00:00:38,560 Insurance. Punkage also has a master's 10 00:00:38,560 --> 00:00:43,280 degree in data science and computational analysis from 11 00:00:43,280 --> 00:00:47,520 Georgia Institute of Technology. He's published quite a few 12 00:00:47,520 --> 00:00:51,440 scholarly papers and he currently has some patents 13 00:00:51,440 --> 00:00:53,760 pending, so it was a pleasure to talk to him. 14 00:00:54,400 --> 00:01:00,040 Let's take a listen. Hi, Ben, I'm good and you know, 15 00:01:00,040 --> 00:01:02,120 thank you for inviting me to your podcast. 16 00:01:03,320 --> 00:01:05,120 Not a problem. And sorry about the technical 17 00:01:05,120 --> 00:01:08,240 issues we had this morning. Just read out the bit of a 18 00:01:08,240 --> 00:01:10,440 background about you. You're an interesting character. 19 00:01:10,440 --> 00:01:12,800 Obviously you've come from a technical background, data 20 00:01:12,800 --> 00:01:16,040 background and you've also dabbled in business analysis 21 00:01:16,040 --> 00:01:21,000 and, and, and you give a unique perspective to the world of 22 00:01:21,240 --> 00:01:24,280 specifically big data and AI We're going to talk about today. 23 00:01:24,440 --> 00:01:28,880 My background is rooted in a strong foundation of education 24 00:01:28,880 --> 00:01:33,040 and experience in the data analytics and industry. 25 00:01:33,040 --> 00:01:38,640 I completed my bachelor's into information technology and which 26 00:01:38,640 --> 00:01:40,760 led the groundwork for my technical. 27 00:01:41,840 --> 00:01:45,080 Right now I'm pursuing my Master of Science in data, data Science 28 00:01:45,080 --> 00:01:48,520 analytics from Georgia University from USA, which is 29 00:01:48,520 --> 00:01:52,080 one of the best, top most university in the USA, which 30 00:01:52,080 --> 00:01:56,240 which strengthen my expertise in this rapidly evolving, you know, 31 00:01:56,800 --> 00:02:00,120 the data science industry. Professionally, I have played 32 00:02:00,120 --> 00:02:03,960 multiple roles throughout my decade of experience like data 33 00:02:03,960 --> 00:02:08,600 engineer, data analyst, solution architect, project manager and 34 00:02:08,639 --> 00:02:12,640 you know, I have work on multiple domains like you know, 35 00:02:12,640 --> 00:02:18,080 the retail, insurance, finance, healthcare. 36 00:02:18,680 --> 00:02:23,040 So I do have experience in bringing the insight from the 37 00:02:23,040 --> 00:02:25,600 big data. I have multiple scholarly 38 00:02:25,600 --> 00:02:29,480 articles and white papers are publishing multiple, you know, 39 00:02:29,480 --> 00:02:33,080 the Google Scholar index journals and the conferences, 40 00:02:33,920 --> 00:02:36,960 which is one of the best thing happened for me as well 41 00:02:37,520 --> 00:02:39,760 personally. I also, you know, as you know 42 00:02:39,760 --> 00:02:43,480 that like I also got invited multiple times on multiple 43 00:02:43,480 --> 00:02:47,520 podcasts and for a panel discussion how the big data and 44 00:02:47,600 --> 00:02:50,680 AI is impacting the different industries. 45 00:02:50,760 --> 00:02:54,280 So I'm very excited here to be a part of your podcast. 46 00:02:54,680 --> 00:02:56,320 No. And, and look, we're excited to 47 00:02:56,320 --> 00:02:59,360 have you. So how did you, why was it that 48 00:02:59,360 --> 00:03:03,000 you were captured by the, the world of data and analytics? 49 00:03:03,000 --> 00:03:05,600 What is it about that? What is it about data and 50 00:03:05,600 --> 00:03:08,480 analytic that gets you up in the morning and excited and, and 51 00:03:08,640 --> 00:03:11,720 gets you to go to university again, Do your masters write 52 00:03:11,720 --> 00:03:13,760 white papers? What is it about that particular 53 00:03:13,760 --> 00:03:16,160 area that you are so excited about? 54 00:03:16,800 --> 00:03:18,800 See, data is not about the numbers. 55 00:03:18,960 --> 00:03:23,080 It's a kind of tools, you know, the, you know, which will give 56 00:03:23,080 --> 00:03:26,520 you the, you know, the insights from the messy database. 57 00:03:26,800 --> 00:03:29,880 And that is one of the challenge you have to, you have to be very 58 00:03:29,880 --> 00:03:31,880 curious. Like when you work in the 59 00:03:32,000 --> 00:03:35,960 database industry or data analytics industry, you are, you 60 00:03:35,960 --> 00:03:38,280 should, you should be ready to ask the questions to the 61 00:03:38,280 --> 00:03:41,640 business. This will help the to build that 62 00:03:41,640 --> 00:03:44,440 business strategies to grow your business, to grow your 63 00:03:44,760 --> 00:03:46,600 individual characters as well, right? 64 00:03:46,920 --> 00:03:51,480 And that, that, that is why I'm thinking about like always, data 65 00:03:51,920 --> 00:03:54,840 database or data science industries are going very 66 00:03:54,840 --> 00:03:58,240 faster. If you have a good mindset as 67 00:03:58,240 --> 00:04:03,600 and when I, I work on, you know, the messy data that's, that's 68 00:04:03,600 --> 00:04:08,040 bring the best out of me. How I can we bring the the 69 00:04:08,080 --> 00:04:13,320 insights from this Messi database and Messi data and what 70 00:04:13,320 --> 00:04:16,600 are the different dashboard I can create throughout the 71 00:04:16,800 --> 00:04:21,560 different data source which is available and how I can help my 72 00:04:21,600 --> 00:04:27,280 clients to understand this data. So that is really these are the 73 00:04:27,280 --> 00:04:32,080 different things which has helped me to spark and to work 74 00:04:32,480 --> 00:04:35,320 on the data science. Yeah. 75 00:04:35,320 --> 00:04:36,920 OK. So you've got we talked about 76 00:04:36,920 --> 00:04:39,640 messy data and we'll probably get to that later on as a good 77 00:04:39,640 --> 00:04:41,400 example. Hopefully I don't forget about 78 00:04:41,400 --> 00:04:42,800 that. People talk about structured 79 00:04:42,800 --> 00:04:44,920 data and then people talk about unstructured data. 80 00:04:44,920 --> 00:04:46,360 I guess we're talking about messy data here. 81 00:04:46,560 --> 00:04:49,000 So we'll get to that in a minute because I, I don't think that 82 00:04:49,000 --> 00:04:52,320 the general person in business really understands the 83 00:04:52,320 --> 00:04:53,760 difference there. So we might get to that. 84 00:04:53,760 --> 00:05:01,040 From your perspective, what does big data and AI truly mean to 85 00:05:01,040 --> 00:05:03,040 business? Today we hear the buzzwords. 86 00:05:03,280 --> 00:05:07,240 What do those two words actually mean from your perspective? 87 00:05:07,800 --> 00:05:11,360 See the big big data is nothing but represent the vast amount of 88 00:05:11,360 --> 00:05:15,280 data which is getting generated from the different data sources. 89 00:05:15,880 --> 00:05:21,400 Right now we have social media, different transactions, sensors, 90 00:05:21,400 --> 00:05:23,240 right? Like there are multiple 91 00:05:23,240 --> 00:05:26,920 invoices, billing systems, legacy systems, the audio, audio 92 00:05:26,920 --> 00:05:30,440 data as well as the video data. All these datas are getting 93 00:05:30,440 --> 00:05:33,560 generated from the different data source and the companies 94 00:05:33,600 --> 00:05:35,800 can now collect and analyze this data. 95 00:05:35,880 --> 00:05:39,240 This data when processed effectively offer the deep 96 00:05:39,240 --> 00:05:42,320 insight into the customer behavior, market trends, you 97 00:05:42,320 --> 00:05:45,520 know the what are the different Flyers we should provide or a 98 00:05:45,520 --> 00:05:48,840 promo code we should provide to the customers so that our retail 99 00:05:48,840 --> 00:05:52,640 business can grow more than expected. 100 00:05:52,640 --> 00:05:57,480 Right out out of this big data companies are more workings 101 00:05:57,480 --> 00:06:01,280 toward the predictive mindset and they are started bringing 102 00:06:01,280 --> 00:06:04,120 the insight from the different data sources available. 103 00:06:04,520 --> 00:06:08,080 If you think about the AI right, AI is nothing but the educated 104 00:06:08,480 --> 00:06:11,160 computer system which which is held to automate all this 105 00:06:11,160 --> 00:06:14,960 process, to analyse your different data points, to build 106 00:06:14,960 --> 00:06:18,600 the algorithms and machine learning techniques to to to 107 00:06:18,600 --> 00:06:22,120 interpret your data. AI system can automate the 108 00:06:22,120 --> 00:06:25,520 complex process and also to provide the real time real 109 00:06:25,520 --> 00:06:28,960 insights from your available data sources. 110 00:06:30,080 --> 00:06:32,560 Right. And so without the big data, 111 00:06:32,560 --> 00:06:35,600 without what we just described, what you just described as big 112 00:06:35,600 --> 00:06:38,680 data, which is bringing out all the different channels of data 113 00:06:38,680 --> 00:06:41,360 you've got sources, whatever it is without that. 114 00:06:42,440 --> 00:06:46,720 Is that a prerequisite for AI to be successful in a business? 115 00:06:48,560 --> 00:06:51,520 Yeah, definitely. Like, you know, if you really 116 00:06:51,520 --> 00:06:55,520 want the AI to successfully in the business, right, we should 117 00:06:55,520 --> 00:06:59,440 have that mindset in our in, in our leadership and adds at the 118 00:06:59,440 --> 00:07:03,200 same time, we also have to train our resources to most and to add 119 00:07:03,200 --> 00:07:05,880 up to the AI, AI technologies. AI. 120 00:07:05,880 --> 00:07:09,240 As I mentioned, AI is nothing but you know, the educated 121 00:07:09,240 --> 00:07:13,120 computer system where you can train them with the available 122 00:07:13,120 --> 00:07:17,120 data to from your traditional database to your cloud base and 123 00:07:17,120 --> 00:07:20,960 your big database. So if you were to explain big 124 00:07:20,960 --> 00:07:24,880 data to your grammar, what why is it different to maybe how 125 00:07:25,040 --> 00:07:26,920 organizations are using their data today? 126 00:07:26,920 --> 00:07:29,480 So I'll give you an example. I'm working for one organization 127 00:07:29,480 --> 00:07:34,000 or one of my clients has a data warehouse, OK, it has part of 128 00:07:34,000 --> 00:07:35,960 their information. They've been on the journey of 129 00:07:35,960 --> 00:07:41,560 creating this data warehouse for years and years and years still 130 00:07:41,560 --> 00:07:46,880 doesn't completely they encapsulate every single data or 131 00:07:46,880 --> 00:07:49,880 the perfect data model for their organization, right? 132 00:07:49,880 --> 00:07:51,640 And they're a large client, they've got a lot of money. 133 00:07:52,000 --> 00:07:55,560 So why is big data different to traditional data? 134 00:07:56,920 --> 00:08:01,600 Yeah, sure. You mentioned about the you know 135 00:08:01,600 --> 00:08:04,720 to to explain the big data to a grandmother, right? 136 00:08:04,720 --> 00:08:08,400 So, so imagine you have a big jar of candies with all kind of 137 00:08:08,400 --> 00:08:10,400 flavors and colors mixed together, right? 138 00:08:11,000 --> 00:08:15,080 Traditional data would be like having a just small jar with 139 00:08:15,080 --> 00:08:18,320 only a few candies. It is easy to count and see what 140 00:08:18,320 --> 00:08:20,680 you have in that small, small jar, right? 141 00:08:21,200 --> 00:08:25,520 But now think about the big data having a giant jar filled with 142 00:08:25,520 --> 00:08:29,880 thousands of candies all mix, you know, all, all all mixed up. 143 00:08:30,240 --> 00:08:33,360 You have candies of many different flavours, colour, 144 00:08:33,360 --> 00:08:37,159 shape and they come from the various places as well. 145 00:08:37,240 --> 00:08:40,320 To to understand what kind of candies are in that jar, right? 146 00:08:40,559 --> 00:08:44,760 You need to sort sort them through very carefully, right? 147 00:08:45,040 --> 00:08:49,200 To use a special tool to figure out things like which flavours 148 00:08:49,200 --> 00:08:51,840 are most common and which colours are most common. 149 00:08:51,840 --> 00:08:53,840 What are the different sizes are available? 150 00:08:54,360 --> 00:08:57,000 What are the different brands are available for that candies? 151 00:08:57,160 --> 00:09:00,520 To divide all these things you need a specific tool. 152 00:09:00,520 --> 00:09:03,560 So that is how the big data come into the picture. 153 00:09:03,560 --> 00:09:06,440 Like you know the big data is different because it is a much, 154 00:09:06,520 --> 00:09:10,880 much bigger and more complex. Instead of just a few few pieces 155 00:09:10,880 --> 00:09:14,560 of information, you have huge amount of data from many 156 00:09:14,560 --> 00:09:17,320 different sources. Just like you know, sorting to 157 00:09:17,320 --> 00:09:21,280 the giant car of candies. Businesses use the special tools 158 00:09:21,280 --> 00:09:26,000 and method to organize and to understand this large amount of 159 00:09:26,000 --> 00:09:27,960 data to make the better decisions. 160 00:09:28,720 --> 00:09:30,480 Yeah. So those same techniques that 161 00:09:30,480 --> 00:09:34,520 work in your small jar of candy are not going to work for your 162 00:09:34,520 --> 00:09:37,080 large jar of candy. You need to start to evolve and 163 00:09:37,080 --> 00:09:40,040 start to work in a different way on a much bigger scale. 164 00:09:40,480 --> 00:09:44,280 What are some of the ways in which big data has helped the 165 00:09:44,280 --> 00:09:47,520 insurance industry? A great example of how the big 166 00:09:47,520 --> 00:09:52,200 data and AI is getting used in the insurance industry is it it 167 00:09:52,200 --> 00:09:55,280 is in the area of, you know, the personalized insurance pricing. 168 00:09:55,840 --> 00:09:59,000 Traditionally, insurance companies are mostly considered 169 00:09:59,000 --> 00:10:01,600 the age, locations and driving history. 170 00:10:01,760 --> 00:10:05,480 But however, with the big data insurance now collect and 171 00:10:05,480 --> 00:10:09,640 analyze a much larger data set that includes your driving 172 00:10:09,640 --> 00:10:13,520 patterns, you're visiting new places, age locations, right? 173 00:10:13,880 --> 00:10:17,360 And all this is happening like you know, for some instance like 174 00:10:17,360 --> 00:10:21,520 some insurance companies are started in installing telematic 175 00:10:21,520 --> 00:10:26,200 device in your car that that helps them to get the real time 176 00:10:26,200 --> 00:10:29,000 real time data from your driving patterns. 177 00:10:29,000 --> 00:10:34,040 And based on that your insurance premiums are getting decided. 178 00:10:34,120 --> 00:10:37,240 So this this approach not only reward the safe drivers 179 00:10:37,280 --> 00:10:40,800 behaviour, but also help the insurance to better better 180 00:10:40,800 --> 00:10:44,680 predict risk and you know, to reduce the losses. 181 00:10:45,320 --> 00:10:48,640 I think some people feel a bit scared around the fact that 182 00:10:48,840 --> 00:10:51,200 we've talked about social media before and the fact that, you 183 00:10:51,200 --> 00:10:54,560 know, to be honest, Mark Zuckerberg better, if they want 184 00:10:54,560 --> 00:10:57,320 to predict what you do next, they can probably do that now 185 00:10:57,320 --> 00:10:59,120 with the data they've got. And we've talked about that and 186 00:10:59,280 --> 00:11:01,720 how scary that is. But the fact that the insurance 187 00:11:01,720 --> 00:11:05,120 company in a real practical way, using this to decide on your 188 00:11:05,120 --> 00:11:09,120 premium levels and risk and what might be a bad driver might what 189 00:11:09,120 --> 00:11:12,760 might be a good driver, what is the benefit for the customer. 190 00:11:13,000 --> 00:11:15,800 One thing is coming to my mind is like the personalized 191 00:11:15,800 --> 00:11:19,240 pricing, insurance companies use the big data to analyse the 192 00:11:19,240 --> 00:11:22,400 detailed customer informations and behaviour for for example, 193 00:11:22,400 --> 00:11:26,240 as I mentioned, the few of the insurance companies started 194 00:11:26,520 --> 00:11:29,720 installing the, you know, telematic, telematic devices 195 00:11:29,720 --> 00:11:32,760 into your car. So the telematic data from the 196 00:11:32,760 --> 00:11:38,000 vehicle insurance can access and drive, you know, to to decide 197 00:11:38,000 --> 00:11:41,120 your premium that that help the customer. 198 00:11:41,760 --> 00:11:45,680 You know, at the same time, big data enabled the insurance to 199 00:11:45,960 --> 00:11:50,120 refine their risk model by, you know, incorporating the wide 200 00:11:50,120 --> 00:11:53,760 range of data sources. Also, you know, the big data 201 00:11:53,760 --> 00:11:57,280 helping them to, to detect the, to detect the fraud into the 202 00:11:57,280 --> 00:12:01,520 claim processing for customers. They, they try to, you know, 203 00:12:02,280 --> 00:12:05,400 submit the fraud claim. So based on the customer 204 00:12:05,400 --> 00:12:09,800 behaviour, their historical pattern, the claims history, the 205 00:12:09,800 --> 00:12:13,440 insurance company can use that data and detect the fraud. 206 00:12:13,800 --> 00:12:16,960 You know the big data also helping the insurance company to 207 00:12:17,240 --> 00:12:21,160 enhance the customer services to take their real time, real time 208 00:12:21,160 --> 00:12:23,280 feedback about the different sources, right? 209 00:12:23,760 --> 00:12:26,680 Like what are the what are the different products the insurance 210 00:12:26,680 --> 00:12:31,760 company should build and how they can how how the how all 211 00:12:31,760 --> 00:12:34,720 these things can help them in a claim processing faster claim 212 00:12:34,720 --> 00:12:36,880 processing. There are a couple of devices 213 00:12:36,880 --> 00:12:40,600 and in fact, now one of my patent is also got approved, 214 00:12:40,880 --> 00:12:42,840 which is related to a claim processing. 215 00:12:42,880 --> 00:12:45,680 I have the that device patent with like you know it is AVR 216 00:12:45,680 --> 00:12:49,200 device which will collect all your data from your local 217 00:12:49,280 --> 00:12:52,840 locations and that will get tied to your policies and claim and 218 00:12:52,840 --> 00:12:56,960 that will help to to process your claim faster than expected. 219 00:12:57,320 --> 00:13:00,680 AI technologies and the big data technologies is helping 220 00:13:00,680 --> 00:13:04,520 insurance companies to automate the routine task. 221 00:13:04,560 --> 00:13:06,840 So we talked about the fact that insurance is probably, if you 222 00:13:06,840 --> 00:13:09,560 talked about a spectrum, insurance is probably right on 223 00:13:09,560 --> 00:13:12,280 the progressive end, right? That started to use big data for 224 00:13:12,280 --> 00:13:14,320 everything now, right? They've started to leverage it 225 00:13:14,320 --> 00:13:17,000 for everything like you said, not just to manage the risk or 226 00:13:17,000 --> 00:13:19,120 work out of your bad driver. They should pay out. 227 00:13:19,280 --> 00:13:21,800 But also now in terms of what products they should create, 228 00:13:22,040 --> 00:13:25,680 you've just mentioned that you've got a patent for VRT, 229 00:13:25,680 --> 00:13:28,560 which is around collecting that data faster. 230 00:13:28,800 --> 00:13:31,280 So almost, I guess, if you're dealing with an insurance 231 00:13:31,280 --> 00:13:34,120 company who's in this space, it's probably in your best 232 00:13:34,120 --> 00:13:36,680 interest to share information with them. 233 00:13:37,200 --> 00:13:40,200 Would that be your assessments that you're better off allowing 234 00:13:40,200 --> 00:13:43,800 an insurance company deep into your data pocket? 235 00:13:44,000 --> 00:13:47,080 They're going to make predictions about you based on 236 00:13:47,080 --> 00:13:51,520 the data they've got. Yes, that is absolutely true. 237 00:13:52,040 --> 00:13:56,760 This as I mentioned all these new technique data, database 238 00:13:56,760 --> 00:14:00,240 data sources, right, They are helping the insurance company to 239 00:14:00,240 --> 00:14:02,160 decide on the new product developments. 240 00:14:02,160 --> 00:14:05,520 And there are most of the companies are working on the AI 241 00:14:05,520 --> 00:14:12,120 models like how they can build the AI model for from policy 242 00:14:12,120 --> 00:14:15,360 creation to claim creation to underwriting process. 243 00:14:15,520 --> 00:14:17,720 It's quite, it's quite scary. I think for some people. 244 00:14:17,720 --> 00:14:19,760 I think they'll come away and think that's scary, but the 245 00:14:19,760 --> 00:14:21,320 reality is it's happening right now. 246 00:14:21,320 --> 00:14:22,680 It's not something you can ignore. 247 00:14:22,680 --> 00:14:25,240 It's going to be part of your life and other industries are 248 00:14:25,240 --> 00:14:26,720 going to catch on and use that model. 249 00:14:26,920 --> 00:14:30,080 So banking I guess is quite a traditional, but I imagine 250 00:14:30,080 --> 00:14:33,800 banking will start to get into a very similar example where they 251 00:14:33,800 --> 00:14:38,680 can probably predict investments and they can predict which which 252 00:14:38,680 --> 00:14:41,720 products are going to be more attractive to a customer. 253 00:14:41,880 --> 00:14:46,160 Yeah, sure. Now, AI is often misunderstood 254 00:14:46,160 --> 00:14:49,040 and you talked about the the fact that AI is just really, you 255 00:14:49,040 --> 00:14:53,240 know, this engine if you like, but could you break up some of 256 00:14:53,240 --> 00:14:56,000 the components maybe from starting with big data? 257 00:14:56,000 --> 00:14:59,800 So let's just assume we've got our data connected in some way 258 00:15:00,080 --> 00:15:04,720 down to literally maybe a chat B, GB T or a kind of a chat bot. 259 00:15:05,000 --> 00:15:08,840 What are what are the different stages or components that data 260 00:15:08,840 --> 00:15:13,160 moves through in order to get it from the big data set or 261 00:15:13,160 --> 00:15:16,400 unstructured data to useful in terms of AI? 262 00:15:16,440 --> 00:15:18,240 Yeah. So you know, like as I 263 00:15:18,240 --> 00:15:22,840 mentioned, the the big data is like a vast amount of data which 264 00:15:22,840 --> 00:15:25,080 is getting created from the different data source. 265 00:15:25,360 --> 00:15:29,160 And AI is about creating, you know, the computer system that 266 00:15:29,160 --> 00:15:32,320 can perform the task usually require the human intelligence, 267 00:15:32,320 --> 00:15:34,160 right. Think of it is a teaching a 268 00:15:34,160 --> 00:15:38,520 computer to to think and make the decision like like a human 269 00:15:38,520 --> 00:15:41,240 word, right. So the other other factor of it 270 00:15:41,240 --> 00:15:44,400 like the you know, the machine learning, this is a key part of 271 00:15:44,440 --> 00:15:46,640 AI. We are confident learn from the 272 00:15:46,640 --> 00:15:49,880 different data. Just imagine teaching a teaching 273 00:15:49,880 --> 00:15:54,360 a child about the fruits as and when like we we teach the child 274 00:15:54,360 --> 00:15:57,200 about the fruits we show them a lot of a lot of different types 275 00:15:57,200 --> 00:15:58,840 of fruits, different types of picture. 276 00:15:59,160 --> 00:16:02,600 We, we, we put all the data inside his mind. 277 00:16:02,600 --> 00:16:05,520 OK, this is apple. This is banana, right? 278 00:16:05,960 --> 00:16:10,280 With with the real food and real pictures, right? 279 00:16:10,280 --> 00:16:12,200 So that's how he started understanding. 280 00:16:12,200 --> 00:16:15,920 So this is what the machine learning will do with the AI 281 00:16:15,920 --> 00:16:20,120 models. They will help the computer to 282 00:16:20,120 --> 00:16:24,520 learn from the available datas, images, videos, links, right? 283 00:16:24,520 --> 00:16:29,080 The audio files, yes. And there is also part of it 284 00:16:29,080 --> 00:16:31,240 like a natural language processing, right? 285 00:16:31,760 --> 00:16:35,280 Yes. So which is mean by NLP help the 286 00:16:35,280 --> 00:16:39,040 computer understand and you know to interpret with the human 287 00:16:39,040 --> 00:16:41,240 language. For example, you know when you 288 00:16:41,240 --> 00:16:45,480 talk to a Voice Assistant device like Siri or Alexa, right? 289 00:16:45,960 --> 00:16:48,640 NLP is what allow you to understand your words and 290 00:16:48,640 --> 00:16:52,000 response rapidly, right. And there is also a part of 291 00:16:52,440 --> 00:16:55,760 algorithm. So in a simple word, algorithm 292 00:16:55,760 --> 00:16:59,640 is nothing but the recipe. So when you consider when you 293 00:16:59,640 --> 00:17:03,320 are trying to to make the cake right, there is a particular 294 00:17:03,440 --> 00:17:05,680 recipes, you have to follow these steps, right? 295 00:17:05,760 --> 00:17:08,000 Same things happened in the algorithm. 296 00:17:08,000 --> 00:17:11,960 It is a set of instruction that guide the computer on how to 297 00:17:11,960 --> 00:17:16,599 learn from data and to make the to make the correct, correct 298 00:17:16,599 --> 00:17:20,680 decisions. Chart GPT and as well as your 299 00:17:20,680 --> 00:17:24,599 Gemini chart boss. Those are really the next 300 00:17:24,599 --> 00:17:28,400 generations AI, I would say, which is really helping the 301 00:17:28,400 --> 00:17:33,040 customers to build a strong model about the AI, understand 302 00:17:33,040 --> 00:17:36,560 about the AI, how AI can help them in a different direction 303 00:17:36,560 --> 00:17:40,200 like the chart GPT, right? Yes, you can easily get a lot of 304 00:17:40,200 --> 00:17:45,360 answers build the code base model using the ChatGPT. 305 00:17:46,240 --> 00:17:50,400 When we when we think about chat bot right like this is this is 306 00:17:50,400 --> 00:17:53,200 also getting used in the claim processing. 307 00:17:53,440 --> 00:17:58,280 The chat bot is helping all the all the customer to get the to 308 00:17:58,280 --> 00:18:03,400 get the data quickly and to to submit their claims as soon as 309 00:18:03,400 --> 00:18:07,120 fast as they can. So we've got the big data and 310 00:18:07,120 --> 00:18:09,280 then we've got the machine learning, which is almost 311 00:18:09,440 --> 00:18:11,960 protective and pattern matching, things like that. 312 00:18:11,960 --> 00:18:14,720 Then you've got the natural language side, which is 313 00:18:14,800 --> 00:18:18,920 converting what we're asking the AI to an understandable language 314 00:18:18,920 --> 00:18:20,920 in which the computer understands effectively and back 315 00:18:20,920 --> 00:18:22,600 and forth. And then you've got the 316 00:18:22,600 --> 00:18:25,800 algorithm and, and the combination of that as well as, 317 00:18:25,800 --> 00:18:28,040 sorry, the interface, which is the chatbot itself. 318 00:18:28,320 --> 00:18:30,920 So those elements, there are actually a number of elements, 319 00:18:30,920 --> 00:18:33,040 right? I think people just think AI is, 320 00:18:33,760 --> 00:18:35,880 there's all of those things. And my understanding is that 321 00:18:36,000 --> 00:18:39,000 each one of these areas have evolved over the years and now 322 00:18:39,000 --> 00:18:41,840 the combination of them are starting to come come out as 323 00:18:41,840 --> 00:18:46,600 what we see as a product like ChatGPT or, or or or copilot or 324 00:18:46,720 --> 00:18:51,000 like you said, Gemini, how do you see AI and human analysts? 325 00:18:51,080 --> 00:18:52,400 This is going to be a big question for BAS. 326 00:18:53,400 --> 00:18:57,480 Human analysts like BAS working together to drive bitter 327 00:18:57,480 --> 00:19:01,440 insights. According to me, AI is helping a 328 00:19:01,440 --> 00:19:05,000 big time for human analyst or a business analyst. 329 00:19:05,120 --> 00:19:09,400 The AI and human analyst working together can really boost how we 330 00:19:09,400 --> 00:19:13,240 can use the data. AI is, you know, great at 331 00:19:13,240 --> 00:19:16,720 shifting to the huge amount of data quickly and, you know, 332 00:19:16,720 --> 00:19:20,880 spotting the pattern and the human analyst, they can, you 333 00:19:20,880 --> 00:19:24,880 know, make the sense out of it. Like, according to me, we should 334 00:19:24,880 --> 00:19:29,440 let the AI to, to do the heavy lifting to bring the data from 335 00:19:29,440 --> 00:19:33,400 the different data source, to clean it and to bring the, to 336 00:19:33,400 --> 00:19:36,680 bring the specific pattern, to identify the pattern between the 337 00:19:37,480 --> 00:19:41,240 available data and, you know, to build, to build the dashboard. 338 00:19:41,400 --> 00:19:44,800 And at the, at the same time, human analyst or business 339 00:19:44,800 --> 00:19:47,320 analyst, they should work on the business requirements. 340 00:19:47,360 --> 00:19:50,360 What, what are the different things we should come out of 341 00:19:50,360 --> 00:19:53,760 this dashboard? Like what story we should have 342 00:19:53,760 --> 00:19:57,600 to build out of this dashboard? And what are the different, you 343 00:19:57,600 --> 00:20:01,480 know, patterns we are able to see or how how to make it a? 344 00:20:02,200 --> 00:20:05,480 Sensible patterns, right? These are the different stories 345 00:20:05,480 --> 00:20:08,760 the human analyst can build out of this big data which is 346 00:20:08,760 --> 00:20:11,280 available. Yeah, it's quite interesting. 347 00:20:11,280 --> 00:20:13,880 So because people say it as a replacement of some of the 348 00:20:13,880 --> 00:20:17,560 activities in ABA, whereas ABA should really be focused on the 349 00:20:17,560 --> 00:20:19,680 business and the analysts. Like you said, it's the heavy 350 00:20:19,680 --> 00:20:21,560 listing. Yeah, the heavy listing side can 351 00:20:21,560 --> 00:20:23,600 be done by AI. It's about knowing the right 352 00:20:23,600 --> 00:20:26,320 questions to ask. Also, like you said, what what 353 00:20:26,320 --> 00:20:28,400 are the requirements, what what are you trying to achieve from 354 00:20:28,400 --> 00:20:31,000 the data that's correct. From your perspective, what do 355 00:20:31,000 --> 00:20:34,640 you think the key the ingredients are for building a 356 00:20:34,640 --> 00:20:38,960 successful data-driven culture? As and when we try to build a 357 00:20:38,960 --> 00:20:41,880 successful data-driven culture within an organization, 358 00:20:42,240 --> 00:20:45,040 according to me, it should start with the leadership support. 359 00:20:45,240 --> 00:20:47,880 All right, without your leadership support, your 360 00:20:48,280 --> 00:20:51,200 employees cannot change their mindset into a data-driven 361 00:20:51,200 --> 00:20:55,000 culture with the set of tones for importance for a data in a 362 00:20:55,000 --> 00:20:58,000 decision making. It is it is really essential to 363 00:20:58,000 --> 00:21:02,720 have a clear vision and specific goal for how data will be used 364 00:21:02,840 --> 00:21:06,240 to drive an organization organizational level strategies. 365 00:21:06,440 --> 00:21:09,120 So according to me, our organizations and the 366 00:21:09,120 --> 00:21:11,640 leadership, they should arrange some training, training 367 00:21:11,640 --> 00:21:14,680 activities for their resources. They should train them, they 368 00:21:14,680 --> 00:21:18,440 should provide the available materials and they should also 369 00:21:18,440 --> 00:21:22,880 explain what is the, what is our mission and objectives as a part 370 00:21:22,880 --> 00:21:26,680 of this data-driven process. And at the at the same time, 371 00:21:26,680 --> 00:21:30,560 they, they should start some recognition, right, like giving 372 00:21:30,560 --> 00:21:35,280 some awards to the employees as and when they bring some kind of 373 00:21:35,280 --> 00:21:39,400 big, big time insights from from the available data. 374 00:21:39,400 --> 00:21:41,880 So that will that will definitely motivate all the 375 00:21:41,880 --> 00:21:44,240 other resources as well. Yeah. 376 00:21:44,240 --> 00:21:48,120 So that's providing incentives effectively to learn this new 377 00:21:48,120 --> 00:21:49,600 pattern. Yeah, I think that's a really 378 00:21:49,600 --> 00:21:52,360 good idea. So I'm going to ask a side 379 00:21:52,440 --> 00:21:54,760 question here because I have been involved in a couple of 380 00:21:54,760 --> 00:21:58,360 executive teams. We are the board and this is 381 00:21:58,360 --> 00:22:00,160 probably happening all around the world right now. 382 00:22:00,560 --> 00:22:03,400 There's probably board members meeting or an executive team 383 00:22:03,400 --> 00:22:06,200 meeting going. We need to get into AOI like 384 00:22:07,120 --> 00:22:08,760 every our competitors are getting into AOI. 385 00:22:08,760 --> 00:22:12,720 They're not sure what AOI is. They're like, we must embrace 386 00:22:12,720 --> 00:22:15,320 AI. And then that may be the just 387 00:22:15,520 --> 00:22:17,720 that's the only directive they've given their 388 00:22:17,720 --> 00:22:20,160 organization. What do you think that's a 389 00:22:20,160 --> 00:22:22,320 mistake? And, and you talked about before 390 00:22:22,640 --> 00:22:25,160 having a very specific mission vision. 391 00:22:25,720 --> 00:22:29,400 Do you think it's enough to just say get into AI, or do you think 392 00:22:29,400 --> 00:22:31,880 you need to be really clear about what you want to achieve 393 00:22:31,880 --> 00:22:34,760 like any other project? Yeah. 394 00:22:34,760 --> 00:22:38,440 I think as I mentioned before, there should be a very specific 395 00:22:38,440 --> 00:22:43,400 objectives why we are moving to the AI, What are our end goal? 396 00:22:44,120 --> 00:22:47,080 Is it our end goal to make our product more, you know, the 397 00:22:47,080 --> 00:22:50,960 client specific or should we have to, you know, the bring 398 00:22:50,960 --> 00:22:54,680 more clients to our organization as and when like you know, also 399 00:22:54,680 --> 00:22:57,880 there are a couple of companies they started building the AI 400 00:22:57,880 --> 00:23:02,120 considering the future, like if they build the AI then they will 401 00:23:02,120 --> 00:23:05,080 have more clients. And there are there are other 402 00:23:05,080 --> 00:23:08,520 types of companies like they have already, you know, some 403 00:23:08,520 --> 00:23:11,400 requests from the clients like to automate this process using 404 00:23:11,400 --> 00:23:14,440 the AI technique. So that's how some, some of the 405 00:23:14,440 --> 00:23:20,920 companies moving toward the AI, but but all the businesses and 406 00:23:20,920 --> 00:23:23,680 companies as well as their leadership, they should have 407 00:23:23,680 --> 00:23:27,720 very clear vision and mission why they want to start using the 408 00:23:27,720 --> 00:23:32,960 big data or AI and what are the different factors how we can 409 00:23:32,960 --> 00:23:36,640 train our resources, what might be our roadblocks. 410 00:23:36,760 --> 00:23:40,920 Because AI and big data, if you, if you are moving toward that 411 00:23:41,240 --> 00:23:47,080 side, then initially, initially there will be some, some XXX 412 00:23:47,080 --> 00:23:50,640 amount of, you know, expenses you have to do to train your 413 00:23:50,640 --> 00:23:53,760 resources to, to buy some new, maybe new tools, new 414 00:23:53,760 --> 00:23:55,680 technologies. Yeah, 100%. 415 00:23:55,680 --> 00:23:58,080 And I guess that's some of the biggest obstacles is probably 416 00:23:58,080 --> 00:24:01,840 that initial investment that you're then paying off along the 417 00:24:01,840 --> 00:24:05,040 way. What in terms of the future of 418 00:24:05,040 --> 00:24:07,560 big data and AI, what are you most excited about? 419 00:24:07,560 --> 00:24:10,640 What do you think the new emerging technologies will look 420 00:24:10,640 --> 00:24:13,080 like? I'm I'm very excited about the 421 00:24:14,280 --> 00:24:16,960 explainable AI, which is called XAI. 422 00:24:17,040 --> 00:24:20,360 As the AI system become more complex, there is a growing need 423 00:24:20,360 --> 00:24:23,760 of transparency in the how they make the decision. 424 00:24:23,760 --> 00:24:29,440 So XAI is more focused on making the AI model more understandable 425 00:24:29,440 --> 00:24:34,040 and more transparent. This technology will help to 426 00:24:34,040 --> 00:24:37,920 bridge the gap between the AI and powerful capabilities and 427 00:24:37,920 --> 00:24:40,680 the need for human to oversight the accountability. 428 00:24:41,080 --> 00:24:43,880 At the same time, I'm also excited about the quantum 429 00:24:43,880 --> 00:24:49,200 computing, which is also on the horizon as the game changer for 430 00:24:49,200 --> 00:24:52,560 a big data in AI. I think with the with the 431 00:24:52,560 --> 00:24:56,480 quantum computing right and it is ability to process the worst 432 00:24:56,480 --> 00:25:02,640 amount of data at unpredicted speed that will that will help 433 00:25:02,920 --> 00:25:08,800 all the businesses right to get the real time feedback faster 434 00:25:08,800 --> 00:25:11,560 than expected. I think the future of big data 435 00:25:11,560 --> 00:25:15,840 and AI is moving towards more intelligence, transparent and 436 00:25:16,120 --> 00:25:20,440 efficient system where this XAI technique right which I have 437 00:25:20,440 --> 00:25:24,840 mentioned explainable AI which will help to bring the gap, 438 00:25:24,920 --> 00:25:28,480 which will help to reduce the gap between the trust between 439 00:25:28,480 --> 00:25:31,600 the customer and the companies. Because I think there's a lot of 440 00:25:31,600 --> 00:25:35,560 concern around ethics in AI and like you said, transparency. 441 00:25:35,560 --> 00:25:39,000 So that sounds like an exciting thing for us to watch out for. 442 00:25:39,640 --> 00:25:42,440 Do you, in terms of young professionals and BA's data 443 00:25:42,440 --> 00:25:45,880 analysts, are there like some top five skills that they should 444 00:25:45,880 --> 00:25:49,200 be learning when it comes to AI and big data? 445 00:25:50,320 --> 00:25:53,800 I would start with the very basic, which is their domain 446 00:25:53,800 --> 00:25:57,160 knowledge, right? So as and when they start with 447 00:25:57,320 --> 00:26:01,040 any, any data project, they should have the domain knowledge 448 00:26:01,040 --> 00:26:03,160 about that. Like you know how the insurance 449 00:26:03,160 --> 00:26:05,760 data looks like, how the retail data looks like, what are the 450 00:26:06,160 --> 00:26:08,760 different factors for banking data, right? 451 00:26:08,920 --> 00:26:12,720 So to gain the to that, that kind of knowledge you will get 452 00:26:12,720 --> 00:26:15,880 when as and when you more study about that particular domain. 453 00:26:16,440 --> 00:26:18,360 There are lot of data literacies, right? 454 00:26:18,360 --> 00:26:22,880 Like what kind of datas comes out of, you know, the, the, the 455 00:26:23,040 --> 00:26:26,400 insurance claim for auto as well As for your personal claim, 456 00:26:26,800 --> 00:26:29,280 right? There are different types of 457 00:26:29,280 --> 00:26:32,320 claims, right? So we have to understand all 458 00:26:32,320 --> 00:26:35,320 those claims. Also at the same time I would 459 00:26:35,320 --> 00:26:39,640 suggest that they should stick to one programming language, 460 00:26:39,640 --> 00:26:44,640 maybe Python or R which is commonly used throughout the all 461 00:26:44,640 --> 00:26:47,560 big data and AI technologies. They should have one 462 00:26:47,680 --> 00:26:51,600 visualization tool, Power BI or Tablou or any kind of 463 00:26:51,600 --> 00:26:55,360 visualization tool. Or maybe Excel is also fine, but 464 00:26:55,480 --> 00:27:00,680 they should have the curiosity and to learn more about the data 465 00:27:01,360 --> 00:27:06,040 to bring the inside of the data. And they also understand like 466 00:27:06,040 --> 00:27:08,880 how we can move the data from one source to other source. 467 00:27:08,880 --> 00:27:12,360 So they should understand the ETL process, right? 468 00:27:12,600 --> 00:27:14,800 What are the different? What are the different ETL 469 00:27:14,800 --> 00:27:17,480 processes are there? Yeah, extract, transform and 470 00:27:17,480 --> 00:27:20,840 load, right? I also I will also suggest them 471 00:27:20,840 --> 00:27:24,520 to focus on the communication skill because you are going to 472 00:27:24,520 --> 00:27:27,440 build your dashboard right out of messy data. 473 00:27:27,800 --> 00:27:30,960 But at the at the end you have to explain that dashboard to 474 00:27:30,960 --> 00:27:34,240 your leadership team. You are C level executive so you 475 00:27:34,240 --> 00:27:38,520 should have that communication skill to be at your top like to 476 00:27:38,840 --> 00:27:41,520 to convince your leadership. OK, this is my dashboard and 477 00:27:41,520 --> 00:27:44,400 this is the pattern I understand and you should go with this 478 00:27:44,400 --> 00:27:46,120 route. Because you're focused on domain 479 00:27:46,120 --> 00:27:48,800 knowledge, which is business, you're focused on explaining to 480 00:27:48,800 --> 00:27:50,160 the executive. That's business. 481 00:27:50,280 --> 00:27:53,480 You've talked about some tools that you can use to talk to the 482 00:27:53,480 --> 00:27:56,160 data, but one of the things you didn't talk about, which I found 483 00:27:56,160 --> 00:27:59,880 interesting, was actually, you know, nitty gritty kind of the 484 00:27:59,880 --> 00:28:01,640 technical side. What you've actually focused on 485 00:28:01,640 --> 00:28:04,600 more is more the business side than the nitty gritty technical 486 00:28:04,600 --> 00:28:06,880 side, which I like because I think that's really important 487 00:28:06,880 --> 00:28:09,160 and that's where BA's generally play. 488 00:28:09,400 --> 00:28:12,880 But the value seems to be using that information to make 489 00:28:12,880 --> 00:28:15,720 business decisions. Why I'm more putting a more 490 00:28:15,720 --> 00:28:18,440 weightage on the business side or a domain knowledge because 491 00:28:18,440 --> 00:28:20,560 you know the technology will come and go. 492 00:28:20,800 --> 00:28:24,760 There will be today there will be a Python or tomorrow maybe C# 493 00:28:24,760 --> 00:28:27,360 or Java, right? Some of the companies, some of 494 00:28:27,360 --> 00:28:29,840 the companies is more comfortable with the Java, some 495 00:28:29,840 --> 00:28:32,880 of the companies are more comfortable with Python today. 496 00:28:33,000 --> 00:28:37,720 Some companies use Power BI, some companies use the Tableau. 497 00:28:37,920 --> 00:28:40,240 So these are the different tools and technologies that are 498 00:28:40,240 --> 00:28:43,440 available in the market. But at the end, we should have 499 00:28:43,440 --> 00:28:47,080 our analytic basics, very strong data reading techniques should 500 00:28:47,080 --> 00:28:49,040 be very strong to understand the data. 501 00:28:49,360 --> 00:28:51,880 That should be very strong. Even though if you don't know 502 00:28:51,880 --> 00:28:57,240 about the Python or R right, you can easily use the SQL or PL. 503 00:28:57,240 --> 00:29:03,280 SQL or Oracle and you can bring your data to your visualization 504 00:29:03,280 --> 00:29:05,520 tool. Thank you so much for your time. 505 00:29:05,560 --> 00:29:10,040 I wanna before you go Pankaj, what is the most important thing 506 00:29:10,040 --> 00:29:13,520 you would like our listeners to take away from our conversation 507 00:29:13,520 --> 00:29:17,520 about big data and AI? To all the listeners, I would 508 00:29:17,520 --> 00:29:21,360 like to say the AI and the big data, these are the technologies 509 00:29:21,360 --> 00:29:24,400 which will be there for a longer period of time. 510 00:29:24,880 --> 00:29:28,800 So get go get used to it this learn more or be curious. 511 00:29:29,120 --> 00:29:31,880 Ask the questions to your leadership, ask the question to 512 00:29:31,880 --> 00:29:34,680 your you know the mentors and make more connections. 513 00:29:34,680 --> 00:29:38,680 There are a lot of organizations or meet ups are meet ups are 514 00:29:38,680 --> 00:29:41,160 happening right? So they should register. 515 00:29:41,160 --> 00:29:44,800 They should volunteers their time and to learn from the 516 00:29:45,120 --> 00:29:49,920 industry leaders and experts. Like how the big data and the 517 00:29:49,960 --> 00:29:53,040 AIS are getting used into a different industries. 518 00:29:53,040 --> 00:29:56,560 At the same time they should also to work on their basic 519 00:29:56,560 --> 00:30:01,040 skills, maybe one programming language, domain knowledge, 520 00:30:01,520 --> 00:30:05,680 visualization tools. So be curious, keep keep 521 00:30:05,680 --> 00:30:09,560 learning, keep studying. Do a networking, learn from your 522 00:30:09,560 --> 00:30:11,920 leaders and on from your networks. 523 00:30:13,160 --> 00:30:16,120 Thank you so much, Pankaj. We'll catch up soon and we'll 524 00:30:16,120 --> 00:30:21,280 follow your journey as you write more around AI and big data. 525 00:30:21,280 --> 00:30:23,280 We hope to speak to you soon. Thank you. 526 00:30:23,280 --> 00:30:26,040 Thank you, Ben. It was a pleasure, you know, to 527 00:30:26,040 --> 00:30:27,200 speak to with you today.