WEBVTT 1 00:00:00.040 --> 00:00:04.160 A man and a child died in the town of Holdenville, where homes were 2 00:00:04.160 --> 00:00:08.560 destroyed. Damage is also being reported in a number of other communities, including 3 00:00:08.640 --> 00:00:13.720 the small town of Morris, where Jeff Moore with Olmugi County Emergency Management says 4 00:00:13.759 --> 00:00:18.640 crews were on standby as the storm hit. After everything cleared, we started 5 00:00:18.719 --> 00:00:22.480 responding to the area to do clean up and search and rescue operation. Governor 6 00:00:22.559 --> 00:00:26.519 Kevin stid Is declared a state of emergency for a dozen counties. The National 7 00:00:26.519 --> 00:00:31.359 Weather Services the outbreak of severe weather also caused widespread damage in parts of Iowa, 8 00:00:31.640 --> 00:00:35.679 Nebraska, Texas, and Kansas. The niece of an Israeli American hostage 9 00:00:35.719 --> 00:00:40.359 being held by Hamas and Gaza believes a hostage deal can be reached, and 10 00:00:40.479 --> 00:00:45.200 a Seagull told CBS's Face the Nation the Biden administration has shown it's committed to 11 00:00:45.240 --> 00:00:48.920 getting the hostages home, and a Seagull is the niece of Keith Siegel, 12 00:00:49.079 --> 00:00:52.719 who's been held hostage by the militant group for more than six months. A 13 00:00:52.799 --> 00:00:57.320 Moss released a video of Israeli hostages over the weekend, including Keith Siegel. 14 00:00:57.679 --> 00:01:00.560 He was kidnapped along with his wife during the octoger over seventh Attack on Israel 15 00:01:02.079 --> 00:01:06.200 set of Republican leader Mitch McConnell says the abortion issue will be sorted out at 16 00:01:06.239 --> 00:01:10.519 the state level. During an interview on NBC's Meet the Press, McConnell said 17 00:01:10.640 --> 00:01:15.040 he doesn't think a national abortion ban would even be able to pass. I 18 00:01:15.079 --> 00:01:19.040 don't think we'll get sixty votes in the Senate for any kind of national legislation. 19 00:01:19.239 --> 00:01:23.840 Democrats have made abortion a top campaign issue heading into the November elections, 20 00:01:23.879 --> 00:01:29.719 as states across the country have implemented strict abortion laws. Following the end of 21 00:01:29.799 --> 00:01:34.040 Roe v. Wade Taylor Swift is smashing records once again. The singer's latest 22 00:01:34.079 --> 00:01:38.439 album, The Tortured Poets Department, is the first album in Spotify's history to 23 00:01:38.519 --> 00:01:42.200 surpress Mandil. It's signed a Stay Sharp Now or anytime is the time to 24 00:01:42.359 --> 00:01:46.560 stay sharp. Have a holiday or a special event coming up. Maybe you 25 00:01:46.719 --> 00:01:49.599 just want to look good for that special someone. If staying sharp is important 26 00:01:49.640 --> 00:01:53.400 to you, then look good with the all new Stay Sharp barber shop in 27 00:01:53.439 --> 00:01:57.680 Redlands. Can a precision Blake cut or Barbara trim from the Master barbers at 28 00:01:57.719 --> 00:02:01.480 Stay Sharp. It's glean, convenient and cool. Watch sports on one of 29 00:02:01.480 --> 00:02:06.560 the many big screen TVs, kicked back with an adult beverage and a warm 30 00:02:06.640 --> 00:02:10.639 towel afterwards. Tapers, crop tops, gentlemen's cuts, fux hawks, skin 31 00:02:10.759 --> 00:02:15.159 fads, combovers, hot towel, shaves and hair design gentlemen. Treat yourself 32 00:02:15.199 --> 00:02:20.800 to the special service as Stay Sharp. Walkins are welcome. Stay Sharp Barbershop 33 00:02:20.800 --> 00:02:23.520 at the Tri City Center just off inter State ten and the two fifteen between 34 00:02:23.560 --> 00:02:28.240 Alabama and Tennessee. On the south side of the freeway. Call nine O 35 00:02:28.400 --> 00:02:31.080 nine two seven two twenty nine thirty one. That's nine on nine two seven 36 00:02:31.159 --> 00:02:38.120 two two nine three one and Stay sharp. Redlands Ranch Market is a unique, 37 00:02:38.159 --> 00:02:44.240 full service international or grocery store that specializes in authentic food items from Mexico, 38 00:02:44.439 --> 00:02:49.080 India, and from many Mediterranean and Asian countries, including popular items from 39 00:02:49.120 --> 00:02:53.080 the US. They offer fresh baked items from their in house bakery, housemade 40 00:02:53.159 --> 00:02:57.719 tortillas from their tortill area, a delicious array of prepared Mexican foods, a 41 00:02:57.800 --> 00:03:00.960 terrific fresh food and juice bar, and a large selection of meats, seafoods 42 00:03:00.960 --> 00:03:06.120 and deli sandwiches, salads, and hellal meats. Their produce department is stocked 43 00:03:06.159 --> 00:03:09.439 full with fresh, local and hard to find international fruits and vegetables that you 44 00:03:09.479 --> 00:03:15.000 cannot find anywhere else. Don't forget to step into the massive beer Cave and 45 00:03:15.159 --> 00:03:20.280 experience the largest selection of domestic, artisan, and imported beers in the ie. 46 00:03:20.560 --> 00:03:23.439 They can also cater your next event with one of the delicious takeout catering 47 00:03:23.479 --> 00:03:28.879 trades of food. Visit them at Redlands Ranch Market dot com. That's Redlands 48 00:03:29.000 --> 00:03:35.520 Ranch Market dot com. Redlands Ranch Market a unique and fun shopping destination located 49 00:03:35.560 --> 00:03:39.439 in the heart of San Bernardino, California. The Teamsters Local nineteen thirty two 50 00:03:39.560 --> 00:03:45.879 Training Center is designed to train workers for high demand, good paying jobs and 51 00:03:46.039 --> 00:03:51.280 various industries throughout the Inland Empire. If you want a pathway to a high 52 00:03:51.280 --> 00:03:55.520 paying job and the respect that comes with a union contract, visit nineteen thirty 53 00:03:55.560 --> 00:04:01.520 two Training Center dot org to enroll today. That's nineteen thirty two Training Center 54 00:04:01.800 --> 00:04:09.039 dot org to hebot Club's original pure powd rcosupert comes from the only tree in 55 00:04:09.080 --> 00:04:12.319 the world that fungus does not grow on. As a result, it naturally 56 00:04:12.439 --> 00:04:18.680 has antifungal, anti infection, anti viral, antibacterial anti inflammation and anti parasite 57 00:04:18.680 --> 00:04:23.000 properties. So the tea is great for healthy people because it helps build the 58 00:04:23.040 --> 00:04:27.639 immune system, and it can be truly miraculous for someone fighting a potentially life 59 00:04:27.680 --> 00:04:30.639 threatening disease due to an infection, diabetes, or cancer. The tea is 60 00:04:30.720 --> 00:04:35.000 also organic and naturally caffeine free. A one pound package of tea is forty 61 00:04:35.120 --> 00:04:39.160 nine to ninety five, which includes shipping. To order, please visit to 62 00:04:39.279 --> 00:04:42.800 Hebotea Club dot com. T hebo is spelled tea like tom, a, 63 00:04:43.439 --> 00:04:46.920 h ee b like boyo. Then continue with the word t and then the 64 00:04:47.000 --> 00:04:51.279 word club. The complete website is to Hebota Club dot com or call us 65 00:04:51.399 --> 00:04:57.240 at eight one eight six one zero eight zero eight eight Monday through Saturday nine 66 00:04:57.279 --> 00:05:01.920 am to five pm, California Time bad no voices of AACAA with an exciting 67 00:05:02.000 --> 00:05:06.839 announcement. Want to hear NBC News or KCAA anywhere you go, Well now 68 00:05:06.879 --> 00:05:11.920 there's an app for that. KCAA is celebrating twenty five years in our silver 69 00:05:12.000 --> 00:05:15.920 anniversary with a brand new app. The new KCAA app is now available on 70 00:05:16.079 --> 00:05:19.680 your smart device, cell phone, in your car, or any place. 71 00:05:19.959 --> 00:05:25.279 Just search KCAA on Google Play or in the Apple Store. One touch and 72 00:05:25.360 --> 00:05:28.959 you can listen on your car radio, Bluetooth device, Android Auto, or 73 00:05:29.040 --> 00:05:32.439 Apple Car Play. Catch the KCAA buzz in your earbuds or on the streets. 74 00:05:32.560 --> 00:05:38.279 Celebrating twenty five years of talk news and excellence with our new KCAA app. 75 00:05:38.519 --> 00:05:45.399 Just do it and download it. KCAA celebrating twenty five years. For 76 00:05:45.560 --> 00:05:48.920 over seventy five years, the Marine Toys for Tots program has provided toys and 77 00:05:49.000 --> 00:05:56.199 emotional support to economically disadvantaged children, primarily during the holidays. But needs are 78 00:05:56.240 --> 00:06:00.120 not just seasonal, and now neither is Toys for Tots. They've expanded their 79 00:06:00.160 --> 00:06:05.560 outreach to support families in need all year long with their new programs, including 80 00:06:05.600 --> 00:06:11.279 the Foster Care Initiative, the Native American Program, and the Youth Ambassador Program. 81 00:06:11.519 --> 00:06:14.920 To learn how you can help visit Toys for Tots. The Village Mud 82 00:06:15.000 --> 00:06:18.199 wants to remind pet owners of the importance of spang and neutering. Shelters overflow 83 00:06:18.240 --> 00:06:23.160 with unwanted pets. Spain and neutering helps prevent this and as many health benefits 84 00:06:23.199 --> 00:06:27.360 too. That message courtesy of the Village Mud at sixty sixty five East Foothill 85 00:06:27.399 --> 00:06:30.439 Boulevard and Claremont for self serve ped washtubs and high end food and treats for 86 00:06:30.560 --> 00:06:34.839 dogs and cats featuring natural and rock. Call the Village Mud nine oh nine 87 00:06:34.920 --> 00:06:46.360 six two four three zero two zero and like them on Facebook. All right, 88 00:06:46.480 --> 00:06:49.279 ladies and gentlemen, Hello and welcome back once again to the only coast 89 00:06:49.360 --> 00:06:55.000 to coast radio show all about the information economy. It's time for inside analysis. 90 00:06:55.040 --> 00:06:58.079 Who is truly Eric Cavanaugh is here with the good buddy of mine, 91 00:06:58.399 --> 00:07:01.839 one of my besties from the Pitts region, Andy Hanna, who has quite 92 00:07:01.879 --> 00:07:03.839 a pedigree. He's done a lot of really cool things. We're going to 93 00:07:03.879 --> 00:07:10.319 talk about the origin story for Blue Street Data. I came across Andy on 94 00:07:10.519 --> 00:07:15.120 LinkedIn, which is obviously the ideal networking platform in social media for business. 95 00:07:15.560 --> 00:07:18.959 Those folks have done an excellent job. They've also spun out some great technology. 96 00:07:19.000 --> 00:07:24.120 In fact, Apache Kafka, which is now everywhere under the stewardship of 97 00:07:24.160 --> 00:07:28.600 a company called Confluence that came out of LinkedIn. That's the engine used to 98 00:07:28.839 --> 00:07:32.920 run LinkedIn was called Apachi Kafka. It's now a huge source and in fact 99 00:07:32.959 --> 00:07:35.720 it's a source of data for one of my other clients, a company called 100 00:07:35.759 --> 00:07:42.040 Osceans. They just rolled out an announcement today, landed another forty nine point 101 00:07:42.079 --> 00:07:46.639 four million dollars of investment and extension of their Series B and Chris glad when 102 00:07:46.680 --> 00:07:48.959 their CEO was telling me the Confluent is one of the big sources of data 103 00:07:49.399 --> 00:07:55.000 for their platform, so LinkedIn good stuff. That's how I met Andy through 104 00:07:55.079 --> 00:07:59.720 one of his master's students at fourteen eighty six Labs. But he's got a 105 00:07:59.759 --> 00:08:03.560 long origin story than that, and we'll kind of dive in. It'll be 106 00:08:03.600 --> 00:08:07.519 fun to remember where we've come from and how we got here. It's kind 107 00:08:07.519 --> 00:08:11.000 of hard to really wrap your head around the changes. They've been so significant 108 00:08:11.160 --> 00:08:16.519 and so compelling over the last two decades. But here we are. So 109 00:08:16.639 --> 00:08:20.040 with that Andy Hannah from fourteen eighty six Labs and the University of Pittsburgh and 110 00:08:20.199 --> 00:08:24.439 oh thought, well, welcome to Inside Analysis. How's it going, Eric, 111 00:08:24.600 --> 00:08:26.120 Thank you so much for inviting me. This is so much fun. 112 00:08:26.360 --> 00:08:30.759 I'm so excited to be on the show this afternoon and can't wait to chat 113 00:08:30.879 --> 00:08:35.000 with you about you know, the past really thirty years and what's changed and 114 00:08:35.480 --> 00:08:41.120 how we've evolved over that time. Yeah. And I remember a friend of 115 00:08:41.159 --> 00:08:45.000 mine came over to my apartment in New Orleans and was all purpose driven and 116 00:08:45.200 --> 00:08:46.360 was like, you are going to get on this thing. I'm like, 117 00:08:46.480 --> 00:08:48.600 what are you talking about it? And it was the Internet and I was 118 00:08:48.720 --> 00:08:52.159 like, all right, I had heard about it through Prodigy. In fact, 119 00:08:52.240 --> 00:08:56.000 when I was a newspaper editor in Lamont, one of my assistants was 120 00:08:56.279 --> 00:09:00.480 all into Prodigy and it was like, oh, that's that is pretty interesting. 121 00:09:00.559 --> 00:09:03.360 And of course it took off very very quickly. But you'd have to 122 00:09:03.399 --> 00:09:09.759 think for a second, how did we get information before the Internet? And 123 00:09:09.960 --> 00:09:13.840 the answer is on floppy drives and you know, other hard medium that you 124 00:09:13.879 --> 00:09:18.519 would carry around. And I was in the printed US business, so you 125 00:09:18.559 --> 00:09:20.840 couldn't use floppy drives. You had to use these big, like portable discs. 126 00:09:20.919 --> 00:09:24.559 And I thought it was so cool to have by portable discs that I 127 00:09:24.759 --> 00:09:28.600 carry around because that was cool. But you were there too, so you 128 00:09:28.720 --> 00:09:31.399 remember the early days. How did we get information around back then? Sure, 129 00:09:31.480 --> 00:09:35.559 I mean you're exactly right, floppy disk, but I remember, you 130 00:09:35.600 --> 00:09:39.720 know, if you were in financial services, you had multiple terminals on your 131 00:09:39.799 --> 00:09:45.759 desk, right, you had a Bloomberg terminal Thompson Reuter's terminal. You had 132 00:09:45.799 --> 00:09:50.600 a terminal from Moody. So it's like you couldn't even combine the sources or 133 00:09:50.720 --> 00:09:54.039 the you know, the systems. In order to have one screen. You 134 00:09:54.159 --> 00:10:01.200 had to have multiple screens. It was a really inefficient time into retrieving data. 135 00:10:01.840 --> 00:10:03.799 I mean, I'm sure you remember if you wanted to give somebody a 136 00:10:03.960 --> 00:10:07.679 file, you had to stick it somewhere out there on the Internet. Literally, 137 00:10:07.759 --> 00:10:11.120 you couldn't attach it to an email. Remember you had the FTP it, 138 00:10:11.639 --> 00:10:15.639 right, and then I have to say, hey, Eric, here's 139 00:10:15.679 --> 00:10:20.360 the secret passageway to where the information is on the Internet. Go out there, 140 00:10:20.559 --> 00:10:24.200 grab it and download it, right. I mean, that's what the 141 00:10:24.279 --> 00:10:31.519 early nineties were like, whenever you were trying to share information. Yeah, 142 00:10:31.600 --> 00:10:35.519 and it's changed a lot. We're watching Young Sheldon now on Netflix and they 143 00:10:35.559 --> 00:10:41.399 were just joking about that, how Young Sheldon was on an FTP transferring files 144 00:10:41.000 --> 00:10:46.519 and that's what you would do. And it moved pretty darn quickly, you 145 00:10:46.559 --> 00:10:48.879 know. Once. I remember the early days, we called them asps right, 146 00:10:48.960 --> 00:10:52.399 application service providers. And I once had a smart guy in my show, 147 00:10:52.440 --> 00:10:56.080 I said, what's the difference between ESPN and suffer as a service and 148 00:10:56.200 --> 00:11:00.720 go SaaS works. I'm like, oh, I get it, smarty pants. 149 00:11:01.000 --> 00:11:07.039 But architectures changed, and there's tremendous innovation that came out of companies like 150 00:11:07.120 --> 00:11:11.440 Yahoo in the early days, like had Dupe that whatever to Google and really 151 00:11:11.519 --> 00:11:16.320 started this or I should say kick start at the open source movement into very 152 00:11:16.440 --> 00:11:22.080 serious data territory, right because Linux was the foundation of open source, and 153 00:11:22.159 --> 00:11:26.559 then ten odd years later you had this whole revolution in data management with had 154 00:11:26.639 --> 00:11:30.679 Dupe and a whole ecosystem around that. Of course, a patche Kofka is 155 00:11:30.799 --> 00:11:37.159 open source and all. So the channels from moving information have been rapidly advancing. 156 00:11:37.240 --> 00:11:39.919 We've got Snowflake, now we've got data bricks. We all have all 157 00:11:39.960 --> 00:11:43.639 these amazing companies that are allowing the use of analytics. But you and I 158 00:11:43.720 --> 00:11:46.519 have been there from the word go, so we remember what it was like 159 00:11:46.600 --> 00:11:50.720 to try to cobble this stuff together. And really you have to constantly change 160 00:11:50.759 --> 00:11:56.639 your mindset about things and recognize that the times have changed and recognized you have 161 00:11:56.720 --> 00:12:00.240 to be using the new platforms as time goes by. And I guess that's 162 00:12:00.720 --> 00:12:03.279 an epiphany that you had as well, right, yeah, yeah, that 163 00:12:03.399 --> 00:12:07.879 was in the mid nineteen nineties. I would say right around ninety five. 164 00:12:07.440 --> 00:12:13.360 There's a really visionary guy's named Gary Muller, wonderful guy who was doing research 165 00:12:13.480 --> 00:12:18.600 in Poland. And this is remember not too long after the fall of the 166 00:12:18.679 --> 00:12:24.200 Soviet Union and the former Soviet countries all wanted to privatize their assets. So 167 00:12:24.279 --> 00:12:30.480 what Gary was doing was doing research about a particular industry and what he was 168 00:12:30.559 --> 00:12:33.600 finding is that there was a dearth of information. And so a bunch of 169 00:12:33.679 --> 00:12:39.320 us got together and said, okay, let's build this information infrastructure about the 170 00:12:39.399 --> 00:12:46.559 former Soviet Union. So countries like Russia, Poland, Ukraine, you know, 171 00:12:46.919 --> 00:12:50.039 the Czech Republic, et cetera. And what we did is we sent 172 00:12:50.200 --> 00:12:56.320 little teams into each of those countries and they mind data. Literally we either 173 00:12:56.519 --> 00:13:03.000 we got it somehow digitally or we went to uh copy machines and actually copied 174 00:13:03.039 --> 00:13:07.639 it. And uh, I'm serious that because it wasn't in existence. We 175 00:13:07.720 --> 00:13:11.000 had to go to banks, companies, analysts, and then we would consolidate 176 00:13:11.080 --> 00:13:16.120 it and all in digital form. And we needed a way to distribute this 177 00:13:16.279 --> 00:13:20.320 information. We weren't going to do it over Bloomberg or Reuters and so we 178 00:13:20.399 --> 00:13:22.639 said, hey, there's this really cool thing called the Internet. You know, 179 00:13:22.759 --> 00:13:28.240 Yahoo seems to be making it work. And so we literally put you 180 00:13:28.320 --> 00:13:31.720 know, we probably created one of the first SaaS based applications out there. 181 00:13:31.799 --> 00:13:37.399 We put it on the one and we got banks and large companies wanted to 182 00:13:37.480 --> 00:13:41.799 investor do business in the former Soviet Union. They connected into these these this 183 00:13:43.039 --> 00:13:48.360 database and they paid per drink essentially, and so meaning that every time that 184 00:13:48.440 --> 00:13:50.600 they went it went in, pulled information up, they would they would pay. 185 00:13:50.639 --> 00:13:54.080 It was a it was a total gas. We had to put the 186 00:13:54.159 --> 00:13:58.320 first Internet Service Provider ISP for all those who are kind of from a different 187 00:13:58.360 --> 00:14:03.039 generation, but we had to put the ISP, first ISPN poland out there. 188 00:14:03.120 --> 00:14:07.360 We actually owned the first ISPN poll wow, And so that was what 189 00:14:07.639 --> 00:14:11.039 it was. It was crazy. It was a wild time, and it 190 00:14:11.240 --> 00:14:16.159 was it was pre infrastructure, and so we developed our own accounting systems, 191 00:14:16.240 --> 00:14:20.919 we developed our own distribution systems. It's kind of crazy. It's a lot 192 00:14:20.960 --> 00:14:24.399 of fun. Yeah. Well, and you were at the very forefront of 193 00:14:24.480 --> 00:14:28.759 what is now called alternative data. And you and I have talked about this. 194 00:14:28.840 --> 00:14:31.399 I've done a lot of research into it, and you know, like 195 00:14:31.480 --> 00:14:35.200 you, I think down the road like, hmm, how is this going 196 00:14:35.240 --> 00:14:39.159 to pan out? And clearly, the amount of alternative data being bought and 197 00:14:39.240 --> 00:14:43.840 sold today is just vast. It's stunning, and a lot of people don't 198 00:14:43.919 --> 00:14:48.120 realize that credit card companies sell what they call exhaust data, which is then 199 00:14:48.200 --> 00:14:52.519 bought by a whole variety of firms, most notably investment bankers. And I 200 00:14:52.600 --> 00:14:56.519 wrote about this a number of years ago saying that it's a pretty unfair playing 201 00:14:56.639 --> 00:15:03.080 field if these investment bankers can all that data at their disposal, because now 202 00:15:03.360 --> 00:15:07.600 they can see, for example, who's going to meet or beat market estimates 203 00:15:07.679 --> 00:15:11.200 on revenue because they have the actual raw data and they can compare it to 204 00:15:11.279 --> 00:15:15.879 a baseline. So why don't we have some publicly facing consumer data leak with 205 00:15:16.000 --> 00:15:18.240 my suggestion, so you could level the playing field and so the rest of 206 00:15:18.320 --> 00:15:22.840 us could all benefit from that. And you're kind of seeing the start of 207 00:15:22.919 --> 00:15:26.080 that in sort of in different places. And of course you also have a 208 00:15:26.279 --> 00:15:31.080 history in the machine learning plus data space through this company called O Thought right, 209 00:15:31.120 --> 00:15:35.600 tell us about that? Yeah, sure, I think one stop before 210 00:15:35.679 --> 00:15:39.240 then I'll talk a little bit about sort of that transformation from data is powerful 211 00:15:39.720 --> 00:15:45.799 to data plus analytics or AI is more powerful. And that was the time 212 00:15:45.840 --> 00:15:48.320 that I spent A good friend of mine started a company called the International Institute 213 00:15:48.360 --> 00:15:52.120 for Analytics. His name is Jack Phillips, along with Tom Davenport, who 214 00:15:52.159 --> 00:15:58.320 we all know is sort of the guru and business analytics. And you know, 215 00:15:58.399 --> 00:16:00.399 about fifteen years ago they together and said, you know what, we 216 00:16:00.519 --> 00:16:06.519 need to be the leading research firm about how firms are using analytics, looking 217 00:16:06.600 --> 00:16:12.080 at the maturity of companies in terms of low versus high, and then whether 218 00:16:12.200 --> 00:16:17.919 or not they outperform companies that actually are less analytic. And you know, 219 00:16:18.080 --> 00:16:22.320 the research holds true, the more analytic you are, the better you use 220 00:16:22.440 --> 00:16:27.240 data the top of the heap you'll be. So good evidence of that is 221 00:16:27.279 --> 00:16:30.639 if we look at the top ten companies of the s and P five hundred. 222 00:16:32.279 --> 00:16:34.440 You know, these are the Metas, the Googles, the Microsoft's, 223 00:16:34.600 --> 00:16:38.960 the Tesla's, you know, they're using data in a very different way Amazon, 224 00:16:41.000 --> 00:16:45.759 and they represent forty percent of the total value of the SMP five hundred. 225 00:16:47.000 --> 00:16:52.639 While so they learned they're the early adopters. They understood power data power 226 00:16:52.679 --> 00:16:57.679 of analytics together and I feel very fortunate to spend time with Jack and Tom 227 00:16:57.879 --> 00:17:03.400 and do some consulting some very large companies and understand this transformation that was going 228 00:17:03.480 --> 00:17:07.880 on. And then me and a guy named John Evattico, a guy named 229 00:17:08.079 --> 00:17:11.759 Jeremy Garvey started said Hey, we're going to do this. So twenty fourteen 230 00:17:12.920 --> 00:17:18.359 we started this company called Othought and the idea about oh thought was, can 231 00:17:18.480 --> 00:17:25.720 we at scale take data about students, high school students and help universities colleges 232 00:17:26.279 --> 00:17:30.079 enroll the best fit students who are going to persist and graduate, and how 233 00:17:30.200 --> 00:17:38.720 can we help those institutions get do better, get more graduated on time, 234 00:17:38.880 --> 00:17:44.799 get better jobs, et cetera. So that's where we took machine learning coupled 235 00:17:44.880 --> 00:17:51.440 with all the rich data that high school students provide to universities, combine those 236 00:17:51.480 --> 00:17:55.920 two together, and we build othought. And that was a seven or eight 237 00:17:56.000 --> 00:18:00.720 year journey. Well well, and just to explain to our audience here, 238 00:18:00.119 --> 00:18:04.920 what you're able to do with data at scale and analytics and especially machine learning 239 00:18:06.640 --> 00:18:10.599 is, of course, identify patterns. But then the real magic is once 240 00:18:10.640 --> 00:18:14.160 you've identified a pattern, let's say, a pattern of a successful student. 241 00:18:14.759 --> 00:18:18.079 Once you've identified that pattern, then you can sort of distill it, understand 242 00:18:18.160 --> 00:18:22.759 it is almost like a recipe, and then apply it to the rest of 243 00:18:22.839 --> 00:18:26.880 the data and see. Aha, So these forty out of one hundred students 244 00:18:26.000 --> 00:18:32.160 actually fit that pattern of behavior in terms of interests, in terms of writing 245 00:18:32.279 --> 00:18:34.279 style. There are all sorts of different bits and pieces you can cobble together. 246 00:18:34.920 --> 00:18:38.279 And that's really the power of this analytical technology, right, is that 247 00:18:38.359 --> 00:18:42.799 you can understand what works and what doesn't work. You can model both and 248 00:18:42.960 --> 00:18:48.119 then analyze current data to say, Okay, according to our model that we've 249 00:18:48.200 --> 00:18:51.640 built, these forty students are going to do very well. These sixty probably 250 00:18:51.720 --> 00:18:53.559 not so much. And you'll never get it completely right. I mean, 251 00:18:53.599 --> 00:18:56.599 there's I think a bit of a misconception that, oh, with all the 252 00:18:56.640 --> 00:19:00.200 best analytics, you'll always be right. No, you won't always be right, 253 00:19:00.440 --> 00:19:03.400 but you will greatly increase your chances of being correct, and you'll create 254 00:19:03.440 --> 00:19:08.319 what's what's called lift right one hundred percent. So that's so let's let's even 255 00:19:08.400 --> 00:19:12.079 maybe easier for the audience to understand, is if we're looking at the probability 256 00:19:12.119 --> 00:19:15.839 of somebody to enroll, right, So what let's say that Eric, we 257 00:19:17.200 --> 00:19:19.839 want you to enroll at the University of Pittsburgh. We see that you're twenty 258 00:19:19.880 --> 00:19:25.079 percent likely to enroll because that's what the machine learning has told us, and 259 00:19:25.400 --> 00:19:29.039 that's because of who you are and how you behave. So we know all 260 00:19:29.119 --> 00:19:32.400 the data about where you went to high school, your grades, you know, 261 00:19:32.480 --> 00:19:34.680 the activities, et cetera. And then we see your behavior. Right, 262 00:19:34.759 --> 00:19:38.880 so you're hitting website, what you're reading, you know, the visits 263 00:19:38.920 --> 00:19:42.079 that you go to. Are you an engaged individual? So you land at 264 00:19:42.119 --> 00:19:45.440 this twenty percent, Well, we want to do if this you know, 265 00:19:45.519 --> 00:19:48.240 we want Eric, we want to be eighty percent. So we look at 266 00:19:48.279 --> 00:19:53.160 the prescriptions. What can we do that's going to increase his probability? So, 267 00:19:53.400 --> 00:19:56.920 for example, we might send you a particular marketing campaign, give you 268 00:19:57.000 --> 00:20:00.799 a particular scholarship. This is you not high school, and all of a 269 00:20:00.839 --> 00:20:04.160 sudden, because we are studying the data in your patterns just like you said, 270 00:20:04.759 --> 00:20:10.079 we see that those activities raise your probability from twenty to eighty percent. 271 00:20:11.400 --> 00:20:15.359 And this applies to every industry. This happens to be higher education that we're 272 00:20:15.359 --> 00:20:18.000 talking about. But this is how Amazon gets you to buy all the products 273 00:20:18.039 --> 00:20:22.039 that it gets you to buy. To that recommendation engine. Yeah, and 274 00:20:22.160 --> 00:20:26.839 it's what's cool is that you can learn about the students, you know, 275 00:20:26.960 --> 00:20:32.200 So it's great for improving efficacy, but it's also great for educating the people 276 00:20:32.240 --> 00:20:33.839 who are involved in the process, right, because when you play with these 277 00:20:33.920 --> 00:20:37.359 models, when you play with the data, that's when you start to understand 278 00:20:38.039 --> 00:20:41.880 how it all fits together, how the tumblers align. And I think we're 279 00:20:41.880 --> 00:20:47.000 at a very exciting time in our business world and the education world as well, 280 00:20:47.559 --> 00:20:49.960 because we've reached a bit of a critical mass in that regard, meaning 281 00:20:51.359 --> 00:20:53.240 we have the compute power now, we have tons and tons of data, 282 00:20:53.680 --> 00:20:57.519 we have smart people who have methodologies, who have published these methodologies that we 283 00:20:57.599 --> 00:21:02.200 can follow and understand. So it's really all coming together in terms of being 284 00:21:02.200 --> 00:21:07.599 able to leverage this stuff. And then you look at artificial intelligence CHAT, 285 00:21:07.680 --> 00:21:11.200 GPT, large language models, which are of course are very very powerful. 286 00:21:11.000 --> 00:21:17.400 They are going to require some effort to govern and to manage responsibly. But 287 00:21:17.519 --> 00:21:19.559 I think we're up to the task now, right because of all this experience, 288 00:21:19.680 --> 00:21:25.160 because what we have is so valuable, and I think the transparency and 289 00:21:25.279 --> 00:21:26.720 ethics are really going to come into play here in the near future. What 290 00:21:26.759 --> 00:21:30.799 do you think hugely, hugely important. I mean, one of the things 291 00:21:30.839 --> 00:21:33.680 that you said that people don't often get is it's not just about predicting the 292 00:21:33.759 --> 00:21:37.799 future. These models that you build allow you to diagnose where you might have 293 00:21:37.960 --> 00:21:44.359 issues. So you can detect bias right through these models. So if there's 294 00:21:44.480 --> 00:21:48.279 bias in the data or biased in the interpretation of the results, you can 295 00:21:48.359 --> 00:21:51.359 look at it and you can do some reverse engineering and say, hey, 296 00:21:51.440 --> 00:21:55.640 what if I change the gender. What if I change does the results? 297 00:21:55.799 --> 00:22:02.440 Do the results change? Oh? Wow, this particular subsetgment is disadvantage because 298 00:22:02.519 --> 00:22:04.880 of that, because it just just by changing the gender, which means that 299 00:22:04.920 --> 00:22:07.759 we have a problem. Right, So then we've diagnosed it. Now we 300 00:22:07.839 --> 00:22:12.400 can fix it. So and I think that that's an area that we're really 301 00:22:12.599 --> 00:22:18.559 just starting to dive into so that we can understand how to fix the problems 302 00:22:18.599 --> 00:22:22.119 that exist. And by the way, we'll never fix all the problems. 303 00:22:22.240 --> 00:22:25.720 We just got to put on the big ones. Yeah. Well, and 304 00:22:25.799 --> 00:22:27.359 it's a journey, right, it's a process. And of course when you 305 00:22:27.480 --> 00:22:30.480 fix one thing, other things break. It's like painting your house. You 306 00:22:30.599 --> 00:22:33.599 paint one wall, you got to paint them all, right, Talking nuts 307 00:22:33.680 --> 00:22:37.079 unless you do that. But you're hitting on something very important, which I 308 00:22:37.200 --> 00:22:44.839 think is probably the most valuable proposition about AI and machine learning is that it 309 00:22:45.000 --> 00:22:51.279 helps you get to understanding the problems. That discovery side is very challenging because 310 00:22:51.319 --> 00:22:55.160 it could be anywhere. I often use the example of you lose your wallet. 311 00:22:55.440 --> 00:22:56.480 You're looking for it all around the house. Is it even in the 312 00:22:56.519 --> 00:23:00.440 house. It's very frustrating because you don't know exactly where it is, don't 313 00:23:00.440 --> 00:23:02.559 know where to look. That's kind of the way it is with large sets 314 00:23:02.559 --> 00:23:06.960 of data too. You have to start filtering it and taking it a hard 315 00:23:07.119 --> 00:23:10.720 look at things. Like you said, let's flip out the gender see if 316 00:23:10.799 --> 00:23:15.039 things change. Oh wow, they change dramatically. That's part of the process 317 00:23:15.160 --> 00:23:18.880 of working with the data, of trying to understand the data. And it 318 00:23:18.960 --> 00:23:22.400 takes effort. I mean it's we're getting better and better with the technologies, 319 00:23:22.440 --> 00:23:26.480 but still it takes human beings in the loop it analyzing the data. I'll 320 00:23:26.480 --> 00:23:29.559 give you a fun observation before a first break. Come e for in a 321 00:23:29.599 --> 00:23:33.160 minute, what good buddy of mine who also runs an analytics institute. He 322 00:23:33.200 --> 00:23:36.759 goes, Yeah, machines don't have the ability to go, huh, that's 323 00:23:36.880 --> 00:23:41.759 kind of weird, but human beings do, right, Yeah, I mean 324 00:23:41.839 --> 00:23:45.720 that's a good point. You know, I thought our chief data scientists, 325 00:23:45.799 --> 00:23:49.160 mart Fordman, he's actually developed what we call the automated data scientists. So 326 00:23:49.640 --> 00:23:53.960 the automated data scientist does exactly what you're talking about. It scans results, 327 00:23:55.039 --> 00:23:59.759 it scans data, it scans changes, you know, model drift, and 328 00:23:59.839 --> 00:24:03.319 it says, aha, it tells our people, aha, we got a 329 00:24:03.359 --> 00:24:06.519 problem over here. So it's like a member of the team, which is 330 00:24:06.599 --> 00:24:08.799 how I think that we need to start thinking about AI as a member of 331 00:24:08.839 --> 00:24:12.440 the team and not an adversary. That's a really really good point. And 332 00:24:12.759 --> 00:24:15.519 we're up to our first break here, but don't touch up now, folks. 333 00:24:15.559 --> 00:24:19.480 We're talking with Andy Hannah all about machine learning and data and the origin 334 00:24:19.559 --> 00:24:22.039 story of blue Street Data. Will be right back. You're listening to Inside 335 00:24:22.039 --> 00:24:34.400 Analysis. Welcome back to Inside Analysis. Here's your host, Eric is Tavanaugh. 336 00:24:37.119 --> 00:24:41.079 Okay, folks, back here on Inside Analysis with the one and only 337 00:24:41.160 --> 00:24:44.759 Andy Hannah of fourteen eighty six Labs. Or get to that in just a 338 00:24:44.799 --> 00:24:47.880 minute. And of course, oh thought, and in the break there, 339 00:24:48.039 --> 00:24:52.599 we're just chatting about the fact that data is changing the nature of data is 340 00:24:52.720 --> 00:24:56.039 changing, and what we do with it is changing. You have these cycles 341 00:24:56.119 --> 00:24:59.920 basically, you know, I'll just throw one fun observation out there that I 342 00:25:00.160 --> 00:25:03.599 seen in the recent past, which is observability. This whole space blew up 343 00:25:03.680 --> 00:25:11.119 called observability, and basically it's just windows into feeds of machine data that's cruising 344 00:25:11.200 --> 00:25:14.559 around of things that are happening, connections to data sets, for example. 345 00:25:15.000 --> 00:25:18.839 And within like a year to two years of observability becoming a thing, there's 346 00:25:18.920 --> 00:25:22.920 already the company that sits on top of your observability data and allows you to 347 00:25:22.960 --> 00:25:26.200 manage it. So there's already a meta solution We had them on the show 348 00:25:26.839 --> 00:25:32.920 last year that allows you to It's called Mesmo, and they allow you to 349 00:25:33.680 --> 00:25:37.799 transform and process and correlate data streams from observability. Right, So these are 350 00:25:37.880 --> 00:25:42.359 new ways of dealing with data. It's not all unstructured data like human texts 351 00:25:42.400 --> 00:25:45.920 and images and things that they thination. A lot of it is machine data 352 00:25:45.039 --> 00:25:48.839 so as they talk to each other, it's capturing that information. But the 353 00:25:48.920 --> 00:25:52.880 point is this is all grist for the mill, as they say, and 354 00:25:52.240 --> 00:25:56.960 what data analysts and business people in general need to appreciate these days is that 355 00:25:57.119 --> 00:26:00.880 it is changing and it continues to change. So you have to kind of 356 00:26:02.000 --> 00:26:07.000 change the pace of your own processes and workflows and understandings to be able to 357 00:26:07.039 --> 00:26:08.440 leverage itself. And you'll never be completely up to date, you know what 358 00:26:08.480 --> 00:26:11.680 I mean. You're never going to be riding the crush of the way for 359 00:26:11.720 --> 00:26:15.559 any significant time because everything is changing all around us. So you kind of 360 00:26:15.599 --> 00:26:18.759 have to be very agile and very open minded and by the way, put 361 00:26:18.839 --> 00:26:22.960 some governance into place. Right. Governance, I think is going to be 362 00:26:22.200 --> 00:26:27.440 a hugely popular topic for many years to come because we're in such a turbulent 363 00:26:27.640 --> 00:26:33.279 time and because the outcomes are still relatively unknown. Well, if you don't 364 00:26:33.319 --> 00:26:37.119 have governance and you flip the switch of aon and your aion and your organization, 365 00:26:37.559 --> 00:26:41.200 you're going to get some crazy stuff flying on the other end. It's 366 00:26:41.279 --> 00:26:44.880 not going to work well for you. But give me your thoughts on that. 367 00:26:45.039 --> 00:26:48.720 The changing nature of the data landscape and how you have to change with 368 00:26:48.880 --> 00:26:52.000 it. Yeah, So we're I think we're really lucky that we've seen the 369 00:26:52.240 --> 00:26:59.160 evolution of this technology and the source of the power of the technology, which 370 00:26:59.200 --> 00:27:02.640 is a data over to pass forty years, right, So if you go 371 00:27:02.799 --> 00:27:06.240 back forty years, we know, you know, Yahoo had just been going 372 00:27:06.400 --> 00:27:10.599 public, you know, or just just formed. So we've come a long 373 00:27:10.680 --> 00:27:15.839 way since those first days. And even if it's only two decades since Yelp 374 00:27:15.960 --> 00:27:18.240 and Twitter have been around, and really the growth of that and the use 375 00:27:18.319 --> 00:27:23.079 of that data is over to pass decade. So the ability and the examples 376 00:27:23.119 --> 00:27:27.359 that you give are wonderful are but the ability to think about the use of 377 00:27:27.440 --> 00:27:33.839 that data that didn't exist, that user level data did not exist until a 378 00:27:33.920 --> 00:27:37.559 decade ago. And so the ability for and we see that as we mentioned 379 00:27:37.640 --> 00:27:44.519 to the top companies in the country using that data to understand behavior and to 380 00:27:44.680 --> 00:27:48.119 change that behavior, the increase probability of buying those types of things. How 381 00:27:48.160 --> 00:27:52.319 do we you know, how does the rest of the world, Maybe this 382 00:27:52.440 --> 00:27:55.880 is a good question for you. We know, it's like the rich getting 383 00:27:55.960 --> 00:28:00.000 richer and how does the poor move up? Right? Because these laws, 384 00:28:00.039 --> 00:28:04.039 large companies are the ones who are pushing the edges of technology and the use 385 00:28:04.119 --> 00:28:08.799 of data when the eighty percent of the companies that are out there are still 386 00:28:08.839 --> 00:28:14.920 trying to understand digital transformation and what can I do with that data they're learning 387 00:28:15.000 --> 00:28:18.799 about the use cases and the applications. You know, are we going to 388 00:28:18.880 --> 00:28:22.599 continue to have this huge divide so that the you know, the true value 389 00:28:22.640 --> 00:28:29.160 that's driven by AI and data is really housed in a small number of companies. 390 00:28:29.839 --> 00:28:32.400 What do you I mean, what do you think, Derek there, 391 00:28:32.519 --> 00:28:36.079 I mean, that's that's a very very big question. I think transparency is 392 00:28:36.119 --> 00:28:38.000 going to be the key. I think this idea through at the top of 393 00:28:38.079 --> 00:28:42.640 the hour of a consumer facing data lake basically would be interesting because it's a 394 00:28:42.680 --> 00:28:47.720 place where we can all dive in and sort of better understand what's happening with 395 00:28:47.799 --> 00:28:52.000 this data and ownership. You know, data ownership is a big issue these 396 00:28:52.119 --> 00:28:56.680 days. You're starting to see some of the bigger companies at least give lip 397 00:28:56.799 --> 00:29:00.880 service to this, but allowing you to own your data, giving you control 398 00:29:00.000 --> 00:29:03.519 of when your data can be used to train algorithms, for example. There 399 00:29:03.559 --> 00:29:07.359 are lots of people coming up with business models to allow you to own your 400 00:29:07.440 --> 00:29:11.160 data and control that. But it's hard. It's going to be hard to 401 00:29:11.440 --> 00:29:15.559 unseat the Amazons and the Microsofts and the Googles of the world. But I 402 00:29:15.599 --> 00:29:19.559 will tell you that the big guys are nervous too. There is tremendous tension 403 00:29:19.599 --> 00:29:23.640 at Google right now. They are very nervous about what just happened with chat 404 00:29:23.720 --> 00:29:29.599 GPT in particular, and then Microsoft stepping in to buy it basically or invest 405 00:29:29.680 --> 00:29:32.400 heavily in it, I should say. And then what happened. The board 406 00:29:32.480 --> 00:29:34.960 for open Ai kicks out Sam Altman, and over the course of the weekend, 407 00:29:36.079 --> 00:29:40.160 Sata Nadela offers to hire him, and you're just like, Holy Christmas, 408 00:29:40.880 --> 00:29:45.839 what is going on? Ye, we do need transparency, and I 409 00:29:45.039 --> 00:29:52.279 think that while I fully respect the concept of intellectual property and having to protect 410 00:29:52.319 --> 00:29:56.119 these sorts of things, to get right down to it, these models are 411 00:29:56.240 --> 00:30:02.599 so incredibly complex that you could offer plet transparency and I don't think you're jeopardizing 412 00:30:02.680 --> 00:30:06.680 your business model really much at all. And in fact, I just heard 413 00:30:06.720 --> 00:30:10.640 today that apparently Elon Musk is going to open source GROC. Of course, 414 00:30:10.720 --> 00:30:15.319 Mark Zuckerberg open source Lama or Lama too, So you know, again you 415 00:30:15.400 --> 00:30:19.000 kind of see these these movements, and I just read today that who is 416 00:30:19.079 --> 00:30:25.680 it? I think Google is following suit and allowing you to exit their data 417 00:30:25.759 --> 00:30:30.599 centers, right because they're all these exit fees. Basically that we're just killing 418 00:30:30.640 --> 00:30:33.559 people. I mean, that was the big joke is you can put all 419 00:30:33.559 --> 00:30:34.480 your data in the cloud for cheap, but if you try to take it 420 00:30:34.519 --> 00:30:38.279 out, those egress fees were going to hammer you. I do think there's 421 00:30:38.359 --> 00:30:41.640 pressure on the big guys, but I think you're right that we need to 422 00:30:41.720 --> 00:30:45.400 be very careful. And you know there was concern what if this kind of 423 00:30:45.640 --> 00:30:51.359 LLLM capability is restricted to a handful of people, Well, that would be 424 00:30:51.480 --> 00:30:55.319 a pretty bad situation. I think we want we all want access to these 425 00:30:55.359 --> 00:30:59.279 technologies, and I think it's good that we have competition at the highest levels 426 00:30:59.319 --> 00:31:04.359 that's pushing towards that transparency. But I think transparency and education are the two 427 00:31:04.920 --> 00:31:08.480 massive keys to success to give power to the little people out there. What 428 00:31:08.559 --> 00:31:12.200 do you think? Yeah, I agree with that. I think that that's 429 00:31:12.480 --> 00:31:18.359 a tull order. I think you know that we were talking earlier about data 430 00:31:18.400 --> 00:31:22.279 warehousing, and you know there's still there's still lots of companies trying to figure 431 00:31:22.279 --> 00:31:27.079 out what's my single source of truth as opposed to thinking about what's the latest 432 00:31:27.160 --> 00:31:33.279 in architecture that I can use both external data and internal data to outcompete those 433 00:31:33.359 --> 00:31:37.279 who are in my space. Right, So it's I just I look across 434 00:31:37.799 --> 00:31:45.640 the spectrum of companies and I am just floored by how the distribution of capabilities, 435 00:31:47.119 --> 00:31:52.200 the distribution of knowledge, and there are so many companies that are just 436 00:31:52.359 --> 00:31:55.240 at the beginning of this journey, and you and I, you know, 437 00:31:55.400 --> 00:31:59.039 understand this the end of the journey because we work with the larger companies are 438 00:31:59.119 --> 00:32:01.319 on the edge. You know, we understand retail media networks, and we 439 00:32:01.480 --> 00:32:06.279 understand data clean rooms and data fabric and all this. But you know, 440 00:32:06.559 --> 00:32:10.319 most most companies are going to Gartner's annual conference to learn about what the heck 441 00:32:10.400 --> 00:32:15.200 these things are, right, let alone being able to gain value out out 442 00:32:15.240 --> 00:32:20.119 of this technology. So I completely agree with you, you know, the 443 00:32:20.200 --> 00:32:24.000 transparency and the education. I just don't know how it's going to happen without 444 00:32:24.160 --> 00:32:29.799 mechanisms to say, okay, here's how you can get value, you know, 445 00:32:30.039 --> 00:32:32.519 and then let's help. Let's help each other, Like on fraud as 446 00:32:32.559 --> 00:32:37.160 an example, we can share knowledge about fraud detection. Right, It's it's 447 00:32:37.480 --> 00:32:40.759 do we want to compete on fraud detection? I hope not? Right, 448 00:32:40.799 --> 00:32:44.759 hope, that's something that we want to share. So why don't we have 449 00:32:44.880 --> 00:32:49.319 a common way of looking at fraud and helping each other in every industry understand 450 00:32:49.319 --> 00:32:54.160 where there's fraudulent customers and or customers that don't even exist. And so I 451 00:32:54.279 --> 00:32:58.640 think we have to figure out a way to get people to that that base 452 00:32:58.759 --> 00:33:01.880 camp of use cases and understand how they work. I like the concept the 453 00:33:01.920 --> 00:33:07.039 base camp of use cases. You're absolutely right about fraud. I've talked about 454 00:33:07.079 --> 00:33:10.400 that on this show even fifteen years ago, with the same exact thought. 455 00:33:10.440 --> 00:33:15.440 It's like, let's share threat detection data at least within industries and the financial 456 00:33:15.519 --> 00:33:20.799 services, in healthcare and insurance and different industries where they're the same business model. 457 00:33:21.359 --> 00:33:25.400 Let's share information about threats because they can spread very quickly and they can 458 00:33:25.480 --> 00:33:30.680 bring companies down. I mean, you were hearing all these stories about ransomware 459 00:33:30.799 --> 00:33:32.680 and just you know, terrible things that are happening with people stealing your data, 460 00:33:32.720 --> 00:33:37.400 stealing access to your data and causing all kinds of trouble. But I 461 00:33:37.440 --> 00:33:40.920 will say the good news is that the young people today, they're all on 462 00:33:42.039 --> 00:33:45.279 their iPhones, they're on their Samsungs, they're on their different devices, and 463 00:33:45.400 --> 00:33:51.799 they understand technology as as a first class citizen. Whereas a lot of us, 464 00:33:51.960 --> 00:33:53.519 you know, middle aged folks. Let's say, you know, we've 465 00:33:53.599 --> 00:33:58.759 had the whole journey of knowing what it was like before even computers for crime 466 00:33:58.799 --> 00:34:01.720 out loud, before that was mainstream, where like reading books, and you 467 00:34:01.799 --> 00:34:07.559 know, you do have to I think every so often gut check yourself and 468 00:34:07.680 --> 00:34:10.119 mind check yourself and say, all right, do am I still lingering in 469 00:34:10.199 --> 00:34:15.840 these old habits and answers? You probably are, So how do you shake 470 00:34:15.920 --> 00:34:17.920 out of that? And I think you do it by listening to people and 471 00:34:19.079 --> 00:34:22.440 just by going to events and going to webinars and things, and just taking 472 00:34:22.480 --> 00:34:27.760 a moment to hear what other people are talking about and to see and to 473 00:34:27.920 --> 00:34:30.199 watch how they interact with data. You know. That's why I'm so excited 474 00:34:30.239 --> 00:34:35.119 about transparency of data sets. And it's really come a long way in the 475 00:34:35.199 --> 00:34:37.559 last i'd say ten years. And you guys, of course are part of 476 00:34:37.599 --> 00:34:42.719 this, but transparency data because it's only once you really start playing around or 477 00:34:42.719 --> 00:34:45.599 as they say, munging the data that you start to get a better understanding 478 00:34:45.679 --> 00:34:50.760 for what that process is like and how it really does take time. Even 479 00:34:50.800 --> 00:34:53.280 if you have the best technologies, it takes time to play around flatten things 480 00:34:53.400 --> 00:34:57.880 move things around, see it in motion, you know. But that's when 481 00:34:57.920 --> 00:35:00.320 you start to get it and you're like, oh wait a minute, and 482 00:35:00.480 --> 00:35:05.559 that's the beginning of a career, right, Yeah, it's you know, 483 00:35:05.679 --> 00:35:07.960 it's interesting. We're you know, a lot of people don't know this. 484 00:35:08.079 --> 00:35:14.159 We now have a major, an analytics major at the University of Pittsburgh for 485 00:35:14.400 --> 00:35:19.719 undergraduate business students. So these students are coming in in their freshman year and 486 00:35:19.760 --> 00:35:23.559 they're learning to program in Python. They're learning how to mind data, they're 487 00:35:23.639 --> 00:35:30.079 learning how to use that data and you know, uh, different models to 488 00:35:30.199 --> 00:35:36.440 make decisions. I mean, so they're they're graduating with with a language that 489 00:35:36.800 --> 00:35:42.199 most people at the organizations that are going into only casually understand. I find 490 00:35:42.280 --> 00:35:46.920 that is a very interesting sort of You have this group of really fluent in 491 00:35:47.079 --> 00:35:53.639 technology new hires and then up to levels people are there. Their their level 492 00:35:53.719 --> 00:35:59.199 knowledge is what they've read, you know, from McKenzie or from you know, 493 00:35:59.480 --> 00:36:02.719 something from you know, towards data science or something like that that they 494 00:36:02.719 --> 00:36:06.599 are getting in the nuance of it that they haven't lived yet. So how 495 00:36:06.679 --> 00:36:10.880 do how do we deal with that? In these companies that divide between sort 496 00:36:10.920 --> 00:36:15.360 of three level I'm not talking about eppmansion. I'm even talking about two or 497 00:36:15.400 --> 00:36:20.400 three levels up in those new hires, and that that that ability to translate 498 00:36:20.519 --> 00:36:24.000 what one set knows versus what the other doesn't. Yeah, well, I'll 499 00:36:24.039 --> 00:36:29.519 tell you one of the lessons of life that I've taken away is you've got 500 00:36:29.639 --> 00:36:32.239 to realize that you do have to reset and recognize there are a lot of 501 00:36:32.320 --> 00:36:37.000 things that you don't know and embrace the change. I suppose you know, 502 00:36:37.079 --> 00:36:39.920 and I try to do it on these shows by talking to smart people. 503 00:36:40.000 --> 00:36:44.519 Of course, you are in the university space, so you're talking to the 504 00:36:44.599 --> 00:36:46.840 next generation of business executives. I mean, that's who these folks are. 505 00:36:46.920 --> 00:36:51.320 If you take a master's in in analytics, you know you're going to go 506 00:36:51.400 --> 00:36:53.719 somewhere in business. I'm quite sure you're going to have a pretty good career 507 00:36:53.800 --> 00:36:59.159 path ahead of you, and it's going to be constantly changing. So you 508 00:36:59.480 --> 00:37:04.000 understand the data understanding. As you've said, the use case is the base 509 00:37:04.119 --> 00:37:07.320 camp of use cases, and that's of course what you're providing with fourteen eighty 510 00:37:07.360 --> 00:37:09.599 six labs, right, Yeah, so that the idea. You know, 511 00:37:09.679 --> 00:37:16.280 we're very fortunate we have a company named Liaison based in Boston, an ed 512 00:37:16.360 --> 00:37:24.119 tech company really really focus on helping the student journey and become successful. That 513 00:37:24.239 --> 00:37:29.119 solved the value in O Thoughts and they purchased O Thought about two and a 514 00:37:29.159 --> 00:37:31.719 half years ago. I still remain as a president of the AI division, 515 00:37:31.800 --> 00:37:37.519 really looking at the next generation of products and so, you know, with 516 00:37:37.800 --> 00:37:43.320 that, I also have a lot of flexibility to be creative. And so 517 00:37:43.440 --> 00:37:47.840 what I wanted to do in combining sort of what Liaison does in creativity around 518 00:37:47.880 --> 00:37:54.199 the data and analytics side and the resources, especially the young sort of very 519 00:37:54.400 --> 00:38:00.880 energetic and very knowledge thirsty students, you know, put together an entrepreneurial lab 520 00:38:01.079 --> 00:38:07.639 where we will develop companies at the intersection of data and analytics. And that's 521 00:38:07.559 --> 00:38:10.400 you know, fourteen eighty six Labs is all about that. So we have 522 00:38:10.519 --> 00:38:16.559 twenty one interns right now and they're building our first company, which is called 523 00:38:16.559 --> 00:38:22.320 Blue Street Data. And Blue Street Data is aimed directly at how do we 524 00:38:22.440 --> 00:38:28.119 make it easier for the buyers of data to understand what they're buying and what 525 00:38:28.239 --> 00:38:31.599 they need to buy in order to make these use cases work? What powers 526 00:38:31.679 --> 00:38:35.559 those use cases, how do we understand it, and how do we get 527 00:38:35.599 --> 00:38:40.280 the right best, highest quality, lowest price data to power those use cases. 528 00:38:42.320 --> 00:38:45.199 Yeah, and that's really important. I can tell you that in my 529 00:38:45.559 --> 00:38:49.199 industry, I've tried a couple times over the years, and I was nervous 530 00:38:49.280 --> 00:38:53.239 each time to buy data sets of contact information. It's a very dicey thing 531 00:38:53.320 --> 00:38:58.079 to do. You typically get very low quality data and it's just hard to 532 00:38:58.199 --> 00:39:00.880 know. I mean, you get all these different companies offering it, it's 533 00:39:00.920 --> 00:39:02.119 like, well, how did you get that data? Where did you find 534 00:39:02.159 --> 00:39:07.760 it? These days, there are technologies like seamless AI that we use somewhat 535 00:39:07.800 --> 00:39:12.159 extensively, which are quite compelling. It's almost like an engine that spins up 536 00:39:12.199 --> 00:39:15.519 a real time Yellow Pages for you to find contact information. And it's real 537 00:39:15.599 --> 00:39:19.960 time. It's not just a repository like a big data set. It's a 538 00:39:20.000 --> 00:39:23.320 set of algorithms that in the moment will reach out and capture information. And 539 00:39:23.400 --> 00:39:27.440 they do persist a lot of it. But you have to ask yourself the 540 00:39:27.519 --> 00:39:30.320 question and like how good is this data? And that's the service you're providing, 541 00:39:30.440 --> 00:39:34.880 is being able to act as a liaison or a shepherd or a schirper 542 00:39:35.000 --> 00:39:38.239 or something to help people understand what are the use cases when do you use 543 00:39:38.320 --> 00:39:42.039 this stuff for? What purpose do you use this stuff? How do you 544 00:39:42.280 --> 00:39:44.639 use it? You have to be careful about how you use these things. 545 00:39:44.719 --> 00:39:47.559 But I can tell you buying email databases, buying email lists very bad idea. 546 00:39:49.119 --> 00:39:52.480 Like you cannot just throw that stuff into production. I mean, if 547 00:39:52.480 --> 00:39:54.360 anything, you can use it as a starting point to start looking around and 548 00:39:54.440 --> 00:39:58.519 finding people. But there's a lot of work to be done, folks. 549 00:39:58.559 --> 00:40:00.800 There's a lot of time and effort that goes into this, and you have 550 00:40:00.880 --> 00:40:02.000 to be careful about how you spend your time. But don't touch that. 551 00:40:02.039 --> 00:40:12.519 That'll be right back. You're listening to Inside Analysis. Respect, Welcome back 552 00:40:12.559 --> 00:40:19.719 to Inside Analysis. Here's your host, Eric Tabanac to share all right, 553 00:40:19.760 --> 00:40:23.440 folks, back here on Inside Analysis with Andy Hannah of fourteen eighty six Labs 554 00:40:23.480 --> 00:40:28.119 and oh Thought and the University of Pittsburgh and of course Blue Street Data and 555 00:40:28.519 --> 00:40:31.239 Andy, you were just staying in the break there that you are now looking 556 00:40:31.280 --> 00:40:36.199 at data quality through some new lenses. That's one thing that machine learning is 557 00:40:36.400 --> 00:40:38.639 very good at, by the way, is finding mistakes. You know, 558 00:40:38.760 --> 00:40:44.320 people are looking for use cases for lllms. Finding mistakes in your code is 559 00:40:44.440 --> 00:40:45.519 really good stuff. You can just throw a bunch of goats, say where's 560 00:40:45.519 --> 00:40:50.039 the mistake, It'll find it. So there are lots of things these engines 561 00:40:50.119 --> 00:40:53.519 do that are not just text or image generative in nature. That's for fodder 562 00:40:53.559 --> 00:40:55.199 for another show. We'll do a show in a couple of weeks on that. 563 00:40:55.679 --> 00:41:00.400 But let's let's drive really dive into the quality converse because quality is so 564 00:41:00.639 --> 00:41:05.920 important with data. Think about when you get an outreach they misspell your name 565 00:41:06.079 --> 00:41:09.400 or something that's not a good customer experience. What are you doing to improve 566 00:41:09.519 --> 00:41:14.519 quality and to vet the quality of data sets. Yeah, so I think 567 00:41:14.599 --> 00:41:17.400 this is a really interesting topic because when we first started to attack this angle, 568 00:41:17.519 --> 00:41:22.519 we went to all the typical sort of metrics around quality, the consistency, 569 00:41:22.719 --> 00:41:27.360 the comparability, the timeliness, et cetera, everything that everybody thinks about. 570 00:41:27.960 --> 00:41:30.599 And as we were exploring it, we came to realize that, hey, 571 00:41:30.880 --> 00:41:37.039 so much of the quality of data is dependent on the particular use case 572 00:41:37.679 --> 00:41:42.639 and how important that use case is to the business. So it becomes very 573 00:41:42.719 --> 00:41:45.440 individualized, it becomes very difficult. Like you said, Eric, it's there's 574 00:41:45.639 --> 00:41:51.639 great technologies out there to do anominally detection, right or to define missing values 575 00:41:51.760 --> 00:41:55.320 or whatever it may be. But you often can't use those techniques unless you 576 00:41:55.519 --> 00:41:59.440 get the full data set. You can do it on a sample, but 577 00:41:59.519 --> 00:42:05.079 as we know, sample rarely represents the full data set. So you know, 578 00:42:05.199 --> 00:42:07.920 one of our advisors is Malcolm Hawker, who you know well, and 579 00:42:08.119 --> 00:42:13.760 we started thinking about how do we need to approach quality? And this takes 580 00:42:13.840 --> 00:42:17.199 me back to sort of some of my days in the company called Plextronics, 581 00:42:17.239 --> 00:42:22.000 where you look at the supplier, right, and you start to evaluate what 582 00:42:22.239 --> 00:42:30.280 is the process that the supplier has in order to ensure that it's delivering a 583 00:42:30.519 --> 00:42:35.400 top notch product, right, So it is the processes behind that. How 584 00:42:35.440 --> 00:42:38.559 many sources do you use before you say, hey, I have a valid 585 00:42:38.760 --> 00:42:45.039 sleeve of data related to a particular attributes, So it is that that as 586 00:42:45.079 --> 00:42:47.760 a transparent do they tell us how they do it? What's the source of 587 00:42:47.840 --> 00:42:52.920 the data, is an ethically sourced data, what's the quality of the company 588 00:42:52.000 --> 00:42:58.599 itself? And then of course we look at the you know, expert opinion 589 00:42:58.719 --> 00:43:02.039 about the day database sort of the data set or the data product, depending 590 00:43:04.239 --> 00:43:07.320 of people who've actually used it before, the pros and the cons. So 591 00:43:07.440 --> 00:43:13.800 when you put that sort of how we make the product. Together with expert 592 00:43:13.960 --> 00:43:17.440 opinion, you have a much different view of quality than we typically look at 593 00:43:17.800 --> 00:43:23.920 when we're trying to find these anomalies or problems with the data sets. So 594 00:43:24.079 --> 00:43:29.960 it's like, to a certain extent, it's an expert crowd sourced view of 595 00:43:30.119 --> 00:43:35.159 quality because it's not just random let's say, Yelp reviewers, and many of 596 00:43:35.199 --> 00:43:38.440 those can be fake. No, you have trusted sources that you rely upon 597 00:43:38.719 --> 00:43:45.800 to give you sophisticated analysis of data sets. Right that coupled with are they 598 00:43:45.920 --> 00:43:50.960 transparent in the process that they use to make the product? If you know, 599 00:43:51.000 --> 00:43:53.519 if you think about the material science world, you know you have your 600 00:43:53.679 --> 00:44:00.119 the spec of the product has to be on and you go in explore the 601 00:44:00.239 --> 00:44:05.599 manufacturing process of the manufacturer. That's what I say is all about. We 602 00:44:05.760 --> 00:44:10.199 need that ISO type of concept here with data production. It's just like thinking 603 00:44:10.280 --> 00:44:16.639 about a product that any company might produce. Well, that's interesting because you're 604 00:44:16.719 --> 00:44:23.239 kind of hinting at or alluding to observability. Right When we talk about observability 605 00:44:23.280 --> 00:44:28.440 in the data world, what you're really talking about is getting windows into streams 606 00:44:28.480 --> 00:44:32.400 of information and then being able to correlate those to understand what happened. So 607 00:44:34.239 --> 00:44:36.840 this is what I love about open source. That's why I always tout open 608 00:44:36.920 --> 00:44:40.440 source and I'm a huge fan for lots of different reasons. One is because 609 00:44:40.639 --> 00:44:45.920 many eyes make few the errors or as the old expression goes bad, code 610 00:44:45.000 --> 00:44:49.039 goes away. Right, that's one of the benefits of open source. Now 611 00:44:49.079 --> 00:44:52.079 there are downsides, which is it's hard to make money, Like if you're 612 00:44:52.559 --> 00:44:54.559 open source in the code, how do you ensure that you get the money. 613 00:44:55.440 --> 00:45:00.960 We were joking a couple of years ago that Amazon was strip mining the 614 00:45:00.000 --> 00:45:04.519 open source community and not doing a lot for it. They have since changed 615 00:45:04.519 --> 00:45:07.039 their tune somewhat and now they kind of understand, which gets us back to 616 00:45:07.199 --> 00:45:12.119 ethics, right into being ethical and how you run your business, how you 617 00:45:12.199 --> 00:45:16.039 run your operations, and you know, there are these sort of competing virtuous 618 00:45:16.519 --> 00:45:21.440 and vicious cycles, it seems to me, and we want to put the 619 00:45:21.559 --> 00:45:27.519 balance of power in the virtuous circles where good virtue breeds good virtue, which 620 00:45:27.559 --> 00:45:30.920 breeds good virtue and it keeps going up. But there is always this downward 621 00:45:30.000 --> 00:45:34.519 pressure of the vicious cycles. And how do you solve that. I think 622 00:45:34.559 --> 00:45:40.159 you solve it through good ethical constructs and transparency and just ongoing work. Basically, 623 00:45:40.239 --> 00:45:44.400 it's never going to be solved. To keep hammering away at it, 624 00:45:44.519 --> 00:45:47.159 right. Yeah, And the transparency, as we've talked about, I think 625 00:45:47.280 --> 00:45:52.360 every single segment here is critically imparent. It's transparency and what the data is, 626 00:45:52.519 --> 00:45:55.800 how you're using it, how it's produced. But you said something that's 627 00:45:55.880 --> 00:46:00.639 kind of interesting that I think a lot about, which is you said use 628 00:46:00.639 --> 00:46:05.360 the words sleeves of data. And the more that I talk, you know, 629 00:46:05.440 --> 00:46:08.199 I probably talked to a dozen people every week in this in the industry 630 00:46:08.239 --> 00:46:13.760 about external data, and one of the most common things are that I have 631 00:46:13.920 --> 00:46:19.599 to typically as a buyer, buy bundles of data. I want streams of 632 00:46:19.719 --> 00:46:22.039 data. I want particular data elements. Yeah, I have. I'll have 633 00:46:22.159 --> 00:46:27.039 to find a way to make sure my matching algorithms are able to to pen 634 00:46:27.199 --> 00:46:30.000 my existing database. I get that, But why do I have to buy 635 00:46:30.199 --> 00:46:35.000 the whole anchilada when I just want you know, the bup, you know, 636 00:46:35.119 --> 00:46:37.760 the uh uh, you know, the guacamole or whatever. It might 637 00:46:37.800 --> 00:46:42.199 be one of bad analogy, but you get what I mean is it's like, 638 00:46:42.360 --> 00:46:45.679 so you want that, you want the ability to slice down into the 639 00:46:45.840 --> 00:46:52.000 individually into the granular level, and that's we really don't have that. That's 640 00:46:52.119 --> 00:46:55.639 not the way that data and information has been sold into pass and so we 641 00:46:55.880 --> 00:47:00.480 have to move to that, to that type of granularity, to the individual 642 00:47:00.599 --> 00:47:04.840 level of data element. And that's going to be hard. That's going to 643 00:47:04.880 --> 00:47:07.119 be hard for a lot of the existing companies who are doing it the old 644 00:47:07.199 --> 00:47:12.719 way to transform. But if we do it in the right way, we're 645 00:47:12.760 --> 00:47:15.719 going to be able to say, oh, for this use case, here's 646 00:47:15.760 --> 00:47:19.960 a particular element you need to buy. You can buy the best data and 647 00:47:20.079 --> 00:47:23.239 it says a streame off of this company that's never sold data in the market 648 00:47:23.320 --> 00:47:29.079 before. Because so we give that opportunity for those types of companies to come 649 00:47:29.119 --> 00:47:32.639 to market with their data with the particular use that's better than anything else that's 650 00:47:32.679 --> 00:47:38.280 out there. That's the way to think about restructuring of the data markets, 651 00:47:38.400 --> 00:47:44.320 which we desperately need if we're going to move to the next level of value 652 00:47:44.440 --> 00:47:50.000 from data. That's very interesting I was mentioning before, you're seeing early indicators 653 00:47:50.079 --> 00:47:53.159 for this kind of of technique or of standard, if you will, and 654 00:47:53.280 --> 00:47:58.400 one of which is like Google with your timeline and how Google will now give 655 00:47:58.440 --> 00:48:00.239 you your timeline, right. I mean, I think Twitter allows you to 656 00:48:00.320 --> 00:48:05.320 download your tweets. Other platforms allow you to kind of download things, and 657 00:48:05.400 --> 00:48:08.400 there is value to that, but being able to reflect back to these Amazon 658 00:48:08.480 --> 00:48:12.599 is actually pretty good at this at being able to capture all of your transactions 659 00:48:12.800 --> 00:48:15.840 very quickly show you where they are so you can go find things. So 660 00:48:15.920 --> 00:48:17.000 you think about all this stuff you used to have to keep track of yourself, 661 00:48:17.480 --> 00:48:20.639 and now you realize, okay, well they're going to keep track of 662 00:48:20.679 --> 00:48:23.000 that for me, like a web portal for the hospital, for example, 663 00:48:23.039 --> 00:48:27.719 I don't have to write down everything from my wife's appointments. I know it's 664 00:48:27.800 --> 00:48:30.119 up in the cloud, so I can go there. And I think the 665 00:48:30.239 --> 00:48:34.320 more we learn to trust those sources and rely on those sources, the more 666 00:48:34.400 --> 00:48:37.400 we'll be able to focus on what we're trying to do and not worry about 667 00:48:37.400 --> 00:48:39.440 trying to do all the other stuff that other people are doing for us. 668 00:48:39.559 --> 00:48:45.559 Right, Yeah, and it gets down you're right. So the way that 669 00:48:45.719 --> 00:48:51.679 we capture information has dramatically improved over the past decade. The way that we 670 00:48:51.840 --> 00:48:55.119 store it so that we can track it down to any individual is what gives 671 00:48:55.199 --> 00:48:58.599 us the power to be able to do this. So it's going to open 672 00:48:58.719 --> 00:49:01.280 up a whole bunch of different revenue models. It's going to open up a 673 00:49:01.320 --> 00:49:07.000 whole bunch of different companies. I mean, we're really flooded with Jenai companies 674 00:49:07.079 --> 00:49:12.360 right now. But I think the underlying sort of hidden gold is what companies 675 00:49:12.400 --> 00:49:16.320 are doing very creative things with the data to make it much more usable. 676 00:49:17.039 --> 00:49:22.400 And when you can combine that value of that usable data with these technologieses, 677 00:49:22.519 --> 00:49:28.519 just like we've seen over the past forty years, you're going to companies will 678 00:49:28.639 --> 00:49:34.039 merge that will disrupt their industries, will disrupt the other companies that they compete 679 00:49:34.079 --> 00:49:37.199 against. Yeah, no, that's right, and it's I think it's going 680 00:49:37.280 --> 00:49:43.960 to be amazing. I think that all the tumblers are aligning right now, 681 00:49:44.639 --> 00:49:49.039 and it's kind of wild, you know. I think the more people get 682 00:49:49.199 --> 00:49:52.800 access to useful data sets like a data sleeve, as you're talking about, 683 00:49:52.920 --> 00:49:57.760 just a little piece to fill in a gap, the more success you have 684 00:49:57.920 --> 00:50:00.719 with that, the better. The problem is that people do get burned. 685 00:50:00.760 --> 00:50:04.719 I mean, one of my partners does content syndication, and I trust this 686 00:50:04.840 --> 00:50:07.800 guy as much as I trust any human being in the world. He's incredibly 687 00:50:07.880 --> 00:50:10.480 professional. He's been around a long long time, and the projects they've done 688 00:50:10.480 --> 00:50:14.239 with us have been spectacular. I mean, the quality of leads is amazing. 689 00:50:14.599 --> 00:50:16.159 But it's still hard to sell. Why because people have been burned in 690 00:50:16.239 --> 00:50:20.079 the past, and once you get burned, you know, once bitten, 691 00:50:20.119 --> 00:50:22.800 twice shy. Basically, it's hard to kind of move past that. I 692 00:50:22.880 --> 00:50:28.239 mean with chatbots, you're kind of seeing this right now. Chatbots are coming 693 00:50:28.320 --> 00:50:30.920 back with some vigor. They first came out, I mean it first came 694 00:50:30.920 --> 00:50:34.840 out like fifteen years ago. In the earliest days, they were just dreadful, 695 00:50:35.719 --> 00:50:37.840 right, it was just horrible. It's like, what is going on 696 00:50:37.960 --> 00:50:40.840 here? These people are making me crazy. So it's like they're you know, 697 00:50:40.920 --> 00:50:44.639 well, it's like Pittsburgh still to this day to people who have never 698 00:50:44.719 --> 00:50:45.480 been here, like, oh, it's not a dirty city. I'm like, 699 00:50:46.400 --> 00:50:49.320 oh my god, No, it's not a dirty city. It hasn't 700 00:50:49.320 --> 00:50:52.480 been thirty from it was from the steel days and that ended, you know, 701 00:50:52.519 --> 00:50:55.159 in terms of the toxins and all that stuff. That's this distant past 702 00:50:55.280 --> 00:50:59.480 now, but it takes a long time for these things to die. Will 703 00:50:59.480 --> 00:51:01.880 podcast segment's coming up next but don't chut that out. You're listening to Inside 704 00:51:01.880 --> 00:51:08.440 Analysis. All right, folks, back your time of the podcast. Bonus 705 00:51:08.519 --> 00:51:13.199 segment here with Andy Hannah of fourteen eighty six Labs and All Thought and the 706 00:51:13.320 --> 00:51:17.119 University of Pittsburgh, and we're talking all about AI and data and ethics and 707 00:51:17.400 --> 00:51:22.880 speaking of everyone's talking about responsible AI and ethical AI. And you know it's 708 00:51:22.920 --> 00:51:25.760 good because we need to have these conversations. And you, my friend, 709 00:51:25.800 --> 00:51:30.440 were recently given the chair of what is it, the University of Pittsburgh's Responsible 710 00:51:30.519 --> 00:51:34.559 Data Science Board? Is that right? Tell us about that? That is 711 00:51:34.679 --> 00:51:39.119 right? So we're we're really fortunate at the University of Pittsburgh where we have 712 00:51:39.239 --> 00:51:45.639 a lot of individuals that are focused on data science and artificial intelligence. What 713 00:51:46.239 --> 00:51:52.760 we're focused on in a responsible data science community is three things. How do 714 00:51:52.880 --> 00:51:59.400 we make sure that the infrastructure behind all these models, etc. Are responsible 715 00:51:59.519 --> 00:52:05.440 From a curriculum perspective, from a research perspective, from a workforce development perspective. 716 00:52:06.079 --> 00:52:10.000 So we want to lead the nation on how we think about the deployment 717 00:52:10.239 --> 00:52:15.719 of all these technologies, regardless of whether you're in academia or you're in business. 718 00:52:15.360 --> 00:52:22.440 My responsibility as chair of the Advisory Board is to bring industry into the 719 00:52:22.559 --> 00:52:28.400 conversation. So that very very thoughtful approach that says that we can't do this 720 00:52:28.480 --> 00:52:31.440 in a vacuum. So if we're working on then we literally use the words 721 00:52:31.480 --> 00:52:37.719 we're working on use cases and data sets to show how you can responsibly use 722 00:52:37.960 --> 00:52:44.360 the techniques that we've been talking about and solve real problems and so and those 723 00:52:44.400 --> 00:52:49.119 real problems come from industry. So we're focused on retail, finance, and 724 00:52:49.280 --> 00:52:59.960 healthcare and i'll what i'll call civic applications of this technology. That's pretty cool. 725 00:53:00.000 --> 00:53:04.360 Well, you know when I think about policies, and I'm glad that 726 00:53:04.400 --> 00:53:07.920 you're saying you're bringing the business world into this, because that's really what it 727 00:53:07.000 --> 00:53:10.239 takes, because you know, you and I can have all sorts of theories 728 00:53:10.280 --> 00:53:14.679 about how things operate in a particular industry, but until you sit down with 729 00:53:14.760 --> 00:53:17.280 people from that business, you're probably not going to be able to figure out 730 00:53:17.360 --> 00:53:22.360 some of the really critical components of processes and workflows. And they will tell 731 00:53:22.400 --> 00:53:24.679 you. I mean, I remember, just as an example, many years 732 00:53:24.679 --> 00:53:28.880 ago, I was on as at a meeting I used to represent the downtown 733 00:53:28.880 --> 00:53:32.679 development district of New Orleans and we're in a meeting with the new executive director 734 00:53:32.880 --> 00:53:37.719 and a whole bunch of other people's stakeholders are there, including a lieutenant from 735 00:53:37.719 --> 00:53:39.920 the police force and a bunch of other people, and the new executive director 736 00:53:40.039 --> 00:53:42.880 was like, yeah, we decided we're going to come up with our own 737 00:53:42.880 --> 00:53:45.800 police force and they're going to handle these stuff downtown. And you can imagine 738 00:53:45.800 --> 00:53:50.199 the lieutenant of the police force in New Orleans said, excuse me, what 739 00:53:50.280 --> 00:53:52.400 are you going to do? That's our responsibility. Why what are you talking 740 00:53:52.400 --> 00:53:55.480 about here? They've never even broached the subject with them, and it's like, 741 00:53:55.639 --> 00:53:59.119 that's kind of a problem. You know, if you sit down with 742 00:53:59.239 --> 00:54:02.760 people in private first and have conversations or even have hearings, for example, 743 00:54:04.400 --> 00:54:07.440 to vet these issues, that's the way to get there and make sure you 744 00:54:07.519 --> 00:54:12.760 have the stakeholders at the table because they're the ones who are going to recognize 745 00:54:12.760 --> 00:54:15.039 those red flags. It sounded like a great idea in the other boardroom, 746 00:54:15.119 --> 00:54:19.599 but when you share it in the bigger context, you realize, oh, 747 00:54:19.679 --> 00:54:22.920 I guess there are some problems with that. That's the point of having these 748 00:54:22.000 --> 00:54:25.360 meetings, right, yeah, one hundred percent, And you know, we 749 00:54:27.000 --> 00:54:31.599 have to look at it from every angle in terms of responsibility. And so 750 00:54:31.800 --> 00:54:37.440 the lens when I think about that concept are responsible. I really think about 751 00:54:37.440 --> 00:54:42.320 it from three different lenses. The first one is purpose. You know, 752 00:54:42.559 --> 00:54:45.639 why are we using this technology? What is the big reason why we're using 753 00:54:45.719 --> 00:54:51.719 this technology? Is it to solve organizational societal issues? Or is it to 754 00:54:51.840 --> 00:54:55.280 cause some type of catastrophic harm? Right? I mean, that's the easiest 755 00:54:55.320 --> 00:54:59.679 way to think about it. So purpose is number one, So how that 756 00:55:00.079 --> 00:55:06.480 that perspective. The second one is all about ethics, So it's what is 757 00:55:06.920 --> 00:55:12.440 what is the intent of the individuals involved in the project. Are they looking 758 00:55:13.039 --> 00:55:20.599 to benefit the society or benefit the company, or benefit or everything that they're 759 00:55:20.679 --> 00:55:23.159 doing. Is it is leaning towards the good as opposed to the bad? 760 00:55:23.920 --> 00:55:31.480 And then the last piece is about you avoiding bias or having anyone treated unfairly. 761 00:55:31.960 --> 00:55:37.800 And that's a really interesting piece of it because that's where we go back 762 00:55:37.800 --> 00:55:40.679 all the way to the beginning of our conversation on a diagnosis. Where are 763 00:55:40.840 --> 00:55:45.840 people being disadvantaged because of this technology? And so if we can combine that 764 00:55:46.039 --> 00:55:52.280 purpose and ethics and fairness together, we're going to have a very powerful and 765 00:55:52.400 --> 00:55:59.159 reliable and trustworthy set of technologies and data that we can then build the next 766 00:55:59.239 --> 00:56:05.920 generations business own. Yeah, you just reminded me of manual cons categorical imperative. 767 00:56:06.119 --> 00:56:08.920 You may you may recall act only on that maxim which you can at 768 00:56:08.960 --> 00:56:15.360 the same time will as universal law, which is a nuanced approach of do 769 00:56:15.480 --> 00:56:17.519 as you would be done by. Basically, it's like, if you were 770 00:56:17.559 --> 00:56:22.400 going to engage in a policy, ask yourself, would it be reasonable for 771 00:56:22.559 --> 00:56:24.480 every other company to engage in the same policy? And if the answer is 772 00:56:24.599 --> 00:56:29.400 yes, then ethically speaking, you're on pretty solid ground, right m h. 773 00:56:29.920 --> 00:56:32.239 And first, you know, I like the physician's oath, right, 774 00:56:32.480 --> 00:56:37.840 So first we do no harm, right, So if you literally have that 775 00:56:37.000 --> 00:56:42.719 in your mindset, and so I think that's you know, the diversity of 776 00:56:42.840 --> 00:56:45.960 people around the problem is a really critical one. So if you have a 777 00:56:46.039 --> 00:56:51.239 bunch of people that are the same gender, same ethnicity, you know, 778 00:56:51.440 --> 00:56:53.719 same level of wealth, whatever it is, trying to solve a problem, 779 00:56:53.840 --> 00:56:59.039 you're not going to get a diverse view of the problem. So the people 780 00:56:59.119 --> 00:57:02.880 around the problem than the techniques that we use to determine whether or not somebody 781 00:57:02.920 --> 00:57:09.440 could be treated unfairly the combination of those two things very powerful, and you 782 00:57:09.599 --> 00:57:14.800 have to kind of maintain an ongoing balancing act, right, because you can 783 00:57:14.880 --> 00:57:19.199 also overreact to things. And that's something we see a lot in our world 784 00:57:19.320 --> 00:57:21.760 is pendulum swings one way and then it swings the other way, and it 785 00:57:21.800 --> 00:57:23.440 swings one way and it swings the other way. You know, for every 786 00:57:23.519 --> 00:57:28.199 action there's an equal and opposite reaction, So you kind of have to take 787 00:57:28.280 --> 00:57:31.679 all that into consideration and just be reasonable. Right. There's something to be 788 00:57:31.800 --> 00:57:37.320 said for being a reasonable person or understanding there are differences and you know, 789 00:57:37.440 --> 00:57:39.000 not everyone's out to get you and all these kinds of things. You do 790 00:57:39.119 --> 00:57:42.920 have to keep that all in mind. But I think the real key is, 791 00:57:43.039 --> 00:57:46.480 as you suggest, bring an ethically oriented mindset, what are we trying 792 00:57:46.519 --> 00:57:51.199 to do, how are we trying to do it? And then honestly gauging 793 00:57:51.320 --> 00:57:53.599 the success is it working? And it is nothing like the law of unattended 794 00:57:53.639 --> 00:57:58.079 consequences to give you a real ham dinger, or you think you're doing the 795 00:57:58.159 --> 00:58:00.880 right thing and it all falls apart. I mean, have you ever seen 796 00:58:00.920 --> 00:58:02.679 the video of Yellowstone when they let the wolves back in? Have you seen 797 00:58:02.719 --> 00:58:06.639 that video? No, I'll have them mind. Look it up, man, 798 00:58:06.760 --> 00:58:10.199 it'll absolutely blow your mind. They did a study and they brought wolves 799 00:58:10.280 --> 00:58:15.679 back to Yellowstone National Park and they said the results were absolutely mind blowing. 800 00:58:16.079 --> 00:58:20.639 You have to get into it to understand, but basically it reset the balance 801 00:58:20.719 --> 00:58:23.400 of power amongst all the different animals that were there. The creeks came back, 802 00:58:23.480 --> 00:58:27.639 all sorts of things happened that absolutely blew people's minds, which shows you 803 00:58:27.719 --> 00:58:30.360 it's kind of like the butterfly's wings effect. Right, One little change can 804 00:58:30.400 --> 00:58:34.880 have a very big impact, but you have to be open to the change, 805 00:58:34.920 --> 00:58:37.360 and you have to be open to tracking the change and just being realistic. 806 00:58:37.599 --> 00:58:39.760 And that's what business is all about. Right In the business, you 807 00:58:39.920 --> 00:58:44.880 have to be pragmatic, like if you go too far in one direction, 808 00:58:45.440 --> 00:58:47.440 you're not going to be in business anymore. But final thoughts, how do 809 00:58:47.519 --> 00:58:52.679 people find out more about Blue Street Data and fourteen eighty six Labs. Yeah, 810 00:58:52.840 --> 00:58:58.039 so the easy go to www Dot blue Street Data dot com or www 811 00:58:58.280 --> 00:59:01.679 Dot fourteen eighty six Labs dot to learn about us from fourteen eighty six, 812 00:59:01.760 --> 00:59:07.440 how we're building next generation companies with really great students, and of course Blue 813 00:59:07.440 --> 00:59:15.760 Street Data, how we facilitate the purchase of high quality, right price NBC 814 00:59:15.960 --> 00:59:22.920 News on CACAA Lomel sponsored by Teamsters Local nineteen thirty two Protecting the Future of 815 00:59:22.039 --> 00:59:37.440 Working Families Teamsters nineteen thirty two dot org. Thanks for tuning in for disposition 816 00:59:37.519 --> 00:59:40.639 of Justice Watch with Attorney Zulu Adli. I am Attorney Zulu Adli with a 817 00:59:40.840 --> 00:59:45.719 Justice Watch crew Rosa Nunez, Michael blau Clark, doctor Kilbasher, and Andrea 818 00:59:45.840 --> 00:59:51.800 Rohdeman. This week, like every week, we'll be discussing critical legal and 819 00:59:51.960 --> 00:59:57.559 social justice issues that are impacting our community. This week we'll be talking about 820 00:59:57.760 --> 01:00:06.119 the O. J. Simpson actually and the actual legacy of actually O J. 821 01:00:06.320 --> 01:00:08.239 Simpson. And I think most of