Latest / The Adults In The AI Room / Dan Turchin on AI and the Future of Work - Episode #21

Transcript

Dan Turchin on AI and the Future of Work - Episode #21 · The Adults In The AI Room

• 0:36
  SPEAKER_01: Good morning, good afternoon, good evening, wherever your time zone is.
• 0:41
  SPEAKER_01: Welcome to today's podcast.
• 0:43
  SPEAKER_04: I'm Chris Featherstone.
• 0:45
  SPEAKER_01: So, Chris, uh uh, I know uh we uh we've been uh recording a lot of these uh pretty consistently these days.
• 0:51
  SPEAKER_01: So thanks for your your time on that.
• 0:52
  SPEAKER_01: I love it.
• 0:53
  SPEAKER_01: Um what's uh how's how's life been treating you these days?
• 0:56
  SPEAKER_04: You know, I gotta tell you, you know, it's um you know, the the fall is always super spectacular.
• 1:04
  SPEAKER_04: I love the fall time because it's you know we we switch over from all the summer activities and then gear up for you know for skiing down this the slopes as fast as possible.
• 1:15
  SPEAKER_04: And you know, so I start getting ski fever and want the snow to come, but I don't want the snow to come right away, you know what I mean?
• 1:24
  SPEAKER_01: So I I love this time of year.
• 1:26
  SPEAKER_01: This is my favorite.
• 1:27
  SPEAKER_01: I I love the fall weather, I love this, and we've had such a spectacular um fall here.
• 1:32
  SPEAKER_01: The the foliage has absolutely been been uh spectacular uh in New England.
• 1:37
  SPEAKER_01: So um, well, let's jump into our podcast.
• 1:40
  SPEAKER_01: So uh today we have a self-proclaimed type A adrenaline junkie.
• 1:45
  SPEAKER_01: Uh our next guest is passionate about the future work, building great teams that build great products and solve hard problems and change lives.
• 1:52
  SPEAKER_01: He's the CEO of People Rain, a company that is using AI and natural language processing to make technology work for humans.
• 1:59
  SPEAKER_01: Joining us from San Francisco, please welcome Dan Turchin.
• 2:03
  SPEAKER_01: Thanks for thanks for joining us today, Dan.
• 2:05
  SPEAKER_04: Gents, good to be here.
• 2:07
  SPEAKER_04: Dan, we still have the podcast.
• 2:09
  SPEAKER_04: We still haven't figured out that you know the cool, you know, like audio insertion of like the uh the loud applause on the back end that once we introduce you, that's coming at some point, right?
• 2:20
  SPEAKER_02: None needed.
• 2:21
  SPEAKER_02: I mean uh let me say something that matters first, and then you can choose whether or not to applause.
• 2:27
  SPEAKER_02: Yeah, right, exactly.
• 2:29
  SPEAKER_01: So, you know, Dan, uh you had me on your uh podcast, which was great.
• 2:34
  SPEAKER_01: And uh uh and you know, we talked a lot, we we certainly share a lot of the same thinking around uh AI, around virtual assistance, around information management, around the use of ontologies.
• 2:46
  SPEAKER_01: But before we get into that, maybe um you could talk a little bit about your journey to uh where you are today.
• 2:52
  SPEAKER_01: You know, how did you get to doing what you're doing?
• 2:54
  SPEAKER_01: And uh, you know, tell us a little bit about kind of your your early interests in uh in the field and really give us a little bit of a thumbnail of of the world, according to Dan.
• 3:06
  SPEAKER_02: Yeah, well, Seth, let me start by saying we've published over 150 episodes of AI in the future of work.
• 3:13
  SPEAKER_02: And uh it's one of the most downloaded podcasts about the topic.
• 3:18
  SPEAKER_02: And of those 150 plus episodes, there's only one where I got to geek out about one of my favorite topics, which is domain ontologies and information architecture.
• 3:28
  SPEAKER_02: So uh man, you and I were like uh two kids in a candy shop.
• 3:32
  SPEAKER_02: That was really a fun conversation.
• 3:35
  SPEAKER_01: Yeah, great.
• 3:36
  SPEAKER_01: I really did enjoy it.
• 3:37
  SPEAKER_01: I I did a little more talking, so that's why I wanted to give you the the podium today to share more of your ideas because we were so well aligned.
• 3:44
  SPEAKER_02: So tell me gratuitous plug.
• 3:46
  SPEAKER_02: I I'd encourage uh uh all of all of your listeners to go listen to Seth's conversation.
• 3:52
  SPEAKER_02: He put on a master class.
• 3:55
  SPEAKER_01: Thank you.
• 3:56
  SPEAKER_01: But you're you're you're too kind.
• 3:57
  SPEAKER_01: Thank you for the buttons definitely check out uh the people rain, and we'll have uh we'll have a uh you know show notes today.
• 4:05
  SPEAKER_04: You know, Dan, the you know, like you know, the the the human experience is always uh you know super, super important.
• 4:12
  SPEAKER_04: And I think, you know, given what you guys like to geek out on, you know, those are the foundations.
• 4:19
  SPEAKER_04: And I would love, you know, like as you're as you're kind of you know talking about, you know, maybe your your journey, why?
• 4:26
  SPEAKER_04: Why, why are the ontologies, why are taxonomies, why are those so why do those strike a special place in your in your heart and your soul, you know, to because I know that it's it's the foundation for performance, but why?
• 4:39
  SPEAKER_04: You know, I mean that's that's probably why a lot of people would love to know, because they miss it in the broad business world for AI.
• 4:46
  SPEAKER_02: That's absolutely the right question to ask.
• 4:49
  SPEAKER_02: Too often in the technology field, we focus on the how and we miss the why.
• 4:54
  SPEAKER_02: And for me, my 25-year journey has always been about the why, and more important, about the who.
• 5:02
  SPEAKER_02: And I'm on a I'm on a personal mission to impact a billion lives and started.
• 5:08
  SPEAKER_02: Uh, this is my seventh company, People Rain, applying the principles of AI and machine learning to improve the employee experience, to make lives better, uh, to make humans better, with not artificial intelligence, but augmented intelligence using technology to make humans better.
• 5:28
  SPEAKER_02: And so it started a while back in the 90s with a company I founded called Aero Prize that was solving a similar problem to what People Rain solves, but for field technicians who needed to do some work in the field, but then they didn't get credit for doing the work until they got back to a to a desk.
• 5:46
  SPEAKER_02: And oftentimes they would print out a paper ticket and uh you know they'd have to file a report when they got back to uh to headquarters to get credit for whatever you know end user they were supporting.
• 5:59
  SPEAKER_02: And so oftentimes, you know, they were less efficient, and you know, the paperwork consumed time that otherwise could have been spent in the field helping more employees.
• 6:09
  SPEAKER_02: And so me and my co-founders looked at that as a very broken process.
• 6:13
  SPEAKER_02: And we decided before there was a smartphone, there was just a dumb pager.
• 6:17
  SPEAKER_02: But we decided if you could take the right attributes of a trouble ticket and take it with you on your pager and then update it from your pager.
• 6:26
  SPEAKER_02: Uh, this I would also say was before WANs, it was before there were data networks.
• 6:30
  SPEAKER_02: So this was a hard problem.
• 6:31
  SPEAKER_02: We were writing software to the firmware of a pager.
• 6:36
  SPEAKER_02: Wow.
• 6:36
  SPEAKER_02: Um, but we really decided that that was a way to improve life.
• 6:40
  SPEAKER_02: And uh that's how it all began.
• 6:42
  SPEAKER_02: So that company ended up uh getting acquired by uh one of the four, one of the at the time it was called the Big Four in Systems Management, a company called BMC Software to improve their systems management tools.
• 6:55
  SPEAKER_02: At the time it was BMC, CA, IBM, and HP.
• 6:59
  SPEAKER_02: And uh since then, the Big Four have turned over many many times.
• 7:02
  SPEAKER_02: But uh that was how the journey started.
• 7:05
  SPEAKER_02: And then through a series of interesting innovations, we got to the point a few years ago where we could really apply AI and machine learning at scale, and we could do it in a way that that would finally allow us to use the right compute, storage, algorithms to fully automate the life cycle of an employee service request.
• 7:25
  SPEAKER_02: And so, fast forward to today at People Rain, we've built uh the world's kind of definitive taxonomy of employee service, and we trained a neural net on about a billion historical IT incidents and HR cases, and so you can actually use NLP sitting on top of that neural net, sitting on top of that, what we call a domain ontology, and have a very fluid conversation in 27 languages with an employee, and almost every time you can autonomously diagnose and resolve their issue.
• 7:59
  SPEAKER_02: And ultimately, Chris, to your question, what that means is that we can give an employee back about an hour a week to do less of what they hate, which is waiting on hold and being treated like a number, and more of what they love, which is creating value, being creative, being empathetic, exercising rational thinking.
• 8:19
  SPEAKER_02: So it's ironic that you know, here we are 25 years later, and we've made the human experience less robotic by introducing AI and machine learning into the process.
• 8:31
  SPEAKER_02: And I think that uh that's why I was put on this earth is to help the next billion employees really feel like the best versions of humans using augmented intelligence.
• 8:43
  SPEAKER_02: That's awesome.
• 8:44
  SPEAKER_02: That's awesome.
• 8:46
  SPEAKER_04: I think I think you you nailed it where a lot of it is, you know, we're utilizing artificial intelligence to augment a human experience and take the mundane out so that it frees people up to be creative, which inherently humans are creative.
• 9:06
  SPEAKER_04: And unfortunately, we get locked in these in these, you know, prop mundane processes, and it kills creativity, therefore, kills the drive and motivation to really want to excel and do more.
• 9:18
  SPEAKER_04: And actually, I think it plays a huge role in the psyche of of a human to actually have all of these just mundane things that they have to go through every day.
• 9:28
  SPEAKER_04: And now we're bringing in these things to say, hey, listen, don't worry about that, you know, that stuff.
• 9:34
  SPEAKER_04: Why don't you focus on what you're good at and using that the most intelligent thing you have, which is that brain of yours, to actually be creative, right?
• 9:42
  SPEAKER_04: Which I love.
• 9:42
  SPEAKER_04: So that's awesome, man.
• 9:44
  SPEAKER_01: It's analogous to you know, all the fears that people had.
• 9:48
  SPEAKER_01: Of course, that does mean there's and you know, at uh the turn of the century, uh, the uh turn of the 20th century, you know, when when we introduced steam shovels, people were worried about ditch diggers being out of work, but people didn't like digging ditches and uh and what was possible with with this equipment of building skyscrapers and highways and track, you know, and cities and and all the other things, of course, with the rest of the mechanization industrial revolution.
• 10:16
  SPEAKER_01: But I think the challenge, of course, is you know, there's you know, the old uh uh the Luddites who who uh uh did not want to, who smashed uh the the looms because of the putting weavers out of out of work.
• 10:30
  SPEAKER_01: But you know, we know there's going to be disruption because a lot of the uh mundane tasks can form entry-level jobs or you know, jobs that are uh that require less uh expertise, maybe less creativity initially, uh, but they're still you know employing people.
• 10:47
  SPEAKER_01: So the question becomes you know, what it what will happen when you start to automate more of those pieces?
• 10:53
  SPEAKER_01: And it does come back to you know, what does the future work look like?
• 10:56
  SPEAKER_01: And if you extrapolate that to things like self-driving cars and trucks, you know, there's a lot of people who earn a living driving, you know, driving machinery, right, driving vehicles.
• 11:08
  SPEAKER_01: So what are your thoughts about uh that kind of displacement?
• 11:11
  SPEAKER_01: And you know, eventually we know there's a shift, we know there'll be new work, but there's going to be uh a displacement and disruption.
• 11:21
  SPEAKER_02: The World Economic Forum says uh AI will create a net new 58 million jobs uh in the next five years.
• 11:29
  SPEAKER_02: And the Pew Research Institute recently did a study of uh employees that said 89% of employees polled said they need to either reskill or upskill in order to be relevant in the next decade.
• 11:48
  SPEAKER_02: Um, I contend that any job that is one of the three Ds, either dull, dirty, or dangerous, is a really good candidate for autonomous intervention.
• 12:00
  SPEAKER_02: But the reason I'm so optimistic is because Seth, to your point, the jobs that are gonna the 58 million jobs, net new ones that are gonna be created, are gonna be in fields that uniquely require human traits.
• 12:13
  SPEAKER_02: So if I'm the human uh picking berries in a field, I would love to be able to invest in a skill that is less uh easily automated.
• 12:25
  SPEAKER_02: That's backbreaking labor.
• 12:27
  SPEAKER_02: It pays very poorly.
• 12:29
  SPEAKER_02: But when you think about some of the roles that AI and automation will create, they're jobs that require uh you know an understanding of data and an understanding of humans.
• 12:41
  SPEAKER_02: Think about you know, when we look at 75% of uh baby boomers will rely on digital assistance or voice assistance in the home to uh order their medications in the next three years.
• 12:57
  SPEAKER_02: So there's a a significant need for people who will educate that aging population about how to configure and use and and order their meds, and and that's a whole field that didn't exist, analogous to, you know, 25 or 30 years ago, uh, you know, there were a lot of print journalists that uh, you know, like the Luddites that you referenced, you know, in the 18th century, were very concerned about what the web might do to print journalism.
• 13:27
  SPEAKER_02: And it turns out, you know, there have been massive fields that require much more uh skill and pay much higher than print journalism did, being things like you know, anything related to web development or SEO, search engine optimization, search engine management, SEM.
• 13:46
  SPEAKER_02: And so what we've learned over the arc of history is um you never want to be on the wrong side of innovation.
• 13:54
  SPEAKER_02: And so I'm I'm more optimistic than ever about the future of humans and the future of jobs because of the way automation is rapidly replacing jobs that are dull, dirty, and dangerous with ones that will take advantage of skills that are uh innately human.
• 14:13
  SPEAKER_04: Well, part of that too is up up-leveling, like you said.
• 14:17
  SPEAKER_04: So I think people I think get maybe a little nervous on the reskilling or upskilling um scenarios, but what they don't take into consideration is the reskilling or upskilling is from a completely, it's not like they're going from ground zero to something different.
• 14:35
  SPEAKER_04: It's they're going from something that they've that they've already built on top of to something that's reskilled or upskilled.
• 14:42
  SPEAKER_04: So for instance, you know, I think about the digital immigrants, you know, in this world that are looking for, you know, like you said, um, ways of doing things that they haven't done anything in the past.
• 14:53
  SPEAKER_04: However, it's not changing their daily behavior.
• 14:56
  SPEAKER_04: So for instance, if I can use the little hockey puck on my desk to order this or talk to that or order this, it's a personified application that I don't necessarily know that I have I'm interacting with this really deep inference of you know short form you know commands.
• 15:15
  SPEAKER_04: However, I'm just speaking to it.
• 15:18
  SPEAKER_04: And let's say I'm I'm you know in my 70s, I'm just speaking to it and still enabling myself to do more than I would have done before.
• 15:27
  SPEAKER_04: But it's because I raised the level of upskilling and they didn't even know that.
• 15:32
  SPEAKER_04: And I think that's part of this too, is that there's a lot of folks that get nervous around the augmentation of AI.
• 15:41
  SPEAKER_04: However, what it's doing is it's already the tide that's rised all these modes.
• 15:46
  SPEAKER_04: And now they're starting from a different, you know, ground truth in actually doing these things as opposed to, you know, starting from their basis of like going back to, well, shoot, I've got to go back to a steam shovel or uh sorry, uh a uh ditch digging type of perspective.
• 16:00
  SPEAKER_04: And now I've got to learn how a steam shovel works, and then I've got to learn how you know a ditch which works, and I have to learn, and it's not that we've already gone past all of that, so now it's just a new starting point and we've moved the line.
• 16:12
  SPEAKER_04: Does that make sense?
• 16:14
  SPEAKER_01: I would, I would uh agree on some terms and I would I would push back in others because I think we still think that when you look at a big uh part of the population that is does not that does not have that baseline.
• 16:26
  SPEAKER_01: So a big pop part of population of people who you know maybe didn't go to college, you know, maybe weren't well educated, but they're doing certain jobs that they learned over the years.
• 16:37
  SPEAKER_01: Maybe they've, you know, generations of of uh of their family have worked in those industries or those jobs.
• 16:44
  SPEAKER_01: I think there's going to be some disruption there where maybe that upskilling skilling and retraining is going to be more challenging.
• 16:51
  SPEAKER_01: But I still think that the same tools and the same technologies that are going to cause the disruption are going to lead to the ability to even retrain people who need remedial retraining or remedial skills or you know, basic, basic skills, right?
• 17:07
  SPEAKER_01: Because again, if you look at our populations, there are underrepresented populations that you know um are having a hard time, you know, across demographics, across um the industry.
• 17:19
  SPEAKER_01: And and like I say, I think customized uh learning applications and the ability to uh understand, you know, where a person is and what's going to resonate with them, what's going to motivate them, what's going to get them excited, uh, you know, is going to have to be determined by algorithms that will take that assessment of who they are and be able to serve up what they need.
• 17:42
  SPEAKER_01: So do you have any thoughts on that, Dan, in terms of uh how the technology would play into that?
• 17:49
  SPEAKER_02: As a society, we need to get comfortable with a dialogue about what it means when uh automation gives every citizen back a day a week.
• 18:00
  SPEAKER_02: Yeah.
• 18:01
  SPEAKER_02: So we'll be just as productive in four days as we are in five.
• 18:05
  SPEAKER_02: And uh that's that's not a threat, that's a promise.
• 18:08
  SPEAKER_02: I mean, that that's within within the next decade.
• 18:10
  SPEAKER_02: So the the who gets the benefit?
• 18:13
  SPEAKER_01: Who gets the benefit?
• 18:14
  SPEAKER_01: Who gets the benefit?
• 18:15
  SPEAKER_02: So the the question for us as leaders and technologists is um how can we ensure that that extra call it 25 20 of our productive time uh goes to improve life, improve society, whether that's you know, pursuing a hobby or being you know a better parent, spouse, friend.
• 18:38
  SPEAKER_02: Um, I I contend that, you know, just given the state of things like you know, global poverty or you know, the climate crisis.
• 18:46
  SPEAKER_02: I mean, if we got 20% of the world's brain power and energy and time that we could feed back into social causes like that, there's no no doubt there's a net dividend that uh that that will pay to society.
• 19:02
  SPEAKER_02: But but now is the time to be having those conversations about what that means.
• 19:06
  SPEAKER_01: And and leadership and organizations need to be intentional about that and saying, well, rather than okay, that only means that means we can cut this percentage, we can cut 20% of the workforce and keep people working five days a week, but now we get 20% you know more to our bottom line, that is going to cause further, you know, um uh a further uh divide, right, in terms of who's getting the benefits and you know who's paying the price and who's you know getting the rewards, right?
• 19:36
  SPEAKER_01: So I think that's kind of a really uh important thing for people to think about.
• 19:40
  SPEAKER_01: It's like what are what is most important to us and and being able to benefit society, you know, is worthy uh in in terms of intent, but is our is our current uh uh set of more and and priorities and and uh value going to reflect that.
• 20:02
  SPEAKER_01: I mean that's that's the big challenge, right?
• 20:05
  SPEAKER_02: I contend that work will change more in the next 30 years than it has in the previous 300.
• 20:10
  SPEAKER_02: And now's the time to be thinking about what that new definition of work means for everyone.
• 20:18
  SPEAKER_02: So whether you're you know a teacher and you'll be augmented by learning technology, or you know, you're a police officer, and you know, you'll be able to identify, you know, the location where a gunshot came from and get to the scene of a crime faster.
• 20:35
  SPEAKER_02: You know, you can be, you know, uh an insurance adjuster, a claims adjuster, and you know, the whole process, the whole actuarial process will be automated.
• 20:45
  SPEAKER_02: So these are all examples of where those careers necessarily won't go away.
• 20:49
  SPEAKER_02: I guarantee we'll always need teachers.
• 20:51
  SPEAKER_02: Yes.
• 20:52
  SPEAKER_02: What it means to be a teacher and what tools you'll you'll need to do your job will change, and they'll probably make it easier to do your job more effectively.
• 20:59
  SPEAKER_02: But it's incumbent on us as leaders in in every profession to embrace the future, these innovative technologies, and figure out how to use it to make everyone in our field better rather than exercising concerns over what might happen, because the fact is we're a good, you know, I'd say 50 years conservatively away from artificial general intelligence, where you'll have machines that'll be capable of doing essentially any human task.
• 21:29
  SPEAKER_02: But right now we're on the cusp.
• 21:30
  SPEAKER_02: You know, some would say, you know, narrow intelligence is having its golden age now.
• 21:35
  SPEAKER_02: So, you know, let's not fear monger, let's embrace the capabilities and use it as an opportunity to rethink what it means to be the best version of a human.
• 21:44
  SPEAKER_04: I don't there's people that still consider the human in the loop concept as part of AI.
• 21:50
  SPEAKER_04: And yet it's super critical, right?
• 21:53
  SPEAKER_04: And it's not necessarily like, you know, like for instance, um, you know, my background, I you know, do language and speech.
• 21:59
  SPEAKER_04: And However, when you get to some certain um like things like solving problems for, let's say, close caption, right?
• 22:06
  SPEAKER_04: Hard one and or subtitling, the, you know, according to you know ADA and some of these other areas where you're actually helping, you know, a general group of a population, you have to have a hundred percent accuracy.
• 22:22
  SPEAKER_04: The only way to have a hundred percent accuracy is to have human in the loop still validate what the machines are doing.
• 22:29
  SPEAKER_04: So if you get to 80, 80, you know, five to 90%, that in in our speak is phenomenal.
• 22:36
  SPEAKER_04: But I'm sorry, it doesn't meet the meet the mustard for what these people need in order to actually live a life.
• 22:43
  SPEAKER_04: Right?
• 22:43
  SPEAKER_04: So you still have these notions where you have human in the loop scenarios where you need the validation from still, you know, a a uh uh we'll call it the meatware to validate what's going on with you know with some of these things.
• 22:58
  SPEAKER_04: So and I think that'll still continue, like you said, for I think forever, because as our enablement rises, our in our ability to um to take on more capacity rises, our knowledge rises, our ability to actually be more creative rises, all that kind of stuff.
• 23:18
  SPEAKER_04: It does rise the tide of all of our boats, regardless of where we're at, so that we're not starting from you know a you know completely different levels.
• 23:26
  SPEAKER_04: We're actually, you know, jumping in to say, okay, now I'm gonna focus on these problems, and that's gonna help this group.
• 23:34
  SPEAKER_04: And it's because I'm at a different, different area of knowledge that's gonna apply over here, right?
• 23:40
  SPEAKER_04: But it's still a human that's actually doing all that kind of stuff.
• 23:43
  SPEAKER_02: Those who uh choose to engage in fear-mongering about you know the bot apocalypse and things like that, I think um really do a disservice because you know, actually, the the example you gave of ASR, automated speech recognition, is a great one where uh most people don't realize that 85% accuracy is pretty good.
• 24:02
  SPEAKER_02: But when ASR gets 15 words out of every hundred wrong, it's an abominable experience.
• 24:09
  SPEAKER_02: Like you said, post-captioning or things like that.
• 24:11
  SPEAKER_02: So without whether it's human in the loop or you know, other kinds of you know, uh augmentation, the current state of AI is, you know, it's actually pretty poor.
• 24:22
  SPEAKER_02: We often compare it to you know the life cycle of a human.
• 24:26
  SPEAKER_02: And the best AI, you know, is call it maybe, you know, it's certainly not a toddler, you know, really an infant in terms of you know its ability to do what uh an adult or an adolescent could do.
• 24:39
  SPEAKER_02: And what one of the one of the other ways I think about this is by going back to you know the foundations of what AI is.
• 24:47
  SPEAKER_02: I think it's there's a broad misunderstanding of what AI is.
• 24:50
  SPEAKER_02: It's really uh you know applying statistics to data at scale.
• 24:55
  SPEAKER_02: Uh it's really it all comes down to being a data problem.
• 24:58
  SPEAKER_02: And what AI is really good at doing is predicting the next event by learning from historical data.
• 25:04
  SPEAKER_02: And so when you strip it down to its foundations, you know, it's a long way from you know, that that conceptual leap of faith where uh you know Skynet is going to appear with human-like capabilities and you know replicate itself and wreak harm on defenseless humans.
• 25:22
  SPEAKER_02: I think you know, we should more think about, you know, maybe using AI to you know improve the you know ability of prosthetic limbs, you know, to help amputees and you know, the ability to predict where a drought might occur, you know, to uh you know, prevent famine.
• 25:40
  SPEAKER_02: You know, those are things that are data problems that are really well suited to today's narrow AI.
• 25:47
  SPEAKER_02: But we're so far away, you know, so many, you know, significant scientific breakthroughs away from anything that would really resemble human intelligence.
• 25:58
  SPEAKER_01: We'll be right back.
• 26:00
  SPEAKER_01: We'd like to thank Marketing Artificial Intelligence Institute for their support of this podcast.
• 26:06
  SPEAKER_01: Marketing AI Institute is a media, event, and online education business that makes AI approachable, actionable, and accessible to marketing leaders around the world.
• 26:15
  SPEAKER_01: They run the world's leading marketing AI conference, MACOM, which brings together hundreds of marketing and business leaders to better understand, pilot, and adopt AI.
• 26:25
  SPEAKER_01: Today's podcast is brought to you by Simpler Media Group, owner of the Customer Experience Decoded Podcast, covering the latest in customer experience and marketing.
• 26:33
  SPEAKER_01: CX Decoded is a semi-monthly podcast designed to deliver insight from leaders driving the evolution of customer experience with an emphasis on digital touch points, customer journeys, respectful data management, and voice of the customer.
• 26:47
  SPEAKER_04: I want to give a big thanks to Early Information Science in support of the podcast.
• 26:52
  SPEAKER_04: Early Information Science is a professional service firm with proven methodologies designed specifically to address product data, content assets, customer data, and corporate knowledge bases.
• 27:01
  SPEAKER_04: Their information architecture expertise has allowed clients from all over the world to efficiently organize, access, and utilize their data, ultimately resulting in better business outcomes.
• 27:10
  SPEAKER_04: They deliver customer-driven solutions that scale and adapt to the needs of a business as it evolves.
• 27:16
  SPEAKER_04: Founded in 1994, Early Information Science has proudly delivered result-driven solutions for more than 100 brands across the top industries.
• 27:24
  SPEAKER_04: You can get more information about Early Information Science at early.com.
• 27:29
  SPEAKER_04: E-A-R-L-E-Y.com.
• 27:33
  SPEAKER_04: Working with customers in AI, the number one question that I get usually is where do I start?
• 27:39
  SPEAKER_04: What data do I need?
• 27:40
  SPEAKER_04: How do I gather it?
• 27:41
  SPEAKER_04: And it's great questions.
• 27:43
  SPEAKER_04: Well, if you want to catch the wave of AI, not be crushed by it, you really do need to read a book written by Seth Early called The AI Powered Enterprise.
• 27:51
  SPEAKER_04: It gives you the ability to look and understand how to harness the power of ontologies to make your business smarter, faster, and more profitable.
• 27:58
  SPEAKER_04: Seth goes into the intricate details of how to build an ontology, develop and employ an ontology as a secret weapon for the basis for how to actually utilize AI as a powerful competitive advantage.
• 28:11
  SPEAKER_04: This book is really the first one to combine sophisticated explanation of how AI works with a practical approach of applying AI to the problems of the business.
• 28:19
  SPEAKER_04: And it's applied to, you know, various use cases from customer experiences to business operations to product development.
• 28:26
  SPEAKER_04: It's a powerful, powerful tool in the arsenal of understanding how to deploy AI to make your business better.
• 28:32
  SPEAKER_04: The AI Powered Enterprise by Seth Early.
• 28:34
  SPEAKER_04: Harness the power of ontologies to make your business smarter, fashion, more profitable.
• 28:38
  SPEAKER_04: Out on Amazon.com today.
• 28:40
  SPEAKER_04: And now back to the show.
• 28:42
  SPEAKER_01: I just wanted to uh we're about halfway through a time.
• 28:45
  SPEAKER_01: I wanted to remind listeners that we're talking with uh Dan Turchin, uh CEO of People Rain and uh uh data strategist and uh ontology geek.
• 28:55
  SPEAKER_01: So let me uh let me reshift this to a little bit to talk about ontologies and talk about, you know, when you look at how organizations can adopt tools like those that your organization is providing, you know, help us understand a little bit more about, you know, what the prerequisites to success are.
• 29:13
  SPEAKER_01: And when you think about how you've trained your models to identify support requests and to be able to remediate, you know, there's a lot that's out of the box, but but does how far does an organization does that get an organization?
• 29:27
  SPEAKER_01: And then what else do they need to do?
• 29:29
  SPEAKER_01: Kind of that last mile of their own specific knowledge and content, data, processes, infrastructure, technologies, and so on, their real, you know, corporate DNA when it comes to knowledge and content and and architecture and of course ontologies.
• 29:49
  SPEAKER_01: Maybe maybe you could define ontologies from your perspective for the audience, and then we could uh answer some of those questions.
• 29:56
  SPEAKER_02: Yeah, Seth, I said all AI is a data problem.
• 29:59
  SPEAKER_02: And on my PAG podcast, AI and the Future of Work, we discussed the role that ontologies play in automating the delivery of IT and HR employee service.
• 30:10
  SPEAKER_02: So I'll go through that use case specifically.
• 30:13
  SPEAKER_02: So in order to use natural language processing to be able to understand an arbitrary intent expressed in natural language, having a structured vocabulary that defines how that service is delivered is essential.
• 30:30
  SPEAKER_02: For example, an employee asking about an airport may have a question about the place that you fly a plane from, or they might have a question about a Wi-Fi product delivered by a vendor called Apple.
• 30:44
  SPEAKER_02: And if the context of their request is around Wi-Fi or an outage or connectivity or something like that, it's less likely that they're referring to an airplane and you know the gate that they're taking off from.
• 30:59
  SPEAKER_02: So having an ontology that maps, we refer to it as a knowledge graph, but having an ontology or a knowledge graph that links these concepts makes it really easy to have, quote, AI or you know, an autonomous virtual agent understand in context how to go about resolving your issue.
• 31:18
  SPEAKER_02: If that structured vocabulary didn't exist, then you know it became essentially, you know, you're talking about like a Google search.
• 31:27
  SPEAKER_02: And you know, the best a Google search can do is map match your query to some keywords and return some documents that may or may not have anything to do with your actual request, but insert a domain ontology, a structured vocabulary behind that request, and all of a sudden you're conversing with an intelligent virtual agent.
• 31:47
  SPEAKER_02: Um, so I think that's, I mean, I know you shared some colorful examples, Seth, about your work.
• 31:52
  SPEAKER_02: But in my world, you know, that that that's how having a mature domain ontology and an information architecture is essential to delivering employee service.
• 32:01
  SPEAKER_01: And so, how far uh does your approach take an organization?
• 32:07
  SPEAKER_01: So uh my philosophy is organizations still need to get their their knowledge and data house in order.
• 32:12
  SPEAKER_01: So when you go into an organization, what are the preconditions for success?
• 32:17
  SPEAKER_01: And you know, this is not necessarily specifically about your product, but you know, put it in the context of your your offering to say, you know, when an organization wants to do these types of things, uh what do they need to be prepared to do or invest in?
• 32:30
  SPEAKER_01: Or what should they be working on today to make these things work better and be more efficient?
• 32:36
  SPEAKER_02: Yeah.
• 32:36
  SPEAKER_02: So interestingly, the deployment process at PeopleRAN involves three steps, two out of the three.
• 32:43
  SPEAKER_02: So literally two-thirds of the prep work to deploy an autonomous virtual agent has to do with data management.
• 32:50
  SPEAKER_02: It's a little bit ironic because uh, you know, we think about just the end result, you know, the the Siri that you talk to or Alexa or whatever it is.
• 32:59
  SPEAKER_02: But in fact, to get to the point in in your enterprise where you can have that level of conversational fluidity, it requires first data aggregation.
• 33:10
  SPEAKER_02: So there's first a significant exercise in figuring out where all the answers might live.
• 33:14
  SPEAKER_02: And again, Seth, I know this overlaps a lot with work that you do, and you gave some great examples on on uh on our discussion.
• 33:21
  SPEAKER_02: Um, but it's first aggregating the data and figuring out where it lives, and then the second is assessing the hygiene of that data.
• 33:29
  SPEAKER_02: So every enterprise has significant gaps in their data, the data is inconsistent and it's incomplete.
• 33:36
  SPEAKER_02: And so the first two steps involve what we call an AI data analysis and a configuration workshop.
• 33:42
  SPEAKER_02: And a hundred percent of those first two steps are just assessing the state of the data because that that then feeds the third step, which is actually configuring the virtual agent.
• 33:54
  SPEAKER_02: But by the time you've got uh your data hygiene at a point where it's usable, it's actually relatively easy to bolt it on to a user experience, but only only once you make sure that it's capable of learning from high-quality data.
• 34:08
  SPEAKER_02: Right.
• 34:08
  SPEAKER_04: There was uh there was an example.
• 34:11
  SPEAKER_04: We were working with a healthcare organization that was kind of a reseller of uh healthcare goods, right?
• 34:16
  SPEAKER_04: Across you name it, um clinical, all the way to just you know, physical health.
• 34:22
  SPEAKER_04: And um they they were trying to get people to understand the you know, like, you know, like the Peloton, you know, kind of uh phenomenon that's that's going on.
• 34:32
  SPEAKER_04: And they were like, we cannot figure out how to get, um, you know, every time somebody asks for a Peloton, it doesn't bring back the right responses.
• 34:42
  SPEAKER_04: And um, and so part of that was like, hey, listen, let's go in and look at your data points and see what is in there.
• 34:50
  SPEAKER_04: And of course, the ASR wasn't pulling back Peloton because it hadn't been been uh trained to know what that word is.
• 34:58
  SPEAKER_01: Didn't know what a Peloton was.
• 35:00
  SPEAKER_04: It was well, it was bringing back pallet on, right?
• 35:03
  SPEAKER_04: Right, right, right.
• 35:04
  SPEAKER_04: So, and so and of course it's gonna miss a hundred percent of the data points that you don't train on, right?
• 35:11
  SPEAKER_04: Not to go Wayne Gretzky on you, but that's exactly you know the the point of this.
• 35:17
  SPEAKER_04: And it's it was super fun because it was um it's always the little little things that actually make the most impact, right?
• 35:24
  SPEAKER_04: And the simple things, I should say.
• 35:26
  SPEAKER_04: And so as soon as we trained it, they were like, oh my gosh, we got 100, we got 100 on this accuracy.
• 35:31
  SPEAKER_04: I'm like, yeah, because you didn't have your data in a way that when you're looking for all these utterances and all of these specific, uh, whether it be issue detection or classifications or topic modeling or whatever, you didn't classify that word as one of the key phrases that you're looking for.
• 35:49
  SPEAKER_04: And they didn't sell Peloton, they were just looking for likenesses of people that were like, I want to bike like Peloton, you know?
• 35:56
  SPEAKER_04: And they were like, Well, how do you bike Peloton?
• 35:59
  SPEAKER_04: That doesn't make any sense.
• 36:01
  SPEAKER_04: So the the inference would come back as sorry, it's failed, right?
• 36:05
  SPEAKER_04: Things like that.
• 36:06
  SPEAKER_02: So best example of what you mentioned earlier about the need for humans in the loop, right?
• 36:09
  SPEAKER_02: A little bit of data labeling goes a long way when it comes to identifying and retrieving documents related to the bike called the Peloton.
• 36:17
  SPEAKER_02: Exactly, exactly.
• 36:19
  SPEAKER_04: Even if there's just the likeness, it doesn't even need to be that thing.
• 36:23
  SPEAKER_04: But you know as well as I do that the inference models will come up with like uh it's not a hundred percent accurate, you know, on actually coming up with that phrase, but it is 85% accurate on what these people are probably asking and the intent behind it.
• 36:38
  SPEAKER_04: So, yeah, that's probably a high likelihood that that's what they're asking for, right?
• 36:41
  SPEAKER_04: But even, yeah, like you said, just even that, just little simple things go a long way to solving that, you know, that core, you know, ask.
• 36:49
  SPEAKER_04: So that's funny.
• 36:51
  SPEAKER_04: But yeah, so the the the data's insanely critical.
• 36:54
  SPEAKER_04: And I think, like to your point, they miss it.
• 36:57
  SPEAKER_04: When you when you go in and talk, because obviously you're not talking to um, you know, you're not selling necessarily to, I would assume, to um a deep technical crowd, you're selling to the business, right?
• 37:09
  SPEAKER_04: Around you know the HR.
• 37:10
  SPEAKER_04: How do you um help them understand the need for that?
• 37:13
  SPEAKER_04: Because obviously, you know, there's got to be the value prop of what you guys provide, and that's the outcome.
• 37:19
  SPEAKER_04: But at the same time, you know, you go back and say, okay, well, listen, two-thirds of this is really gonna be about your data.
• 37:25
  SPEAKER_04: How do you get them over the hump about thinking about it correctly so that you know, because you probably walk into so many scenarios where, just like Seth, uh, your data is a goddamn mess.
• 37:36
  SPEAKER_04: And it's gonna take a significant amount of time to actually even find the data points we're looking for to be able to train this.
• 37:42
  SPEAKER_04: So, how do you help them understand that or get the business to think about that?
• 37:47
  SPEAKER_01: Or do you say, let us know when you have your act together?
• 37:53
  SPEAKER_02: I mean, Chris, you started the conversation focusing on the why and not the how.
• 38:00
  SPEAKER_02: And so, um, for example, just before we start recording this podcast, I was doing a workshop with uh a higher education organization, a university on the East Coast.
• 38:09
  SPEAKER_02: And the problem that they were describing is that when faculty members experience a technology outage in the classroom, like during a lecture, it impacts the learning experience.
• 38:24
  SPEAKER_02: And so the why is students who pay a lot of money to get educated by world-class professors don't expect an outage, you know, a network outage to impede their learning experience.
• 38:38
  SPEAKER_02: And so we 80% of the conversation was focused on why does that matter?
• 38:44
  SPEAKER_02: How often does it happen?
• 38:47
  SPEAKER_02: What is the impact to the the educational institution's mission when something like that happens?
• 38:55
  SPEAKER_02: It's less about you know, why'd the network go down or you know, how many geeks does it take in a server room to diagnose the issue?
• 39:03
  SPEAKER_02: It's really about what's the human experience and you know what what what happens when you know to the students and what what does the faculty member do?
• 39:11
  SPEAKER_02: And then once we fully diagnose the uh the the why it matters and we understand the impact on in this case on the learning experience, only then can we step back and think about what a technology solution might look like to automate, let's say, the monitoring of the network health or introduce you know some predictive analytics to look for, you know, use AI to look for the patterns in data flows that might indicate when the network's going to go down.
• 39:42
  SPEAKER_02: But 80%, literally 80% of the time is spent understanding the business problem, the human impact, because then and only then can you understand where's the data?
• 39:54
  SPEAKER_02: Where does it live?
• 39:55
  SPEAKER_02: What, you know, what what it's what's its current state of hygiene?
• 39:58
  SPEAKER_02: What does the AI model need to look like?
• 40:01
  SPEAKER_02: How do we use humans to label data to augment the accuracy of the solution?
• 40:06
  SPEAKER_02: But it's incumbent on us to first ask the right questions about the human experience so that we can propose the right technology solution.
• 40:14
  SPEAKER_04: Yeah, I think there's there's, you know, for the listeners out there, I think that was exactly what um what I was hoping you'd bring out is it's literally about the quantifiable impact, positive or negative, uh, of not doing this or of doing this.
• 40:29
  SPEAKER_04: And once you can come to that conclusion and understand, just you know, maybe it's a little bit of a hypothesis, but that's what the data is telling you, then you can work backwards and then nail it, right?
• 40:40
  SPEAKER_04: Because at least you have a North Star that you're working towards that you've worked backwards from to say this is exactly what we're gonna try to solve, whether it's a positive impact or negative impact.
• 40:49
  SPEAKER_04: This is what the data we need to make sure that we're hitting this.
• 40:53
  SPEAKER_04: And I think, you know, part of this too is in my experience, a lot of people just miss that and they just start throwing stuff at the wall to see what sticks.
• 41:00
  SPEAKER_04: They think that this is gonna be um, you know, the the silver bullet to solve all problems, and it not isn't because it takes an evolution and it continues to evolve and it never stops, right?
• 41:12
  SPEAKER_04: Never stops learning because you're gonna get if you get more data, add more data points, it's always gonna be learning, right?
• 41:18
  SPEAKER_04: But if you give it a finite amount of data, then you've limited your ability to progress.
• 41:23
  SPEAKER_04: So it's you know, it's that chicken and the egg of just making sure that you know you work towards what the what you believe the pause or negative impact will be and work backwards.
• 41:32
  SPEAKER_04: So that's awesome.
• 41:34
  SPEAKER_01: And and to that point, um I imagine when you're beginning with a with a customer, you're identifying those scenarios and those use cases and then using those to say where do we need to focus our attention around remediation?
• 41:48
  SPEAKER_01: Because I think too many organizations, you know, when they're presented with uh this uh large scale problem of well, we need to get our data house in order, our knowledge house in order.
• 42:00
  SPEAKER_01: We need to be able to document all these things.
• 42:02
  SPEAKER_01: It's it's too broad, it's too big, it's too difficult.
• 42:05
  SPEAKER_01: But it also misses the point that Chris was making in that we're really looking for a very specific intervention, right?
• 42:12
  SPEAKER_01: We're trying to understand a process.
• 42:14
  SPEAKER_01: We're not trying to replace a human, we're not trying to replace an entire process, we're trying to understand where the gaps in a process are and where we can have a remediation, where we can have an intervention, and then focusing completely on that, building out those use cases, and then being able to understand the data on that.
• 42:33
  SPEAKER_01: What does that usually take in terms of you know, a realistic assessment?
• 42:37
  SPEAKER_01: So how many use cases do you typically begin with when you define?
• 42:41
  SPEAKER_01: I mean, can we get down to kind of a specifics around what an engagement might look like?
• 42:45
  SPEAKER_01: And again, just for people to kind of get in their heads that hey, this is feasible and we can begin this process.
• 42:51
  SPEAKER_01: We can begin by analyzing you know that that customer journey, that employee journey, that uh life cycle, whatever it is that our target is, and then and then start to define those people.
• 43:02
  SPEAKER_01: So maybe you could talk a little bit more about the details of some of that.
• 43:06
  SPEAKER_02: Yeah, you may find this as well, but when you initially speak with a customer, um, they'll have a dozen different problems that are good candidates for uh AI AI machine learning.
• 43:19
  SPEAKER_02: And when you start whiteboarding uh, you know, every way that this kind of technology could benefit their business, it quickly, you know, it's a list that grows without bound.
• 43:28
  SPEAKER_02: We feel like it's it's our job as the experts to actually narrow down that list.
• 43:33
  SPEAKER_02: So we feel like it's uh addition by subtraction.
• 43:36
  SPEAKER_02: So we we we will never commit to a project unless we can uh commit that we'll deliver value within 30 days.
• 43:45
  SPEAKER_02: So what that means is picking not just the right business problem to start with, but also understanding uh the KPIs or the key performance indicators, so the success metrics that the customer will use to measure the success of the project.
• 43:59
  SPEAKER_02: So So at People Rain, we have four applications for three different personas, and they can all be deployed independently.
• 44:07
  SPEAKER_02: We have a virtual agent for the employee, we have two applications that assist the caseworker, the live agent, and then we have one application that does predictive analytics for the service owner.
• 44:19
  SPEAKER_02: And we do not come in with any kind of assumptions about which application to deploy for which persona in which order.
• 44:29
  SPEAKER_02: We listen.
• 44:30
  SPEAKER_02: And like in the case of that university I mentioned, the recommendation is really to start with predictive analytics.
• 44:37
  SPEAKER_02: So even though a lot of times the virtual agent is the shiny object, sometimes it's not the right tool.
• 44:43
  SPEAKER_02: If in this case the business problem is to proactively remediate network issues, it turns out using the virtual agent is not likely to improve that KPI with the same level of efficiency as a predictive analytics approach, because the student who's impacted or the educator who's impacted, their inclination or the best user experience isn't to pause the lecture and ask a virtual agent how to troubleshoot the latency in the network, right?
• 45:17
  SPEAKER_02: That's the last thing you want to have happen.
• 45:19
  SPEAKER_02: What you do want though is for someone in information systems or IT to get proactively notified, let's say, you know, four hours before there's any uh performance degradation in the network.
• 45:32
  SPEAKER_02: So that was a case of where it starts with a problem or a large set of problems.
• 45:36
  SPEAKER_02: We isolate one business problem that can be solved within a 30-day uh kind of path to value.
• 45:43
  SPEAKER_02: And then and only then do we propose a tech a technical solution and then a path to deploying that technical solution.
• 45:51
  SPEAKER_02: Um I think it's important to think about it as you know, first in terms of a business problem and a success match.
• 45:57
  SPEAKER_01: And then you're bounding the problem and you're bounding the domain and you're bounding the knowledge and you're really bounding the use cases, and you're saying, let's make this achievable and demonstrable in terms of what that return will be.
• 46:09
  SPEAKER_01: And then then over time, these things grow.
• 46:11
  SPEAKER_01: And I like to recommend that clients build libraries of use cases according to their different segments, and those become a value, right?
• 46:19
  SPEAKER_01: Those become increasing value.
• 46:20
  SPEAKER_01: You can use them to test all sorts of things, right?
• 46:23
  SPEAKER_01: Use cases are testable.
• 46:24
  SPEAKER_01: And so you can build evidence.
• 46:26
  SPEAKER_01: Can we can we support that use case?
• 46:29
  SPEAKER_01: Can we support that task?
• 46:31
  SPEAKER_01: And it actually goes into role definition and job definition and so on.
• 46:36
  SPEAKER_01: And the more we can support those, we can start building those capabilities that become more and more sophisticated and that can further augment uh any of those job roles.
• 46:47
  SPEAKER_01: And of course, then we start looking at commonality or cross roles and and then beginning to extend those.
• 46:53
  SPEAKER_01: So it sounds like it really becomes a discipline that the organization needs to build some maturity around and some capability around.
• 47:00
  SPEAKER_01: Is that would that be an accurate uh way of uh describing that?
• 47:05
  SPEAKER_02: Absolutely.
• 47:06
  SPEAKER_02: And further to that point, it's a learning process for everyone.
• 47:09
  SPEAKER_02: So give you an example of uh an insurance organization in uh Europe that has been using PeopleRAM for a long time, years.
• 47:17
  SPEAKER_02: And initially, um, their vision was just to um automate the IT trouble ticketing process.
• 47:26
  SPEAKER_02: And they started by assisting live agents.
• 47:29
  SPEAKER_02: So the live agent gets a screen pop when the phone rings with an employee on the line to uh decrease the research time.
• 47:38
  SPEAKER_02: The AI is basically showing them what to do next.
• 47:41
  SPEAKER_02: Um, and so that was just augmenting the intelligence of the live agent to reduce the the downtime for the employee um and just for IT issues.
• 47:51
  SPEAKER_02: Well, now fast forward a few years, and now there are 75 different departments within this insurance organization that span 25 countries, but they didn't really know how those other 74 uh uh use cases could be none of us, us or them, just you know, knew what the art of the possible was until we saw this first use case working.
• 48:13
  SPEAKER_02: And then all of a sudden, you know, they shared the success that they were having in IT, and facilities had a need for something similar, and finance had a need for something similar, and legal and sales operations, and all of a sudden, you know, this mushroom cloud of you know use cases emerged, it all really relied on the same technology, right?
• 48:34
  SPEAKER_02: But until we understood what the art of the possible was, it was hard to get beyond that first use case.
• 48:39
  SPEAKER_02: So, you know, we maintain that you know, time bounding it with 30 days and one use case and one success metric is almost always the way to catalyze organizational creativity to be able to have a roadmap that has any kind of uh you know integrity.
• 48:57
  SPEAKER_04: Well, I mean, I can't tell you how many organizations I've run into where they were like, we want one self-service assistant or bot to rule them all.
• 49:05
  SPEAKER_04: I'm like, good, that's great, because that's a pipe dream.
• 49:08
  SPEAKER_04: That'll never happen.
• 49:10
  SPEAKER_04: And you know, it and it's easy to think that way from you know, this is me speaking to them to from your perspective because all you're doing is seeing it from your perspective, and you're not seeing it from the global perspective, the departmental perspective, the organizational perspective, where you almost have to have a network of these things.
• 49:28
  SPEAKER_04: It can be built off of the same structure and it should be, but it's different data points coming in.
• 49:34
  SPEAKER_04: So that's great.
• 49:35
  SPEAKER_04: But guess what?
• 49:36
  SPEAKER_04: What happens when you span you know service for a customer and HR and you need to share context?
• 49:43
  SPEAKER_04: How do you do it?
• 49:44
  SPEAKER_04: Right.
• 49:44
  SPEAKER_04: And we're just now getting into the realm of being able to hold context over between these different, you know, scenarios and uh and then be able to, you know, like somewhat come to a conclusion of, oh, here's all the things that you talked about, Dan.
• 50:01
  SPEAKER_04: Here's all the groups that you wanted to get information about, here's the transcript, and you went to five different departments for that, you know what I mean?
• 50:08
  SPEAKER_04: And it wasn't, you know, we didn't, you know, it was just based on a network of all these different bots that were or assistants that were sharing this information as opposed to one to rule them all.
• 50:19
  SPEAKER_02: You know, it just doesn't work that way.
• 50:21
  SPEAKER_02: I can't tell you how many uh technology marchitecture diagrams I've seen that place some version of a cog in the middle, and then you know, a bunch of you know technology features kind of orbiting around the cog.
• 50:36
  SPEAKER_02: And I want to vomit every time I see one of those because we're missing an opportunity to put the human in the center of that technology ecosystem.
• 50:46
  SPEAKER_02: And that's not just a foundational flaw in marchitecture diagrams, it's a foundational flaw in how we think about the value of technology.
• 50:56
  SPEAKER_02: We should always think about the human experience first.
• 50:59
  SPEAKER_02: And if we do that, it becomes very easy to take the technology and map it to the business process and the human experience rather than what we've historically done.
• 51:09
  SPEAKER_02: We just take the technology and tell the user to map their process or how they do things to what the technology can do.
• 51:17
  SPEAKER_02: I think if it, you know, but that that's that's what I want on my epitaph is uh, you know, the right version of that architecture diagram.
• 51:25
  SPEAKER_02: Yeah, yeah, that's great.
• 51:27
  SPEAKER_01: That's awesome.
• 51:28
  SPEAKER_01: Well, we're getting to the end of our time for today, and it's been a real pleasure, Dan.
• 51:34
  SPEAKER_01: And uh tell us what's what's on uh your agenda these days.
• 51:37
  SPEAKER_01: So I know you're uh you're an athlete, you uh like to compete in triathlons.
• 51:42
  SPEAKER_01: Do you tell us a little bit more what did you already do your training this morning?
• 51:46
  SPEAKER_01: I guess I have to ask.
• 51:48
  SPEAKER_02: I certainly did.
• 51:50
  SPEAKER_01: What's on your agenda?
• 51:51
  SPEAKER_01: Yeah.
• 51:52
  SPEAKER_02: Uh so yes, I am a bit of an adrenaline junkie.
• 51:55
  SPEAKER_02: Um, but I think that you know, that that's the fusion of work and and uh and life that makes us the best version of humans that we can be.
• 52:04
  SPEAKER_02: It's just that desire to be our best and you know, compete at the highest level.
• 52:10
  SPEAKER_02: And I always, you know, I always kind of coach myself that you know, if I can make myself the best version of myself, then I don't have to worry about anyone else.
• 52:19
  SPEAKER_02: And like I said, I I I firmly believe I was put on this planet to help make a billion employees have a better experience at work.
• 52:26
  SPEAKER_02: And uh, you know, I'm I I take no greater sense of pride than being able to build teams and build products that change lives by applying the power of technology that can sometimes seem uh you know more like science fiction.
• 52:42
  SPEAKER_02: But you know, if we we think the right way about the human experience, we can almost always take the fiction out of the science and uh you know use it, you know, to make all of us better.
• 52:51
  SPEAKER_02: And so whether it's in my my training or you know, whatever else, you know, the way I parent my daughters or you know, always try to think about how to be the best at whatever I do, but always with the goal of you know trying to trying to put a little debt in the universe by applying some of what amazing technology can do to impacting lives.
• 53:11
  SPEAKER_01: And I'll tell you, it is, you know, if you look back 20 years, the stuff we have today is science fiction, right?
• 53:17
  SPEAKER_01: It was science fiction back then.
• 53:19
  SPEAKER_01: The greatest scientific minds uh in the world, you know, would say that this would have been a physical impossibility, right?
• 53:27
  SPEAKER_01: But um it's and and there are physicists out there who think that, and that's because of quantum mechanics and you know, all the work that Einstein did, and there are physicists that believe we will have more breakthroughs of an equally uh uh uh impactful and and paradigm-changing manner.
• 53:45
  SPEAKER_01: And uh it's funny, I was at uh uh a presentation the other day talking about single photon um light uh detection uh that enable cameras that can look around corners and see through documents and uh just it's I think what what technology can do for humankind is just uh it's it's it's unlimited and it's phenomenal, but we have to have that intent to to benefit people and to help people and to really think about how it's going to impact society.
• 54:16
  SPEAKER_01: So I really commend you you know on that value and that thinking.
• 54:21
  SPEAKER_04: So thanks for the passion, Dan.
• 54:23
  SPEAKER_04: It's palpable, it's awesome.
• 54:26
  SPEAKER_02: Thank you so much for right on the cusp of something that uh will transform how we as humans enjoy life.
• 54:34
  SPEAKER_02: And I and I think it's it's important.
• 54:36
  SPEAKER_02: We we as leaders have a responsibility to educate everyone out there about the potential for AI and machine learning to make life better.
• 54:45
  SPEAKER_02: And I think if there's you know one thing your listeners take away, be optimistic, be proud, be a leader, but think about ways that we can help other humans embrace the power of these technologies.
• 54:56
  SPEAKER_02: Yeah, it's a responsibility for sure.
• 54:58
  SPEAKER_01: Awesome.
• 54:59
  SPEAKER_01: Thank you again.
• 55:00
  SPEAKER_01: It's been a pleasure and uh thank you for the inspiring.
• 55:04
  SPEAKER_01: Great conversation.
• 55:04
  SPEAKER_01: Thanks for having me, guys.
• 55:10
  SPEAKER_00: Thanks for listening to the Early AI podcast with your host early and Chris Father.
• 55:16
  SPEAKER_00: We hope you enjoyed the program and took away some ideas that you can start implementing data.
• 55:25
  SPEAKER_00: Early data, information about the information, and then we're gonna go.