hello and welcome to premium bites where we shine a light on The Unsung world of insurance technology leaders I am sudar Krish machari one of your co-hosts and with Samir Ahmed our other co-host we're here to have meaningful conversations about the evolving technology landscape inside the insurance industry let's acknowledge it being an insurance firm CIO or Chief Information officer is one of the toughest leadership roles in today's world you've got a RIS discovers mindset baked into to the ver
y fabric of the industry a legacy estate that must complement the modern tech and a talent crisis from both the retiring side and nextg recruiting side and then you got the slower piece of innovation we see it because we're right there in the trenches with us Consultants day in and day out our mission is to Deep dive have conversations that carry substance offer practical insights and help you validate and navigate the relevant challenges of being an effective CIO in the insurance sector so than
k you for joining us on this adventure let's jump right into our next conversation welcome guys um I hope everyone is doing well um as um we discussed last time I think uh some of you who have been catching up with with our podcast I think some of you might know that we talked about uh it and analytics and uh the the organizational constructs behind that kind of touched on uh uh some components of AI there but before that we got some great feedback from one of our own colleagues who attended a G
artner conference about um how that topic was relevant and probably will resonate with a lot of folks given the priorities uh but uh Samir welcome back I know you and I are also going on the conference circuit but I'll let you uh chime in and I know you're on the east side I'm going west side we I'll let you say a word or two about that yeah yeah so uh uh the spring conference season is in full circuit I already have uh one event under my belt from February I mentioned it last time but in Earnes
t the spring season is just now we we don't we don't come Canada in our accounts no just kidding yeah yeah yeah so um yeah that that'll continue and Go on into June pretty much but hey uh speaking of uh you know analytics uh so today uh we've got a guest with us right that is right it's um I'm really excited this is the first time in this format we're bringing a a guest and we're super excited this is something that you and I have been talking about um so yeah I'll let you uh introduce uh this s
pecial highly talented very uh special person for our first guest honor yeah yeah so our guest today he's uh the director of AI and ml at xby 2 uh he with over 25 years of experience uh in applying state-of-the-art research to complex and critical problems uh in a variety of domains uh for the past 15 years he's been leveraging Ai and deep learning methods um recently he's been applying these techniques to problems that are facing insurers uh he's created and developed award-winning uh AI soluti
ons to revolutionize engagement equity and impact in a variety of settings he also holds several patents in the fields of anomaly detection and classification so let's please welcome Mitch Quinn hey hey Mitch awesome hey Mitch hey hey guys thanks for the introduction Samir and sudar it's good to join you here so excited to be part of the podcast so yeah great to have you right that's right great to have you but and and Mitch I'm for the first time I'm hearing words like uh paytons and stuff like
that in your profile although we we've been working together didn't realize you have Pon on your name wow impressed uh and and uh 15 years 25 years something along those lines working on AI and you know I'm just going to ask a very naive question I thought the the whole AI was uh just uh you know fell from the heaven only 16 months ago so uh tell us a little bit about your journey yeah so the 25 years in applied research really starts with um the work that I did in uh image compression and and
video compression technology using wavelet based uh uh wavelet Theory signal processing and so forth but the specifically transitioning to AI you know AI has been around for a really long time um there there have been several AI winners where it fell in and out of favor um [Music] and regards to the 15 years ago we as part of a start up and we built solutions for info security um where we were detecting anomalies uh in the behavior of computers across an entire Enterprise Network and that involv
ed building a model that described and understood how the different configurations in a whole corporate network setting behaved and changed over time and to identify anomalies uh things that weren't normal uh in regards to the rest of the population and then to classify those anomalies in terms of how threatening they were right and so that was a very early form of what we would call unsupervised learning where you're characterizing um the behavior of a large thing to try to discover patterns or
unusual things and I involved a large amount of data so that that was an ad hoc version of of unsupervised learning algorithms um really have been working uh so you're one of those uh experts who really have been working for as long as uh you know computer-based uh AI modeling has been around yeah and well I mean so it was you know sort of by happen stance it was as far as the need to solve the problem that we were trying to build in in the startup that we were working in um the the other side
of it is that it eventually evolved as the techniques and the and the became more well-developed and became you know the Frameworks to develop them became more um Advanced we then got into using what we would now consider to be formal like deep learning to identify un you know threatening or things that we should block execution of in the network and then transitioning into more more um traditional techniques that are that that are now part of the language models that we see in practice today so
yes it's it's been a long it's a long way and it's certainly a a um interesting time to to to see come about as as having been done this for for quite some time yeah Mitch so so that's fascinating uh that that you've been working with deep learning for uh this long of a time uh our our audience obviously is uh you know Tech folks uh in insurance insurance so uh most folks will be familiar with what you're describing but perhaps uh just for anybody that's been under a rock for a little bit uh ma
ybe you can clarify you know how all of these things they fit together the space of AI where deep learning fits where generative AI fits uh and what may be of value to insurers sure yeah so there are the the two main so there are three main disciplines in traditional AI or machine learning there's reinforcement learning there's unsupervised learning and there supervised learning reinforcement learning is is and and unsupervised learning are really sort of nent in their in their applications and
and sort of Novel and how and the value that they can bring the the core value right now that is making big uh impacts to what companies can do with AI lies in an area called supervised learning where you know what the answer is and you can look at a large amount of data and you've labeled them and you're training a computer to determine the patterns in the input data to classify them in certain ways right so you either are predicting you know is this a is what is in this image or things where y
ou can start with known examples um and then new to the that is making the biggest impact in the news or has the highest you know hype associated with it of course is generative AI where AI systems are trained to predict the next word and then extended to generate um to generate text or or or or images or so on so forth and while they generative AI is providing some of the most interesting and most newsworthy uh results today um the area that is still providing the most value to the broadest ran
ge of domains is in uh supervised learning because we understand how those systems perform and can Target them at very specific um business problems oh that's great Mitch I most of our audiences are you know senior technology leaders on the insurance space and and many of them have already started their Journey on on data and analytics and to a degree in a lot of this from a machine learning perspective um from your experience um do do you do you see a an evolution of maturity in the insurance a
nd maybe even health insurance Healthcare space that we all dabble with is in terms of the data analytics and the supervised learning and uh the machine learning uh capabilities what's your take on that Evolution and where we are today as far as adoption um curve right so there's um it it definitely being in healthcare and insurance where the stakes are high and they have some fairly well established disciplines and it's very conservative and it's naturally in its approach um there's still a lot
to be explored and how those techniques can be brought to bear uh to to make a difference in in those domains um so it's I think there's it's one of those fortunate situations where the um where the field has progressed significantly and matured significantly and you know we understand pretty well how to implement and deliver those Solutions it's just a matter of this is a a well-formed technology is finding you know the solution that can be married with those Technologies to deliver like outsi
zed value um so it's it's an exciting time and I think one that uh that holds a lot of potential you know I think the danger is that we not get distracted in in some of the some of the consumer level uh types of applications um you know while those generative AI approaches or those those Solutions have the ability to provide real value um at the edges so to speak where you're sort of improving the quality of work for a lot that a lot of people are able to produce um and many times I don't believ
e it's going to provide a true meaningful competitive advantage to areas like Insurance um where the the real danger is if you over invest in in those applications then you may to um deliver a huge value which is finding exploring that opportunity landscape of applying this really mature uh and well understood techniques to your core business uh issues um and so I think exploring that is one of the critical things that may be overlooked today um but I think we need to you know I think a great Po
int focus on think it's probably one of those things that uh I mean we we we in talking to to folks out there um most people are looking at either uh these consumer level tools or we mentioned conferences at the top of the show here uh there'll be you know companies out there with AI products specifically for insurance uh so I think these are the two categories uh there's also naturally the you know rle your own kind of build your own AI model so perhaps you can give some perspective Mitch on uh
you know how an insurer should think about each of these three categories general purpose uh tools Insurance specific products and then when you're building your own all right sure so you know the way I like to think of it is like how um H how finely tuned do you need the solution be to be to have a have a big impact right if it's generating emails the the whole industry and everybody who writes emails can benefit from that so that's not necessarily something that that you know it's a tool righ
t it's just now it's just part it's like your editing tools um and so while those can help accelerate SP very uh generic and and not very specific tasks they are useful and and you should consider how to deploy them um safely and effectively right when it comes to buying solutions that are already built right you competitors are also able to buy and build those Solutions as well if they're generic enough um to your domain right so but it's a little bit more specific and you know I think there is
a risk of of overlooking things that can provide a those are sort of table Stakes right for for the domain is to be able to use some of the well established um tools that are widely available in for AI for and applied to insurance and then it comes to like the specific example specific ones that are custom to you and and and your um your Niche or your business you know the the the market that you have carved out um and there is where it becomes really tricky right because uh you have to think a
bout things such as you know can I attract the talent to build maintain support and and deliver these and many times these projects fail and so the the the main question comes well how do I approach building these Solutions and how do I how how do I make that a a competitive Advantage for me the the solutions I build that are customed to me uh are my business and how can uh how can we you know explore that landscape of what has the potential to make a meaningful change and do you think uh insura
nce companies should invest to build only on that third category or I completely get it the first category you know no insurance companies should probably think about building that because those are on the edges very generic common but the middle I'm curious if if I think the middle you you have to evaluate right you have to have a reasonable you have to understand stand and be able to evaluate there are a lot of vendors out there that make a lot of promises um and and so you have to be able to
evaluate those vendors well to see if they can bring value if the if the cost is worth the value that they provide um and I think that's the really that is a critical aspect of of you know deciding whether or not to to build versus buy and the question is can it be customized or does it provide you with a unique competitive advant AG or is it just merely you know or is it something that you have to do in order to stay on par right are you not trying to be competitive but reaching parity with wit
h other people in the in the domain who you might be competing with in your market and then the real building your own right I think yeah yeah yeah yeah so I I think that's all uh making a lot of sense uh so let's say you know I I I I I'm so on on your advice and I'm gearing up to to do this um in the last episode Suther and I talked a little bit about the evolution of analytics sort of moving around we saw it being a function of of it then we've seen the emergence of a chief data officer we've
seen emergence of a chief analytics officer uh now with the you know AI based Solutions how how should uh if I'm an insurer right I think about uh is this something that is a new toy for uh it folks to sort of start playing around with and then convince their business partners hey look what we can do and maybe you should think about using it or maybe a different approach you might suggest so perhaps some comments on that would help yeah you know in my mind I think it depends on um how integral t
hese AI Solutions are going to be sort of infused into the you know overall operations of the company if it is just a tool then you clearly don't need to have like a lot of you know it doesn't need to be an integral part of your company it's just like well we need to adopt this tool or we don't and that's that um the more integral it becomes then clearly the more strategic you know like fundamentally strategic it becomes as as the direction of the company and then that suggests the need for a hi
gher leadership or a higher leadership position that has the you know experience to avoid some of the pitfalls and everything and can provide um good direction and whether or not that's a full-time role or not is is sort of um Up For Debate or up for you know discussion sometimes that may lie with this CIO or the CTO if they have the competency and experience many of them have experience in the past um so they can act as fractional part of their of their role being focused on AI for the the for
the full company but in other cases where you're trying to introduce AI it makes it may make sense to have something like a fractional Chief AI officer um which is something that we're seeing a lot more um going forward where they uh you know in in some capacity advise the CEO and the rest of the executive leadership on how to uh solve their business who understand the problems because you're always starting with the problems first uh and and then you're you're seeing where AI might provide an o
utsized advantage or a competitive advantage in the solution to those problems and I think that that requires an intimate relation ship with the the executives and the problems that the company as a whole are facing to find how those things match up just like every other big technology now we now back to this business problems right yeah yeah yeah so so this uh this uh fractional AI officer or even full-time AI officer that you are speaking of maybe uh you can highlight you know what character i
stics this individual should typically have and you know when is it something that an insurer can say hey John over there you know repurpose what you've been doing now this is your new responsibility versus no we actually don't have this within our four walls so we need to get it from the outside so how how should somebody think about this yeah you know I think you should always start with the problems first right I I this this is AI is really like a fundamental uh capability that needs to be in
fused in the DNA because it's a new way to approach problems it's not the right solution for every problem it's not you know the right solution for everything um and so I think it helps to have that education at that level or to know you know general what strategically makes sense to to approach or you know like we may get some lift in this area versus that one with with with AI um you know I would I would expect them to have some experience in delivering AI Solutions and understanding the compl
exity or in integrating them um because they are kind of a different Beast uh in in in a lot of ways uh you know it's kind of this moving it's it's a it's an evolving discipline it's not the Cobalt code that lives in a on a server in a closet and runs quietly and and effectively for for 50 years until somebody forgets and unplugs it from the from the outlet um it these are thing these are like living uh it's like a living manifestation of of what's going on um and so like it it does require some
expertise in knowing how to keep them um performing well and to also maintain and deliver those Solutions effectively so I have a kind of a provocative question so we've seen a lot of Chief data officers Chief analytics officers do you think it's just a natural progression for those guys to be like oh you you're now a chief AI officer or yeah I guess that's a good question it I think delivering ml or AI Solutions is sort of a prerequisite for that role it's very different than delivering an ana
lytic product like dashboards or or things because uh it's it's it requires a a significant increment of of like Operational Support I if you think about how hard software is to write um software is really hard to write and maintain AI is another you know building AI Sol is another one on top of that where you're manag managing the data to write the patterns to build the software so it's sort of like your one step removed and the complexity is is much is is significantly higher and there's some
unique concerns there so I think it does require that expertise or or at least familiarity with delivering these specific um AI Solutions yeah sounds good Mitch well look uh I know we're uh on limited time so appreciate you're joining maybe we're quick sort of running back uh what you're sharing right so uh don't follow the hype yes take advantage of uh things like co-pilot and chat GPT but this is just going to raise uh the floor for everybody don't expect any kind of competitive Advantage out
of this uh Point Solutions in the market everybody has access to them so let's say it's a verisk fraud detection for claim submission uh just don't get left behind but again don't expect anything significant in terms of Competitive Edge so that Competitive Edge is going to come from building AI models and solutions uh within uh the company based on unique specific problems and to do that you have to build the capability and the team and that you mention is a little bit of a depends on the contex
t and situation you may be able to start with what you have you may be able to start fractionally with somebody from the outside but over time you see this growing and uh clarifying how it goes so that that about uh right yeah I think you summed it up amazingly well Samir I appreciate the uh the opportunity to share with you guys my perspectives I appreciate it yeah all right and a lot of our audience I think will be curious to perhaps uh maybe in a future uh conversation we'll double click on u
h you know what does it mean to run all of this over underwriting or or run all of this over a claims or you know there are a bunch of uh potential areas within the insurance domain that are prime for work on data and analytics and machine learning and models which they have already been doing but I think you know are the newer capabilities adding you know a a 10x value over this or is it just a progressive value ad over that right yeah like as a as a as a final like thought to leave you with is
like you know this not exploring the opportunity space within your environment effectively enough of finding how AI can provide this outside advantage that really is the risk that you're not exploring that that opportunity space effectively enough and that perhaps your competitors are and if you know if they are doing this more effectively than you then there will come a time when they have found a singularity moment where they are be they will have this outsized competitive advantage and you m
ight not might not be able to react to it in time yeah yeah sounds good Mitch so look uh if any of you out there are in the Life Insurance space or maybe in like uh work comp or something like that where you're dealing with medical records I think Mitch has been doing some a lot of great work on uh how to leverage deep learning for uh for for that specific uh sort of uh problem space or if any of you are in commercial Auto and you're scratching your heads on you know why your books are upside do
wn last two years uh we have some interesting things uh that we can talk about as well so Mitch if you don't mind I'll put your contact info in our show notes and maybe folks can reach out to you and perhaps have a conversation as well great sure thanks Samir for having me well looking forward to it I'm sure this will always be an uh Evergreen topic so expect to have you back again shortly great I appreciate it all righty so so I think that's a wrap uh thank you all for for joining and listening
in and don't forget to like comment subscribe and we'll see you in the next and while Samir is off to the east coast I'm off to the West CO but uh by the time you all listen to this we probably would have already uh spoken a little bit about that on our LinkedIn channels so looking forward to catching up again and that's a wrap for today's conversation on premium bite where we shine a light on The Unsung world of insurance technology leaders thank you for joining us on this adventure and being
part of our community don't forget to like comment subscribe share and follow us until next time take care
Comments