Sports AI symposium | Panel Discussion: The Future of Sports Technology and Fan Engagement

Sports AI symposium | Panel Discussion: The Future of Sports Technology and Fan Engagement

Show Video

our next panel is being moderated by Scott  Rosner Professor Scott Rosner he's a professor   of professional practice in the sports management  program uh at Columbia of SP SPS uh I don't think   there are too many people at Colombia outside  of the business school or maybe even inside   the business school who know more about the  business of sport more than Scott so Scott Scot I   appreciate that um and so I'll welcome the uh the  the panelists up to the stage um and we'll kind   of go one by one we we were going to prepare and I  was thinking about walk up music uh for everybody   uh but decided ultimately not to uh and so as we  I'll just bring everyone up so uh am Sharma Brett   of aradel Christina Chase um and FL Ortiz uh if  you could please uh come on up and we we'll get started all rightone all right so we are in effect the  anchor leg uh today and so want to do a very   quick introduction uh of everybody see their  bios really quickly out there uh am Sharma   is the CTO of dream Sports um and that and his  bio was just drop the mic moment you go a little   bit deeper uh and uh also previous experience at  Netflix Yahoo um NBA uh from UMass Dartmouth or   Ambassador I'm s computer science from um dmouth  sorry uh additional degree from from Santa claraa   uh undergraduate from University of Mumbai uh  and so welcome uh brettel um is uh executive   director Dr BR bar it's executive director of UNLV  International gaming Institute associate professor   uh in unlv's uh incredibly well regarded H College  of hospitality uh and an affiliate position at the   University of Sydney science brain and mind Center  in the school of site um does a lot of work around   academically around video games um Esports and  gambling sports betting uh operations technology   responsible gambling uh and community relations  by the way I have friends who can use your help   in probably all of those areas uh um so uh and  and sits on numerous boards um and has done some   really phenomenal work uh undergraduate from Brown  University and Ms and PhD from unlb uh Christina   Chase is managing director and co-founder of  the MIT Sports lab uh where she works with   professional teams Global Brands uh Elite Sports  organizations to tackle key questions at the   intersection of data Sport and Engineering uh has  appointments uh in the the School of Engineering   at MIT where she teaches Sports technology and  Innovation and Entrepreneurship uh and was the   first e entrepreneur and residence uh at MIT where  she's helped hundreds of Founders go from concept   to company uh Tech Stars ventor uh South by  Southwest uh accelerator judge um and an Advisory   Board uh former cyclist um has trained at the US  Olympic Training Center in the springs uh color   springs uh taught dville skiing generally a badass  U it's probably the best way to to explain what   what Christine actually does uh Christine actually  does for for a living um and then Ortiz uh is an   assistant professor of electrical and computer  engineering at Rucker University um but I think   you know what what I think the general population  would think that what your uh side hustle is in   a way uh is much for uh is the AI team lead for  baseball operations for uh the New York Yankees   expertise so right like oh okay and so expertise  in lies in sensing systems interaction and M   multimodal learning uh within heterogeneous  densely sensed environments which are a lot   of really big words for someone who teaches  sports management right um so um but uh and   has a phenomenal background um I look forward to  the conversation of all but Dr or earned his Ms   and PhD uh in computer science from Berkeley uh  and uh his BS in computer science and engineering   from MIT um so uh welcome to everyone and uh uh  like I said my name and like I was introduced   so Nic my name is Scott Rouser I lead the sports  management program uh which is over in the School   of Professional studies uh Ms degree granting  program um and uh I've been here for six years   prior to this I was at the warten school for 15  plus years uh help leave the research and create   and leave the research center there um as well so  uh excited for this conversation I'm admittedly uh   a AI neoy um you know and so in a lot of ways um  you know it it represents what the sports industry   as a whole and people on the practitioner side of  the industry um are involved in so it's actually   really helpful to have these conversations um as  we kind of try to bridge the gaps and and and as   AI becomes more infused in the day-to-day  so with all of that said that Preamble uh I   mean I'll start with you so let's talk about dream  sports right and how is dream Sports leveraging AI   to enhance user experience and engagement in the  fantasy sports sure I'll take some time actually   before I talk about anything else I wanted to  thank everyone for being here and like being kind   of the host dream Sports we this name throughout  the day uh I'll take the next few minutes just to   explain who we are what we trying to actually  do here and um but very proud to be associated   with colia Columbia's um you know search abilities  access to pity year for us our industry experience   and the kind of investment that actually want  to make in the sports Ty ecosystem I think we   can really solve a lot of fan problems there today  right in sports ecosystem and I think earlier in   one of the panels today I heard someone say that  the amount of research in acemia in the sports is   not enough right maybe some of the larger teams  of the world got lots of money are like doing a   much better job but that's where we without our  te Sports foundation and among other things and   these collaborations we want to do a lot more  um so who who are we right so dream Sports our   vision our vision is like make Sports better  simple so we identify um in the sports domain   what are the problems that we can solve for fans  and we actually go ahead and our Flagship product   is J which is a fantasy sports platform it's  alone on its own it has 220 million users today   and around 20 to 30 million of them actually play  in the largest like IPL which is the largest CET   tournament um in India we also have fan Cod which  is a content and merchandise platform where users   can watch tier two tier three content which  is probably not as popular for the masses and   then you have merchandise where we can it's not  in India it's um some of the things that I talk   about today have been Nuance because we serve the  Indian market primity so sports merchandise is a   lot more mature here for example in the States  but back in India that's not the case you look   get an authentic Sports Jersey of a player that  you want is not that easy so fan Cod provides a   sports um you know merchandise platform that you  can go and buy the jerseys customize it to yours Etc then we also have an In-House Gaming studio  called Dre game studio so that's what we're   saying is so even today cricket being one of the  very popular games there's no one AAA quality um   game mobile game for Cricket anywhere there are  some of them are the quality if you compare it to   the of of the it's not there right so we want to  create something like that um what about sports   travel a lot of people want to go around the globe  people follow Sports globally these days how do we   actually provide them the opportunity to actually  identify the games that they want to go give them   behind the scenes experience so dream set go is a  travel arm that actually does that these are the   various examples and then the foundation that I  spoke about so that's who we are um when it comes   to AI I think our journey actually has been like  started a few years ago when we historically if I   look at it in two buckets historically we've  invested a lot in machine learning um so for   example some use cases would be demand forecasting  so an interesting thing about fantasy sports   platform is a lot of users come to your system in  a very short uh interval of time so you know for   example our Peak concurrency is 15 million users  are on the plat at the same time and they're not   there just to consume content they're there to  actually make you know technical transactions so   they're there to create their teams or edit their  teams they're there to add money to their wallet   they're there to join different contests that  exist on a platform a contest joint right on a   platform can go as high as 30,000 contest joints  per say so you know so we need to be prepared to   handle such volumes so we need to prale our  system so what is going to we have to predict   the user Behavior and the expected users on the  platform and have all our system because in real   time I don't have enough time because of the spiky  nature of the platform I don't have time to scale   up my cloudform so I have to pre pre scale my  platforms so that's what our demand forecasting   actually comes in um it's a again app so then we  have to make sure that there's no the Integrity   of the platform is maintained so there's a lot  of work that we do to make sure that there are   machine learning algorithms that identify any  kind of collusion between Pairs any violation   or terms of service so heav invest in that um  personalization obviously is key I all good apps   of the world I think end to end I talk more about  maybe lat than end user journe personalization   matches the contest that they join the rewards for  them the payments instruments that they use so we   personalize those and uh so that's historically  and all these and what makes it a lot more complex   is the scale at which we operate as 15 million  confident users all of these use Cas will become   a lot more complex um we've also been investing  heavily in caal infuencing over the last one   year uh you coming from like the companies in the  industry I can tell you that making decisions is   very is very difficult I mean you may think that  today we have so much data analytics capabilities   so much data about our users but even today we  sit in the Board Room or sit in conference room   and still struggle to make decisions um so that's  where like things like a comes into the play so   it's like directionally we relying on a lot the  understanding the relationship between various   metrics and variables to try to figure out what  to prioritize because just like everyone else   we also believe we have very limited people  I think mostos will tell you they don't have   enough engineers and also like them so I have  to prioritize what to do so that's where these   decision making capabilities come into picture  forward looking we very excited about J gen so   we've done some prototypes two tracks there one  is the some prototypes to solve user problems they   help them create their teams using uh you know by  looking at all the information that's available   in the in the public sphere and then bring it all  together for a particular user and the pro side   also we trying to do a lot of things hack inside  not just focus on technology but all functions   okay marketing HR to bring so I'm try to bring  everyone up to speed with what's the capability   what's the possibilities and you know how make  them independent so they're actually not relying   on the teams and Engineers but they're on their  own using these platforms so yeah a lot of work   happen yeah that's AES and if you had followed if  you play fantasy football for the NFL uh you'll   notice as you use the ESPN platform that they've  started to incorporate something similar uh with   Watson AI stuff and uh particularly uh Watson  X stuff and uh really interesting right I mean   if you comments on you know everything from is  this the correct trade to make uh and feedback   that you'll receive on any trades they make or  proposed trades um how good or bad your draft   was uh in a particular uh in any particular  year uh it'll make game predictions obviously   for uh for each of the athletes individual  performance so it's basically um the the dream 11 appone yeah it's fascinating by the way they think  very highly of Scott's top so far in the 2024 uh   fantasy football season just so you know um so uh  BR um so you do have done some some fascinating   work and really cool research how do you see AI  in data analytics shaping the future of sports   betting and gaming uh particularly in terms of  responsible gambling practices just like we' heard just like we've heard all day long uh no  okay it has changed every single f asset of sports   betting Eng gaming And to clarify because I'm use  a whole bunch of different terms and I think the   different people on this panel will use a whole  bunch of different terms um the very generic   definition of gambling is risking something of  value on an outcome that's uncertain we talk about   legal gambling the legal definition of gambling  that things that you're risking in the prize you   might win at the end that's something of uh money  or money's worth so when we talk about gaming   that might something that theoretically looks  like gambling but when we think about it under   the law it might be a better fit underneath the  definition of gaming so I want to make sure when   I talk about betting and gaming we're clear all  of this is in essence Gaines of Uncertain outcomes   it's just a matter of what the use of that game  what goes into that game what are you risking   and what might you get out of it those are some of  those other elements as well I'll also talk about   skill I think a little bit later because that  I think is a very interesting future use case   of some of this I but collectively within this the  area of betting and gaming again just about every   single facet of this has some sort of impact  from data science and and in particular AI so   I'm only going to talk about a few briefly here  we can delve into others as the subjects come   up but I'm going to start with of course the  subject that's on the screen behind me which   is fan engagement one of the most important  things to recognize about betting and gaming   is just how much it has an impact as a whole on  fan engagement one of the best examples of this   is to go all the way back now I used to say the  other year now it's nine years ago to 2015 when   we first started seeing the true explosion of  daily fantasy sports we had had especially in   North America fantasy sports for a very long  time and we saw just how much it really got   people engaged with new teams with new leagues and  how that impact really helped Foster new interest   in these games in these teams in these Leagues  with daily fantasy sport we saw at an even more   granular level level because as you start forming  your teams and you pull players from various   different parts of the league onto your particular  team for that particular game people stay longer   you're engaged through more of the game I brought  this up with some ESPN folks the other day and one   of them interrupted me and raised their hand  and they said we were so happy in the fall of   2015 again lar part because of how this game  this game of uncertainty allowed players to   stay involved because it no longer matters if the  game is a blowout what it matters is individual   performances even within a team setting so one of  the ways that we're seeing AI show up now in this   fan engagement setting in betting and gaming is  particularly in the area of personalization I know   you wanted to get into that a little bit later so  I'll be broad on this now which is this concept of   taking the ability to place Wagers whether that's  again betting or da buying tickets versus buying   a betting ticket and providing the ability to  cater that to each individual Choice some of   the lightning talks that we saw earlier looked  at this what contest do you want to promote to   individuals because it might be a better fit  to what that individual's interests are we see   this in areas of things like marketing tactics we  see it in terms of providing parlay bets that br   in multiple different areas of interest for  individuals you see it in bonusing are there   areas that might be more of an incentive to place  a wer versus others usually people like free money   this is not a phrase that you'll see permitted  by too many regulatory agencies but it's used   all the time nonetheless uh which makes this a  particularly key area for use of analytics and   the capacity to again CER to individual interests  another major of this is in the Integrity of the   game we look for anomalies not just in game  play but also in suspicious betting patterns   sometimes they're very obvious so if you are  for example a gambling operator and you see a   ton of money com in on an amateur football game  in Romania real example um it's probably not   going to be a particularly clean game there's  probably fraudulent activity there and that's   something that can very easily trigger uh activity  something that gets a little bit more nuanced   might be something like example Christine I were  talking about earlier which is tennis tennis went   through a major Scandal about eight years ago nine  years ago and they had to completely revamp what   these uh organizations looked like what did their  competitive scene look like because there was such   a big betting related Scandal so something that  might be a little bit easier to tweak would be   something like double false double FSE is a major  area of fraudulent activity in tennis and these   can much more accurately be detected by Ani now  than by individual activity some of these other   areas in terms of kind of you know moving more  into this consumer protection and game protection   space include areas like compliance uh beding and  gaming are two areas that are extremely Reliant   for their framework on the regulations that  create the framework in which they exist and   we're going to pause for a moment and say offshore  and illegal activity different different story but   anything that fits within legalized Frameworks  has to fit within that legalized framework um   if I were to say for example I'll turn to all of  you what is your most valuable asset as users the   most valuable asset to a betting operator is their  license because without the license they have no   users right so so it's extremely important that  you maintain that in place so Ai and it's it's   still a bit rudimentary but it's with anything  we do it's getting refine every single day uh   is used in compliance settings as well are you  doing something that might be pushing you out of   compliance in certain parts of the world betting  is regulated very very highly and in North America   it's done at the state and Canada provincial  level which means that if you operate in the   United States and more than one state you have  more than one legal and Regulatory framework that   you have to manage so AI is used not just in terms  of tracking compliance but also ensuring that you   fit within all of these different compliance  Frameworks and then I'll move into the last   element of this here the number one priority  for most of these compliance Frameworks is   consumer protection so that includes most of what  I already said but it also includes the very very   important area of what you somewhat generically  call responsible gambling responsible gambling   is the concept of preventing harm we talk about  problem gambling that comes up quite a bit that   is the that's the moment where you're in har  already where you're doing you need to seek   help you need to be able to access treatment where  it's available um so more generically one of the   not more generically but but on the prevention  side then one of the major areas that AI has   been a major contributor and was one of the most  or really is one of the first areas that we saw   AI applied in the betting and game space is in  uh behavioral tracking behavioral activity for   anomalies that might be driving toward harmful  activity for individuals and that is uh it's   incredibly difficult to achieve it's still sort  of considered the whole Grail we have come a   long way since the beginning but even within that  to speak to something that Natalie said earlier   right it's the humans and the loop concept  the hardest part of tracking human behavior   for potentially you know activity that might  be driving you toward problematic play that's   actually the somewhat easy part the hard part is  then communicating that how do you actually tell   somebody you have a problem because if you do  that there's usually you know a few different   outcomes that can happen the first one is screw  you that's the most common response to you think   that you might be going for the problematic  play but you also want to make sure that you   time this correctly because if you intercept  somebody at the wrong moment if they're not   actually edging toward harmful or problematic play  you may be driving away a customer and if they   are but they're not ready to seek help you might  actually push their help seeking further down the   line and so that's one of the major areas where uh  data analytics has become incredibly important is   because it really helps with that timing element  and it also helps test out messaging around the   communications element of that yeah I mean it's  almost like without going too deep into it this   point will see to send a note like to your mom  right um kind of thing that that you know you're   headed down a bad path and to get the kind of you  know response that you're hoping to get but it's   not there yet right um Christina you all are doing  really really cool work um at MIT Sports lab what   emerging Technologies coming out of the work  that you're doing at the lab do you think will   have the biggest impact on fan engagement in the  next five to 10 years yeah that's a great question   so I'm sure we all heard about um the immersive  external um experience that the future fans G to   have and we've got a bun a number of sports teams  concerned that it's going to be so good that folks   aren't going to want to go to Live Events um I  disagree I think it's really hard to emulate the   experience of all of those humans being excited  at the same time um the roller coaster of emotions   um that said I think one of the things to think  about with say engagement really comes back to   what we've been all been talking about is that  personalization and it's actually we talk about   it as Big Data it is awkwardly medium siiz data  so what does that mean it means you've got a few   areas that you got nice high resolution and you  got a 70 to 60 to 70% of it that has to be thrown   in because of questionable Integrity um uh unsure  um unclear uh how it was measured also a lot of   inconsistencies of that and this is where on  the fan site you would think that pro basketball   teams Pro um Pro uh baseball teams all the resorts  know who their fan base is you'd be shocked they   don't know who's in the sa right it's it's the  number is like 40% and that's good like that's   a bigger Improvement that's exactly and so well  in marketing efforts we capture all that right   well does an open rate actually help you better  understand what the best personalized message is   for someone who's 15 25 45 maybe my 6 demographic  where so I think better understanding what we're   continuing to here um how do we better understand  our our future fan and how do we understand our up   and cominging fan who will then convert to season  tiic the only reason I'm stepping back from the   technology pieces if we put the human in the loop  if I can't afford to go to a Warriors game can I   afford to buy a VR headset and pay for the content  I probably because the content is expensive to   produce for that experience to be better in my  living room and then do I have enough people   who are also going to be in that with me so I  think if I was to say I almost think we've got an   opportunity to sports has an opportunity to step  back and just say can we just better understand   what the lowy fruit is and if I was to look  actually in gaming they've done a really good job   capturing being sticky on a platform now how do we  better understand what the sequence of events are   that are leading to a higher engagement reason  I mentioned that is this goes back to what can   we do in stadium what does that look like and can  we better understand who those folks are during   the time so I think we're going to actually see  more technology going to the smart Stadium smart   venu so that they can better understand who's  in there when and I understand if my Saturday   afternoon or evening games are we driving  families are we driving Bachelor parties   what those are and then can we customize it in  inating experience and so um you know baseball   has very uh Rich season so also those tradeoffs  do I trade revenue for butts and seats essentially   because I want that energy there and so I think  we're going to continue see that and I think   finally if I was to say where where did we see  these future Technologies you have to have an   infrastructure a a fairly technical groups who  can also just maintain it right so the easy part   is the buy the hard part is everything all the  infrastructure own that goes with calibration   um prop betting I think is going to be a big one  in stadium how many of you had a reliable Wi-Fi   experience in stadium right and these profs have  to right so I think we're going to see that but   we're also gonna it's gonna have to be balanced  with can my user base financially it does it make   the broadcasters half like those deals that's  where Technologies are are being used first um   and can I better understand is it infrastructure  bottleneck or C bottleneck or am I just completely   missing how my 14-year-olds who are going to be  my future fand Bas are consuming sport today so   I you give you a lot yeah sure it's VR AR is g to  come first which is and I have a in a jersey and   have unique insights that are coming up in stadium  but I also think we have to keep a a mind toward   just because I have a drone is me flying it  through the stadium going to actually help or   hurt what the experiences reason I mentioned this  I was it uh so dfb who uh is the German Bali they   did one of those uh asual uh I was going to say a  real time experimentation that they're like okay   we're going to show you how incredible drone be in  stadium they had us all blocked off for safety so   we couldn't move around this the stadium at all  now imagine doing that in any type of Stadium   environment where we're already running into flow  challenges so I think that we're going to see it   more on theit from those owners who are trying to  understand who they are and then if we can better   understand who they are then we can actually use  the Technologies to optimize the engagement yeah   and and to give everyone some perspective here  because most folks in this room are making the   Assumption have some pretty significant Tech  chops right um and that is not necessarily what   goes on with professional sports teams right and  so to give you an example um about six years ago   I did a Consulting project with a major league  baseball team that owns owns its own stadium that   had an IT department of three That's Not Unusual  let that sink in for a second of three and one   was directly responsible for the president of the  team's it he was his own personal help desk right   so it's really the other two for everybody else  so then think about how much you're really going   to be able to do from a tech perspective with an  IT department of R right so speaking of baseball   teams um we'll turn to Jorge um Jorge we're  asking to divulge State Secrets here um can   you share insights on how the New York Yankees  are using AI to improve um onfield performance   and to the extent that you're engaged in it and  so with respect of play performance to the extent   that you're engaged with at Fan Experience um  so I I sort of wear a dual hat as a faculty   and being involved with the Yankees at a at a high  level I think with with the Yan and just like any   other sport when you're trying to assess player  performance you want to make you want to make use   of the you know the big part of the data P which  is really video and and there's a ton of video   that's being collected by the Yankees and every  other uh you know MLB sports team and so the the   the key the the main sort of job that that I've  had since I've been there has been trying to mine   that data to provide more insights about player  performance um and of course you know within the   quantitative team that's that's really the the  perspective um there is this interesting thing   as an academic to see that you know historically  it's had this strongly Quant background to it and   and and we're moving into this future of way  more data than any of the Beijing models can   handle and frankly for which the infrastructure  is not ready for it um and so slowly our our   team has been growing on that front we've been  we've been really focusing on that and I think   you know I think another another challenge being  part of the the sort of AI and computer vision   team there and leading that team is is you know  you don't always see the impact right away you   know where it's you certainly extract more um you  know new data sources from which new statistics   dve than everything else and it's somewhere down  the line getting getting baked into someone's   decision which at the end of the day could  either get folded into a decision or dropped   on the floor CR right and I think that's true in  any situation where you have this much data and   you're trying to synthesize it and figure out  who that's um in terms of fan engagement it's   a completely orthogonal process to what we do  um you don't think I mean at least I haven't I   haven't seen us really you know prioritize that in  any way it hasn't really come up in conversation   it's mostly about understanding the you know  frankly the player an asset and what are they   capable of I think there are a lot of really  interesting technical questions because these   are like you know athletes represent the tale of  the distribution in terms of human capabilities   and so really trying to pin that down having a  high level of accuracy and precision we often   have questions I I bring this up all the time like  what's our ground truth right it's it's not very   clear um but but we but we do you know like like  like I said a lot of other sports teams use the   standard Technologies use the data feeds that we  get from MLB um you know I I think there is a lot   of opportunities you know if we were to somehow  uh provide maybe aggregate statistics of some kind   to to fans I think they would really appreciate  that I think there I think in terms of you know   we we've been i' I've hinted at causality being  something we should look into um and maybe someone   in the department is working on it but it hasn't  been an explicit point of discuss well you so you   bring up an interesting qu really kind of chungles  around May CA is something we should look at but   you're taking and and it is kind of the modern  take on the the old school vers new school um   conversation that is largely pass us in a lot  of the kind of more traditional ways but it's   it's still there right right um but we think about  then the storytelling for you so what we're really   focused on here so you have so you generated some  Pro of really interesting insights right um on   particular players and performance and and things  of that nature but then Translating that for   the coach for the manager for the athlete is an  entirely different story right um and if you talk   a little bit about the storytelling process talk  about humans in Loop right so how do you translate   that I mean there's like the first level of  translation is literally going from I'm a computer   scientists I'm I'm talking to statisticians and  and so they think of the world as priors and   hierarchical modeling where like everything is  known or at least you have like a very explicit   assumption and I say no no we're we're like orders  of magnitude past that in terms of the amounts   of data that we now have and so yes we absolutely  need those because those those give you insight to   decision making but if you want to get it right  if you will or or maybe extract insights that   there's no beijan model for which you're going to  express a priority and learn something from you   really have to rely on these data sources that  you have in build data driven models and that's   been an ongoing conversation the entire time I've  been there yeah I mean I remember going back when   some of my former colleagues have really been  instrumental in in UCR and range Factor kind of   these new defensive metrics that were that are now  you know far more common um and the conversation   was brought the research had shown that D Jeter  at that stage of his career was no longer a really   good defensive shortstop and they presented the  material to Jeter um and Jeter's response publicly   was computer must be broke man right um so get  getting the the athlete to understand that and   then and then getting the athlete to again having  the actionable insights right is really different   so concern about information overload about impact  of the athletes confidence uh Factor things of   that nature uh particularly challenging yeah  and you know it's I think and there's new like   llms are huge right they're every um even just  having an llm to serve as a way of potentially   translating this information into like the tidbits  of like the main takeaways of the results of the   analysis I think is something that's not utilize  that we could to be to be fair to the Yankees   though I am a huge Yankee fan and I've been with  them for a long time so while I'm very critical   because I'm very critical um yeah so um you know  they they have been moving in that direction and   there has been more interest it feels like  there's a momentum growing and slowly through   these conversations both within the department and  from the Department to the coaching staff and the   players and frankly the rest of the organization  it's slowly trickling in but it's really a culture   change oh for sure can you give an example of and  it could be generic I you don't need to tell us   who the athlete is right uh but can you give us  an example of something where you generated some   insights through your use of AI that ultimately  has been been implemented um bottom clote that's   a hard one yeah I I I have hard time actually  pointing to an example of something that we've   discovered because it's so silent yeah and and  it's interesting that that's not uncommon for   that to be refra it's like oh the but they're  not really doing it right um it's really not I   mean it's it's funny early on early on so think  about six seven years ago we'll time it uh when   the Miami Heat had both Shane Bader uh who's very  data driven statistically oriented player um and   Lebron James playing for them and one of the data  points that came out was hey LeBron every time   this you're guard this player who you're guarding  in the upcoming game um 81% of the time I think   was the number that this player gets the ball in a  one-on-one U on the left Baseline he go he drives   left right and so he tells LeBron he tells LeBron  this um and the first time the guy drove by him   right okay um it was the 19% unraveled all trust  absolutely have one shot that's it right but and   so he said something to him because he got dumped  on right and and everything so uh the next time   like trust me the next time he get it he's going  to go left and he did it again and the guy went   left right and then he did it again went left  again LeBron came with a block and a seal and   they came to the bench at the next time out is  like hey what else you got man right um and so   having the success rate right starts to to impact  things as well so the hum can I just pick you back   um how many of you are working with pro sports  or in sport awesome um how many times have you   found a really unique insight and shared it  with someone you didn't ask for it and it went somewhere that's amazing so because I think  one of the big things that we find is scort   is exceptionally exceptionally conservative  you can explain away a loss if you've changed   nothing but you change one thing you go for the  three points and you don't do it I it does not   matter what the statistics say you called it  Incorrect and now you are in the media for the   next week so and or you get fired right I mean  it's these are exceptionally public conservative   of Industries it might say industry but they're  they're owned by usually family and so one has to   also understand that you in fan engagement youve  got more experimentation on the team side you have   a lot less wiggle room for what that has to get  to and usually it's when a team is not doing well   it forces so we have an Al who assistant manager  GM um the head of r& for that diamond they were   not doing very well for we went in baseball news  it and he said look it had to get to that point   for it to feel comfortable for the organization to  actually make some big changes and I think they're   thinking about that as their fan base as well  because they don't want to lose anything yeah I   mean it's easier to fail Dum than fail smart right  in in some ways right in in the traditional way of   thinking about thinking about sports I wanted  to add that when something is picked up it's I   think it gets really sticky and and and speaking  from a from a fan perspective you know obviously   you're trying to optimize for the performance  of your team and so you come up with strategies   and it takes years for the coaching staff and  everyone else to gain trust in those strategies   and then they Implement them but they Implement  them sometimes for instance in baseball to the   extent where the game got more boring I mean I  know that I I'm I'm gonna wear my professor hat   and my fan hat from that but baseball was boring  before they change started getting really boring   when they didn't have the clock and they  were shifting all the time it just wasn't   fun I would watch my favorite 80s and 90s clips  on YouTube because those games were much more   dynamic they weren't shifting it wasn't really  clear that uncertainty led to the excitement   small shifts in the rules of the game completely  changed the dynamic of the game qualitatively   and it made the experience much more fun my  son who's 10 years old didn't want he was not   interested at all in baseball and now because  of its Pace can sit and watch a few Innings of   he's not quite watching a full game but he's  getting there and I think it's in large part   because of the pitch clock and the lack of  the ship that got rid of the ship I want to   ask that's I don't know soccer is not very popular  but sometimes it can the opposite as right so in   soccer what happen is now the V you know they do  action in real time to get the offside decisions   and soccer uh it's a game which 90 minutes of  action you know fans are used to it and now with   every decision with every goal all the players  everyone is like every single goal is like checked   every single penalty call is checked and they like  look at the way the you know the one1 of the shoe   may be in an offside position that AI is it looks  on television with AI is confused end just making   a judgment call oh you know what I don't know  this one but this is outside but that KS skills   it right sometimes you can use AI technology to  actually in hands and make it better but it has   a reverse impact on the the and the engagement  absolutely what's it's really interesting to hear   this first I love you said the uncertainty that's  a bit more exciting of course it has but this is   such a great point with soccer and a great point  with uh with baseball in terms of the pace and   said shock but what there's really interesting  applications of this in terms of the data flow   and what that means for what we can do for live  sport I know you asked the question about live   sport twice so I'm G address it now so we we  see this a ton in betting and gaming spaces for   things like prop betting Inplay betting is often  used term for this props can Beil at any point   um things like Inplay betting micro prop betting  uh these are not unique they're not unique terms   there's overlap between all of these These are  low latency high frequency Nets these are things   that can be done they can theoretically they're  not yet but theoretically these can be automated   these systems can be automated which eliminates  for those of you who are unfamiliar with sports   betting operations actually the terminology is  quite similar to finance which Dean Chen mentioned   earlier as an application for AI uh Trading desks  is the common term for sports betting operators   if you're unfamiliar uh which makes it a lot  easier in many ways and harder for trading desks   to maintain the ability to provide this on in the  moment live betting opportunity or live engagement   opportunity as we think about this through gaming  and what's interesting about some of these data   flow situations is for such a long long time  soccer I I give a lot of talks in Europe so say we'll go with that but but with bookall  this was for such a long time considered to   be a continuous flow sport so when V was  introduced it threw a massive wrench into   trading desks because we had finally gotten to a  point where you're starting to see a lot of really   inding Automation and things had to change um we  started to see it we kind of we think about sport   transitions I don't know how you feel about  hockey here these days we're still pretty keen   on it in Las Vegas um but hockey again another  one of those more flowing Sports whereas when   you talk about things like basketball which was  a great example from earlier when we talk about   things like baseball we're talking about a lot of  discreet instances that keep happening the flow of   data is very different than it is or was in a game  like football or football and and when we think   about those capacity we have to change some of the  ways that we engage we have to change some of the   ways in which we offer some of these immediate  engagement tools throughout the game itself so   when baseball started getting to the more of a  flow that also changed the way that we engage in   real time and we're still somewhat figuring that  out again I think such a great point about there   being so much data and it's sort of funny that  comes from the gaming side of things and hear   people compliment for sticking this because in  our field like oh God nobody ISS going to do it   what what's happening we here why why aren't you  putting more people right three people comment on   on this sort of area um because this really is an  area where fan engagement can us to amplify Big   Time whether that's a beding product whether  that's a gaming product even whether that's   just an information piece of information that you  can share with a fan this all becomes incredibly   important in terms of how we're learning from the  data we have how that data itself is changing yeah   and it's interesting too I mean again you think  about the fan experiencea which you talked about   before um and The Fan Experience in baseball from  an accelerated game so you think about the user   experience in stadium so my first job in sports  when I was a high school kid I was a vendor at   veteran stadium in Philadelphia so I walk around  and sell soda and ice cream right uh during the   games great best high school job you could ever  imagine right um know fan Behavior oh believe me   um so but you would think about it so you think  about faster games for a vendor uh oh right I   mean that's problematic there's less time for  you to sell you're going to take a cut uh in   your it's it's commission based uh business  so you're going to you're going to take a cut   what they've found is fascinating right so so  far the FMB actually has increased right fans   are more engaged they're consuming more in  a shorter period of time now in some areas   alcohol that might be problematic of course  right but when you think about the the overall   engagement it's much higher even though you would  think given that the time is shorter right that   you be spending less money when you're at the  ballmark so so fascinating kind of insights um   that come from it and some surprising ones right  for sure I wanted to SP one point I think because   AI is so great in technology evolv so much  and we probably talk about revolutionizing   everything but there are not g steps that  we can take to make things a lot better as   so I had gone to the Super Bowl game last year and  I I think they gave us something to wear right and   which migrated and had lights and it's a very  small application that's Run in real time with   the music and that in the half time show it was  a great experience right like the ability to go   in and out of Stadium smoothly the crowd control  that happens as to which Gates you get based on   you know or a simple example of a p sentiment  analysis after game which sections of the fans   in the stadium had a good time which which are the  ones where like the view was not good for example   so behind the scenes there's a lot going on using  AI in small small different ways in identifying   that how can we make fans experience better  which probably doesn't come up front in front   of everyone because it's not like glamous any  because we we completely think of that there how   do we get like VR set for everyone $3,000 and give  them that experience right but that's not only you   can do a lot more with technology which is more  affordable as well right not just the things yeah   so like the Portland Travers right now um are a  team that's really taking a lead on AI uh in the   NBA on the business application side of it right  so one of the things they're doing specifically   is with respect to customer surveys um where they  can now they use AI to flag problematic responses   right from fans who had a bad experience at the  game and can you know really then forward identify   when it needs to be raised to a level of concern  where it needs to be forwarded to someone in their   customer service department to respond to that  particular complaint now there's a lot of teams   that do a lot of survey work and historically  they haven't had the resources to even read   the surveys to turn them into actionable insights  right so I mean that's that's a pretty remarkable   development now of course fan surveys are a little  that's heavily impacted by whether or not the team   won or loss uh when they're doing well so the  beer is always colder and the hot dogs always   are a little better um the game presentation by  the team is always much higher um when the team   has won versus when they lost even though it's the  identical thing The Identical product of course   but that's a really important Insight I think to  take where where AI has been extremely helpful   on that frun right can I offer up two extreme use  cases please um every single pro sports team we've   spoken to who's trying to optimize for the inating  experience what do you think is the number one   issue for their concern of the fan experiences  when going to a game a match whatever any guesses I'm trying to feed our  athletic director the answer here   park you have now just optimized for  the the frosty in middle of the Oreo   cookie but how much is it bookended by  the hassle to get there and then the hle is the bigest hey if you guys resolve that  everyone will love you forever but this is a   it it seems like you should be like a VR headset  will not solve that um it's interesting so I would   say for those um who may not be in the field don't  hesitate to ask a lot of what you might think are   dumb questions because we go in my me  and my colleague go in we say please be   patient with us we're gonna ask you a lot of dumb  question this is how we found out parking is the   biggest challenge right and now you've got likea  fraudulent parking passes I don't know where I'm   supposed to go I can't even get my right share  card in the right place that's a big one but   then if we kind of think about to where are the  winds happening well if we look at these younger   audience gaming is actually doing the best because  they're providing small Snippets of content that   are asking for some level of interaction and to  point now my team's not just focus around own   one team I might have six teams or essentially  entire lineup that I'm now following I think   there's a lot of interesting opportunity then to  start to look at other fields like mobile games   which are exceptionally good at keeping people on  the platform and then it's free and then you start   buying tokens right so I think there's opportunity  Sports Media or these package deals start coming   out and you start to think about what are their  ways to feed this future fan in the way that   they're currently consuming I ask my students  all the time how many of you are paying for K what's anyone who raises their hand and said is it your  parents yes they're watching Clips in Short and   they're saying I watched the G that's the other  thing the verbage itself is being applied to I'm   trying to figure out what my future season ticket  holder looks like but I have to work Way backwards   and I can assume that the way I'm selling today is  going to get me the same of result I think that's   where I'm really interesting is personalized  content in a small format and I think there   could be really an interesting opportunity to take  away the $80 monthly subscription fee that I'm not   going to buy at as a college student likely and  look at all of the free content and find ways to   then maybe have that at a higher resolution  a little bit more unique but then find small   are ways to pay for it yeah and the the old of  statistics on the games like like for they fantasy   sports platform and if you like created for game  and then you're watching on the mob phone and you   see your scores your ranks move up and down it's  just like a very different experience than just   sitting on a television and just not like you know  people need to be interactive so if you give them   that experience on the mobile you can do so much  more like the prediction given what's going to   happen next um so some examples like how you seen  these Sports and they're trying to do a field goal   like what's the probability of hitting it or not  it just keeps the engagement going throughout the   game as well so that passive viewing and that's  what fantasy sports kind Sol that passive of   games nothing to do is like I don't think people  that want they need a lot more engagement and   that's why these these everyone from of the world  everyone is investing in those apps right how do   we make the apps to make a lot more engaging how  do we create a connect with the user from our fans   and understand them a lot and then use technology  to actually sell them things get more imers into   experiences Etc yeah I mean I remember having  a conversation years in years ago with a senior   executive from the NFL when Fantasy Football was  really taking off and the phrase was very simple   and straightforward and one I remember it's like  this is like Mana from heaven from us for us right   um and so the engagement level is just so much  higher and then obviously with the with with   Pasa uh you know being overturned and then you  have the um you know the legalization of sports   outside of Nevada um and across the the the 50  states um leading up to the states to do so um   even more so right um because gambling is in many  ways like fantasy sports on steroids right um if   you think of it in in that context right so BR one  of the things that we talked about though this is   not all great right so we talked about microing  and a lot of the issues um you know how quickly   it needs to be done and everything else and some  of the fraud issues and integrity issues that come   with it but there's another side to the micro b um  issues um in particular and that's the impact on   player and player mental health um and sometimes  physical health but really mental health um if   you could talk a little bit about that um you know  just to inform the audience of that I know that I   did say if you primed me on this I talked about  it for a while so we can put a pitch clock on so so in particular one of the areas that Scott  and I had talked about earlier is player abuse   but before I talk about that Scott you actually  frame this question in a really ideal way because   you generically talked about players and so here  I'm going to talk about not just the players and   the athletes but also the players those who are  buying the tickets those who are placing the   bets and in particular with micro betting I want  to point out I just spent you know five minutes   talking about how engaging this is how important  it is how great it is to be able to have this live   interactive engaging tool Well turns out Frankin  perfectly Scott micro betting is also one of the   highest risk uh bets that is possible for when it  comes to predicting problematic part so you have   both the highest engagement but it turns out that  the highest engaged vs are also those who might be   more prone to problems so that's something you  always need to keep in mind when we talk about   engagement there's a cap on how engaging you  want to be especially when you're talking about   incorporating any of risk product you can talk  about that in terms of risk management that's   the very common business phrase that might be used  you can talk about it in terms of harm prevention   that's the more public health or psychology  phrase that might be used how you could talk   about in terms of compliance that would be your  law and Regulatory space same concept whatever you   want to do when you engage any sort of tool that  has risk involved with it is it's a balance you   have to make sure that you're optimizing what you  offer to everybody who plays in a healthy capacity   also limiting the capacity for any sort of harm  it's extremely difficult you will never get   optimal because human kind is a risk-taking being  that's part of the human in the loop P there uh   but that's an incredibly important piece to keep  in mind moving then to the player side athlete   side of this player of use is one of the things  that we've seen as this very big growth in in play   betting has risen is player use somebody you know  just betting on the outcome of the game we talked   about right it's betting on individual uh moments  within the game so let's say I bet on field goal   doesn't matter which sport we'll just go with  field goal as a whole and uh and it misses I go   online and I say oh screw I'm gonna kill you next  time I see you you lost me I've seen everywhere   from $5 to $5,000 or more uh that's it uh you're  you're dead next time I I conf find uh these are   real these death threats are very real bullying  as a whole is very real and this is something   that's incredibly important to to think about  because these are all both data generated types   of activities the human peace of it of course or  data is not telling people to write these things   but there also are ways to use data to look at  what might be happening to identify some of these   threats before hopefully they even get to the ath  themselves I'll also note this is not a solved   problem if you're looking for something to work  on maybe you're a student looking for a project   that you want to use to apply for the next round  of dream fellowships uh prevention of athlete   abuse is an incredibly valuable thing that needs  to be solved because it's incredibly harmful not   just for the individual who's the victim of this  bullying and abuse but also the bully themselves   we need to find out who this person is we need  to be able to punish them and we don't really   have a good mechanism for that yet there are  certain reporting structures in place you might   complain to the compliance director underneath  the Athletics director but even that that's not   solving the problem that's identifying it after  the fact so one of the major areas in which AI   can help is um identifying and predicting where  that kind of behavior might come from and again   not a solved issue yet something to be done in the  future so we're we actually don't have that much   time left based on the the clock that that I'm  looking at we've only got a few minutes so again   feel free students now if you others to to chim in  with questions um but with students present always   curious and when I when I hear career advice kinds  of pieces um so so first I I read a statement   recently um that AI is not going to replace you  right someone using AI will or hey what is your   thought with respect to that specific comment true  false absolutely um yes uh you know as llms do   more and more these days and you know generative  AI in general um they're not perfect obviously   they're getting they're certainly getting a lot  better but the human in the loop in that process   is really really important um providing that  feedback providing that level of judgment I think   there's something very subtle two things that  are very there one is that because they produce   such good content both from a coding perspective  from a writing perspective and everything else

2024-10-18 12:29

Show Video

Other news

11 Linux Distros That Feel Like Windows (2025) – Perfect for Windows Switchers! 2025-02-11 19:59
X1 vs XL: AMS vs Toolchanger - What's better? 2025-02-11 16:00
10 Amazing Facts About Tech & Science in the Baroque Era [The Well-Tempered Cast] 2025-02-08 23:39