Research in the Intelligence Age: Staying Relevant in the AI Revolution

all right hi everyone it is noon so we will get started with our next presentation uh i'm erica i'm a senior project manager at accelerant thank you all so much for joining us for our second day of the quarantine virtual insights conference just before we get started i wanted to go over our ground rules one more time so just keep in mind that we're at the mercy of the technology we might have some speed bumps here and there but we'll always get back to you so stay tuned if we run into any problems be respectful to each other and to our presenters in the chat in the comments um questions just be respectful of each other we're all adults here and use this time to network and interact with one another you can use our hashtag qvic2020 on linkedin and twitter to connect with other attendees and the presenters as well and please feel free to leave questions in the q a box because we will be going over those at the end of the presentation and with that i'd like to introduce our next speaker tanya franklin she's the vice president of digital analytics client and market intelligence at lpl financial and she'll be talking to us today about research and the intelligence age staying relevant in in the art of artificial intelligence revolution sorry about that and with that tanya i will stop sharing my screen and hand it over to you thanks so much for joining us oh thanks for having me are you able to hear me okay yes we can hear you just fine okay let me close and share okay sorry because i wasn't at the beginning of my presentation um okay so thank you guys for uh taking some time out of your afternoon i know for those of us on the west or on the east coast it's a lunch hour so appreciate you taking the time um my name is tanya franklin i've been with lpl for a little over three years now three years in july formerly was at lowe's here in charlotte the charlotte area for about 10 years and as was mentioned i'm going to not talk too much about like research methods um but really when i came to lpl um we are a part of the insights team is a part of a larger group that also has data science applied analytics business intelligence we also have data strategy and governance and so the insights part of our team is you know a little bit of a an add-on to a lot of this really heavy data focus intense um type of teams and so uh just in the last three years working within that environment it's given me you know increased appreciation for the data side of things and we were asked earlier this year to pull this type of content together and and so i i'm repurposing some of this a little bit to to bring to you today but the question that was raised was really you know as we're looking at the um just the evolution of data and you think about artificial intelligence and machine learning um you know how not just how the company needs to adjust that and leverage it to add value but you know even for our insight team um you know how do we live alongside of something like that how do we see how just the the tools and the the modeling around data are evolving and then you have these traditional research and insights methods sitting alongside of it so how can they live cohesively so the bulk of this presentation is going to give you just a journey down that artificial intelligence road to show you uh and i'll move over to the agenda here to show you what like what type of environment we're in right now it shouldn't be too much of a surprise for for those of you on the call but then like what are the driving forces that are making it happen within artificial intelligence um and then uh taking a little bit of a side path into just how ai within the computing space has evolved and how it's being manifested today and then what does that mean for consumer research um because it you know some folks could feel like the two are unrelated and then i think there's even um you know conversation out there of would this totally eclipse consumer research one day so i i just want to show how um that doesn't necessarily need to be the case especially if you address it in a way that's more complimentary instead of duplicative so getting into the intelligent revolution itself this is a slide that salesforce presented at one of its conferences and i thought it was telling because it shows how you know when we think about these different industrial revolutions that have happened especially within the united states you see how just this exponential growth and development has spurred as a result you know it started with the steam engine and then again when electricity became more prominent and ruled out and then we went into that third revolution of computing and computational power now what they're um saying is that we are in this fourth revolution that has to do more with intelligence so when you think about um this intelligence transformation and it's driven by a lot of a lot of the technology that came from the computing revolution but now it's there's like these uh divergence of different forces that are at play making it into you know a truly new uh generation of a revolution here within the states and across the world so what's what are those drivers that are behind it um this slide it's a few years old but it still um tells the story so you can see where um you've got three factors that are pretty hey let me turn this off sorry i don't know if there was a question i'll pause if there's a question otherwise maybe just mute your line i i hear a little bit of feedback and you might also hear my dog so i apologize we're all dealing with different environments right now um so these these different forces have been at play for quite a while but only recently um have they really converged and bolstered themselves into enhancing this artificial intelligence revolution so you've got processing power we all know um the just how the processing speed has been increasing over time um we also know that score storage costs are going down but what you really see is you know a much more sharp uptick in the just um sheer growth of data and we all we're just consumed by it today and these three things that are at play right now they really have a lot to do with just what we're capable of using all this fork so has the costs have come down as the storage capacities have increased and then also as the data has become more available that all leads to us being able to have um you know much more of a an environment that can foster this type of artificial intelligence evolution and with that you know a big part of it i think what i read was about 80 or over 80 of the data that is being produced today is unstructured data so that's like written feedback and the pictures that we post and and those types of things and so when you think about it you know there have been advancements in just research methods being able to capture and analyze that type of unstructured data but it's a process and it can be a very long arduous process especially without the benefit of like a machine learning solution or an artificial intelligence solution so it would be unrealistic for us to think that just traditional research methods alone could keep up with this this pace of change within this just exponential growth around data so with that um you know what we also showed on that graph was the proliferation of computational power so just the processing speeds and it's gone to a point now where um it can basically mimic the brain and not to get too deep into this but um they call these things neural nets so these algorithms and i i don't pretend to be the expert but um it mimics it mimics how the brain and the neurons in the brain operate and so what it can do is recognize the underlying relationships within a data set and it's a process similar to how the brain would also recognize relationships and so what they're saying is that the networks the neural networks are able to adapt um to really help generate the best possible result without necessarily having to tweak or change the output criteria so there's a lot of you know magic technical sauce that kind of happens in between that i'm not going to get into today but um especially in the financial and wealth management space a lot of this is being applied to help with things like forecasting fraud detection risk assessments even stock market price predictions so the other thing too is that you know i want you to kind of hang on to this term of human intelligence because you know while i'm talking about artificial intelligence there's also this element of human intelligence that that is sitting alongside of that so what the artificial intelligence is doing is trying to mimic the human intelligence and i think mimic is the key word here so when we think about the types of technologies that can be produced as a result of this you know there's a lot of disruptive technologies that are benefiting from ai and machine learning and when you think about how traditional research has played a role in some of these things historically it does make you wonder will the historical approaches either be um circumvented or or replaced altogether or is there going to be just like a really strong mandate to integrate these types of aiml solutions within an insights learning environment so moving on and if you have questions feel free to jump in but we also have time at the end but moving on to how this whole area has evolved we're going to i'm going to take you on a little bit of a historical journey here for a second and this is that that side path that i was telling you about because i think also embedded here while i'm talking about how this applies to um the insights profession i think also um especially given the environment in 2020 it was a very interesting uh learning for me as i was exploring this just how much uh diversity of thought so you think about that human intelligence just diversity of thought and and just how much diversity and the the differing of opinions played a role and how ai evolved to get to where it is today so when we when we go back to some of the very uh beginnings of what we see today some of you may have heard of this person charles babbage so he was he was from england and he had been he was a mathematician he also invented several different things he was the first in the early 1800s to come up with like a mechanical way of calculation so if you think about math math calculations being able to do that mechanically on a device versus um just with the human brain and so he created a couple different machines in order to to develop this he's being credited with conceiving like the first automatic digital computer um and he created something called the analytical engine so this was like the the ancestor of the the digital computer that we used today um and i remember not to age myself but i remember in school like learning um like computer programming how we learned about punch cards and things like that but this shows you like even back in the early 1800s babbage had a a design and a unique way of how he was envisioning how this analytic engine would work and it was based on these punch cards and so long story short you know he developed the idea and the schematics for how it would all work but never really brought it all the way to fruition a lot of folks sort of were skeptic at the time and felt like just based on what he was doing it was pretty much infeasible and he stopped working on the engine in about um the mid-1800s but along the way he befriended a woman by the name of aydah lovelies and she's been known as like the founder of computer programming and really what happened is she took some of his early ideas and renderings for this analytic engine and even some papers that were written on the topic and and progressed them to a state that they hadn't been up to that point and so what she was able to do was to be be able to figure out that it was more of a transition from calculation to computation so that there's there would be an opportunity for these types of machines to do more than just add up numerical figures and to do mathematical calculations that it could actually move more into this computation space so it it took even though babbage is credited with the invention and is known as like the father of the digital computer it took this woman years later to really bring it to life and she is more known as like the the mother i guess of computer programming um and then the last thing i was going to mention here which is also like an interesting uh history lesson on how everything had come to the current state is a fellow by the name of alan turing he was also in england and some of you may have heard of him already i think his name might be a little more well known than uh babbage or ada but uh he's all he's known as the father of artificial intelligence so where he first kind of came to into his own was um he had built a machine so that at the time that this was about during world war ii and there was this uh coding machine that the germans had called the enigma and nobody could crack the code and there were even movies and documentaries made about this and about alan turing but um so he was able to with along with other cryptographers was able to come up this with this machine to basically break the enigma code and it was estimated that as a result they saved quite a number of lives during the war and also were able to shorten the war by an estimated two years just because they were cracking the codes an interesting movie about this if you if you want to know more about it but fast forward so getting back to the the ai piece of it after the war he started focusing more on comp computing and artificial intelligence so he came up with something that's known as the touring test um it's also known as the um the imitation game so it's basically you have a person asking questions both to a computer and asking questions to a person and that questioner is trying to determine if the answers are coming from a human or from a computer and it tests the level of intelligence and basically the the um like the underlying premise was how well can a computer really function to the point where it's almost imitating a human being and he had theorized that you know by the 20th century computers could probably be mistaken for humans up to 70 percent of the time and i don't think we're anywhere close to that yet um but there have been some um some computers that that do get pretty close and like chat box i think a lot of chat bots some of them are you know getting a little bit closer but nothing has really reached that um that higher level of um really behaving and acting like a human and part of the the interesting thing too is like it's capturing dimensions of what i was referring to as that human intelligence but it re it's not necessarily um taking into account or substituting for that human level of consciousness and emotions and being able to be you know much more reactionary and this kind of gets into um you know like the continued importance of the human element of data and of research so with the exponential growth and the accelerated pace of change um you know this is kind of interesting because and this is more of that that continued innovation uptick to where you know they're saying that well this this chart says that by the by 2023 they were estimating that uh just processing power of the computer will surpass the power of the human brain and that we would eventually get to a point of what they call singularity which i say is aka the matrix because they're basically saying that you know it's a hypothetical point where the technology growth will just kind of take a life of its own and it'll become uncontrollable and you won't be able to reverse it and that's what all of these dystopian type movies are based on is like the the computers and the ai taking on a life of their own and really humans no longer having control over that computational power i feel like that's you know it's a ways often the fictional part of what we hear and see on the screen can make our minds wander a bit but it is a very interesting thing to think about though just the sheer processing power potentially eclipsing that of the human brain so the manifestations just quickly you know there are definitely a lot of benefits as we all know and some of you within your own firms might be adopting a lot of machine learning techniques and ai even within your insights practices but definitely perhaps within your data science teams and your analytical teams and we've seen some of this in in news almost on a monthly daily basis where uh different industries different companies are using ai to really revolutionize how they how they work and how they do business so there's a lot of benefits but i think also on the other side of things there's also a lot of risk and this is the stuff that usually makes more of the headlines than the benefits so and this is what strikes fear into the hearts of people is that um you know how are we going to continue to harness ai and ml um especially if you know we're saying that it might reach a point where we can no longer control it but um what what i want to talk about here is that while it presents a huge promise we do have to be mindful about how we use it and again that's how this human intelligence really overlays over top of this and why we it can't just be a this or that type of way of thinking when you think about data and insights generation so with the proliferation of data you'll have the positive you'll have the negative but ai the way that it works it's kind of like your brain so turing who i mentioned earlier he said that you know when when you're born it's almost like you have a uh like this whole computer in your mind it just has not been programmed and so through life you are programming your mind and that's the same as it is with artificial intelligence so you don't just create something and then it knows everything you have to train it so it has to have the right type of data coming in in order and and not too biased of data coming in in order for it to generate the types of results that are going to be beneficial if you're having things that are biased going in then the the output is also going to be biased so it needs you need to train that data just like how you would need to train a child so this is a a news it's a news clip from a few years back um i was just going to play this hopefully it comes through if for some reason you don't hear the audio just like chime in and let me know and i'll just stop it and tell you what what it's referring to but this is just a manifestation of some of that those watch outs why we still need that human intelligence portion to help within this ai environment hey tanya we can't hear it okay sorry okay can you hear me while it's while it's playing yes okay so i'll just talk over it um what they're showing here is basically a video that was created of president obama but it's not president obama they basically use technology to imitate obama so what you're seeing right here it sounds like him looks like him but it's not him and jordan peele who is a director is the person who is basically saying this and they're they're making it look like obama is saying it and um so this just shows you it it was getting more into like the idea of fake news but when you think about it it's the ai that is really powering the ability to do these types of things and so it just shows you that you know even when we're like in our environment even if you are looking at social media or you're trying to understand what it is your clients or your customers are consuming if there's things out here like this that are being created that are not authentic you can actually be interpreting things very incorrectly so it's just something to to think about i'll let me pause this um so this was an example of you know they had this girl ripping up a shooting range bull's-eye but they augmented it to look like she was ripping up the constitution and um like that jordan peel snippet they were showing him they were showing president obama um like saying something about supporting terrorists and so it's it's it's kind of scary but at the same time it just shows you that there's a level of oversight and a level of that that human intervention that we still need to we can't just take everything that's being fed in through social media everything that's being fed in through what what people are reading what they're seeing as at face value especially with some of the ai that's that's out there today so yeah i mean it just kind of sums up that with what they say with um with there's a lot of responsibility with things that are really powerful and so the way that you treat this it can it can be a good thing or it can be a bad thing um so when we think about how we want to move forward you know i i feel like it's important to continue to note that it's a it's a potentially very useful tool ai it i i shouldn't even say potentially it is a very useful tool companies are using it daily and but it will continue to require guidance in order to thrive we shouldn't just assume that it it's going to just circumvent the human intelligence piece of it and that will no longer be relevant so kind of going back to our our history lesson here you know i had already mentioned as far as aid is concerned you know that the um diversity aspect of it that you know there was a woman that was very instrumental into how computer programming evolved originally and then with touring um the interesting note about him is that his sexual identity he kept a secret uh back in the you know 30s and 40s and 50s it was not very popular if you had a sexual identity that kind of went against the the mainstream and so as a result of that he actually went through quite a bit of trial and turbulation um was even um you know lost his um his ability to be on different boards um was basically stripped away of some of his credentials and that led to him um taking his life and so i i only keep this part of the story in here because i i think as i mentioned that side path conversation not only is the human intelligence portion of it going to continue to be important but also within that the diversity of thought the diversity of experiences i you know that's always been something that i try to as a leader continue to value but this is clearly showing you that when you have that diversity of different perspectives and you know even to the point of different genders you don't just completely discount um that contribution because it's not a mainstream type of uh of a contribution so i just i just thought that was very um interesting and timely given the environment that we're in in 2020. so wrapping it up as far as implications for consumer and market research um hopefully i was kind of leaving those breadcrumbs along the way about the human intelligence piece so i feel like in order for um this to be something where in our profession we feel like it's a compliment not a threat there's a couple different things that uh or a couple different ways in which we should be treating it one is that we should be looking at it as working in working alongside and working in harmony with artificial intelligence and machine learning so ai and human beings you know to have that harmonious collaborative synergistic relationship and each of them would basically be augmenting on their strengths so one way to do this is to embed it within your process today so to actually take advantage of ai and machine learning within your insights function the other would be to actually complement it and sit alongside of it so i was just going to call out a few ways that you would that you would do both so this slide really speaks to more of that integration so integrating ai within your insights practice to evolve to evolve how you work how how you deliver information how you ingest data and deliver those outputs to your stakeholders so you know as we were seeing further upfront in the in the deck just the sheer amount of data it would take the human brain and just laptop computational power way too long in order to deliver those types of timely insights on an ongoing basis and so the benefit of integrating an ai solution is that the time or the speed to insight is is so significantly reduced um you know you cut down on your need for the the ability to have dedicated resources just combing through the data and the computer can do it you know exponentially faster so it can help you to automate tasks to perform your job more effectively and to quick more quickly generate those insights and then also um and i didn't really talk much about this but like the natural language conversation so a lot of that unstructured data um the ability to to use the technology to help pull out those insights um a lot faster than what we could do on our own without it so it would the ai piece of this enables businesses to to continually survey the market so it's not necessarily you know the point in time or the um like the dedicated engagements that we typically do but it's more of an ongoing constant stream and flow of information and insights that are coming in and out of that ai and ml solution um they're also you know creating those insights in seconds and it used to take it would take um so much more time in order for us to do it without so that's the one the one strategy and approach is to actually integrate the ai into how you do your your job and and how you create and generate insights the other way is to actually use it to complement what you do so you're basically enhancing so and and it's not to say you do one or the other both of these can exist at the same time but what this really speaks to more of is leveraging your traditional methods to sit alongside of the ai so instead of only using ai as a solution you're actually helping the the development and all of the the fringe details that go go around um creating an ai solution or an ml solution you're helping to refine that with the human intelligence part of what you can bring to the table so an example that you know we're in the midst of right now we're working on a proof of concept related to some artificial intelligence solutions that we're looking at bringing on board within the business you know we've been talking with that proof of concept team to really understand okay well what is it that we really even want this thing to do so to refine the question and to really help prioritize what it is you're trying to solve for the human intelligence insights part of that can really help to understand that it's a lot more straightforward if you're doing something like trying to predict stock prices but if you're trying to predict you know like next best action for your advisors or what the biggest opportunities would be for like content to be provided to your customers and and how that might be received like those types of things you might be able to build some foundational insights up front that can help the business prioritize where they should put the ai where they should should invest the time to develop the machine learning and the ai to really enhance the business across the board so that upfront prioritization is is definitely a benefit also helping build out the hypotheses so you know this might be more qualitative to really talk to your customers to understand okay if you think about what the ai would bring to the table hypothesize around is that something that would benefit your customers and how would it benefit your customers so that might be talking with them to say if if we were to able to provide you with something like this how how do you feel you might integrate that into your into your daily life or for us as a b2b it's how would you integrate that in to better serve your customers um the other thing too is to help ensure unbiased inputs so you know when if you have a infrastructure where you have a closer relationship with your data science teams partnering with those teams to understand okay well what it what are you going to have feeding into your modeling for for the um the outputs that are going to be predicted and you know even just from a human standpoint uh just kind of pressure test those to make sure that you're not feeding in um biased data or data that only takes into account like one segment or portion of your customer base so making sure that it's more of a balanced data set that's feeding into that development and the other thing too is to try and help verify and confirm authenticity so what we saw with that obama clip there's ways that we could go out and from a from an insights gathering standpoint understand like how is that information really being received and translated and is it authentic so there's some some investigative work potentially on the data inputs that are going to be used to build out those models the thing that we've talked about a lot is filling in unknown data gaps so if we know that there are different sets of data that we already have access to what don't we have access to that maybe it would or we think it would be beneficial to help with the development process of a machine learning or ai solution is there a way for the insights teams to help go out and get that gather that data and feed that in on an ongoing basis so filling in those data gaps is a potential opportunity and then just ongoing contributions for future refinement so once it's established and rolled out how can you you know talking with your data science teams and your stakeholder teams how can you as an insights professional help to gather ongoing feedback to see how the contributions from the ai and ml solutions are being received so that might require the the model itself might be refining based on the data that's feeding into it but how useful is it is it being received and so you can have those types of feedback exercises and activities with the stakeholders who you're intending to to influence so um you know bottom line the human intelligence piece of it does still have even even if you are only looking at integrating it into the methods that you're already using it's still the human intelligence side of insights gathering still has the power to to generate additional insights that can live alongside of an ai or ml solution and then you know we shouldn't discount the fact that we have the ability as people not as a machine but as people to still have that creativity to deal with these unforeseen obstacles or changes and and really change course very quickly in a way that perhaps the the ai modeling could not um so with that i will close um thank you for your time and thank you for listening and i don't know if there are any questions i'll stop sharing at this point but um if there are any questions happy to try to answer no questions come through yet but we will give it a couple minutes if anyone has any questions now you've wrapped up but thank you so much for sharing tanya i certainly learned a lot i think there's so much about ai that i just don't know at all so i definitely learned a lot and i do want to go back and watch that obama clip um i'd never seen that before it's just so creepy that yeah i can do that yeah and i like i said earlier i don't pretend to be an expert on this we have a data science team and they are like rock stars and we have a real rocket scientist on our on our team um he literally was a rocket scientist but um we you know it's definitely interesting just trying to better partner and collaborate with that team you learn so much over time and it helps you to start thinking and crafting ways and ideas of how you can collaborate not just sit alongside as a sister team but really integrate some of the solutions that you're trying to bring to the table yeah definitely let's see i'm not seeing any questions come through so let's see nope all right well we can just give you a few more minutes back in your day then thank you so much for joining us and thanks so much for a great presentation