at insights consulting being an innovative Church agency will really value very much early on experimentation with new tools new techniques new methods and also new technologies and we partner up with all kinds of teams and universities to do that and one of the universities or business schools we work with is Isaac Business School in lille france and steven is doing his PhD there and together today we will share with you some of the experiments that we have done in the field of artificial.
intelligence in the past couple of months so one of those new things that we experiment with is what we hear a lot about at conferences at conferences nowadays within the market research industry you hear a lot about deep learning automation machine learning big data artificial intelligence and I get the feeling sometimes that we are really using a lot of buzzwords I have the.
feeling sometimes that we are hyping all those beautiful words so lets do a little bit of Investigation are we really hyping all these terms first characteristic I think of a hive is that were using the same word for different things and different words to stay the same and I think we are doing that again I think if you look at all the different.
words that were on the second slide they all say about the same thing they all talk about what we call as an umbrella term in artificial intelligence and if you look at one of the definitions thats being given to artificial intelligence I really like this one artificial intelligence is the broader concept of machines being able to carry out tasks in a way that we would consider smart in a way that we would.
consider as something that humans cannot do or cannot do at the same speed and if you really look very carefully at all the words on the second slide it all is about its about machines doing things for us the second characteristic of a hype is lots of talking but not a lot of experiments not a lot of people who are already doing it not a lot of people who share case studies and if you look at the latest great reports by Green Book we get a little bit an insight into whats going on in terms of the talking.
versus the doing we see that 77% still look at artificial intelligence as something that might maybe have an impact on our industry but we are not quite sure what that impact will be we are not quite sure what it actually is and how to use it its only 23% who thinks it will be a game-changer for our industry and I think we can draw two.
conclusions from this little graph the first one is that most people seem to be talking about something they actually dont know what it is and most people talk about it and have never done it we make fari we make a lot of assumptions about the future I think even within the 23% that think it will be a game-changer theres probably only a handful of people who actually is doing it and I think its pretty dangerous as well that we as an industry that most of us believe that it will not have an impact.
on us but if you look at lots of scientists they all predict that it will have a huge impact on lots of different industries it will have a huge impact on our society but we as researchers still think that it will not hit us I think thats a very dangerous kind of situation to be in so we are inside consulting and Isaac.
School of Management we thought well lets stop the talking lets start experimenting with this lets try to see what artificial intelligence can do for market research lets start really doing it and its take the future into our own hat dont thats the type of stuff need to come up with if youre right before lunch so we said lets build AI applications for market research lets create prototypes lets create MVPs lets stash them in.
real life in real projects wet clients lets learn from the mistakes that we make lets make those tools better that we have created and especially lets start talking about it at conferences lets share our learnings with others because I think if we all start to build things – things learn and share it with others we will find out together what artificial intelligence can do for us market researchers so we are here today to share with you two case studies two.
little experiments where we have used artificial intelligence to make our work different better the first case study that Stephen is about to share is applying artificial intelligence to the world of research communities so generating insights engaging with participants like we saw in some of the previous presentations and we are going to bring a very specific application there that helps us to make sure that we can keep on engaging with participants not for the period of days but for a.
period of months and years and that moderators get a clue on how to keep the community healthy and a machine is going to help moderators to do that the second case study that we are about to share is around turning insights into impact so giving employees within organizations tools artificial intelligence tools so that they can access the insights that.
they need all the time and that they have the insights available when they need to make decisions why do we pick those two domains within market research because we believe that those two are changing the market research industry communities become the dominant platform or are becoming the dominant platform that we do market research true if you look at 2016 its already 5 percent of the global turnover in the market research industry and if you listen to Ray pointer he predicts.
that in the next 10 years 7 it will grow towards 70% of the budgets that are spent on market research will be done through a form of community platforms or community research so thats why we think if we apply AI we should apply it to one of those future methodologies secondly we have chosen the field of insight activation turning insights into action because according to a study that we did this is the biggest problem that clients.
see within our industry its not getting to better insides always its especially leveraging the insights that they already have because they say that only half of the insights that are generated are actually used within organizations so if we can try to fix that problem or if we find a solution to fix that problem that would also be great so two case studies using artificial intelligence in the world of communities and using artificial intelligence to make insights available and to turn insights more into action within organizations so Steven thank you Tom so for me its very easy to explain to you.
our two case studies using too futuristic movies the first one is Minority Report and the second one is the movie ER so lets start with a movie Minority Report so to summarize what movie is about it describes the world where murder can be predicted and be prevented so to recapitalize all movie is about let me start with a small movie trailer.
okay Jan whats coming double homicide one male one female killers male white $0 40 up a perimeter and Toma on roof Im placing you under arrest for the future murder of Sarah marks here the man has had the future can be seen all we have to run on are the images that they produce we see what base hasnt been a murder in six years theres nothing wrong with the system it is.
perfect I agree can be stopped tell me exactly what it is youre looking at well if we get any false positives we are resting individuals weve broken the law but they will the fact you prevent it from happening doesnt change the fact that it was going down the system cant be wrong okay to make the link of the movie with our case study let me make a cheesy analogy where murder is a threat for the road Tom Cruise lives in remember this engagement is a threat for online.
research communities when not enough members participate and a topics that are being organized in a community low quantity or whatever they say doesnt really say anything at all low quality you as a moderator may not be able to derive useful insights from a community anymore so therefore as you want to sustain your community on a long term its important to battle memory disengagement and the moderators can do a pretty good job about it but it still requires a lot of energy to detect and to correct it but what effect what if you can adopt.
the ideas from a movie Minority Report what if we can also come up with some type of Oracle that automatically alerts of future destructive behavior well meet our new approach called proactive community managements proactive community management uses prediction models using two months of historical data to predict for each member whatever future behavior will be in the upcoming two months future can be seen thereafter the moderator can act on that informations to take corrective actions for predicted member the engagement.
remember this engagement can be stopped so how do we do it so we leverage the data rich environment of the community and use machine learning techniques so here we adopt AI to identify useful predictors or explain future disengagement behavior we looked at all sentiment and activity variables of all the American members moderators in a community and we analyzed about seven million data points to come up with two prediction models to predict low quantity and low quality behavior now how can we make this practical in a community context if we combine it two.
dimensions quantity and quality with respective activation levels high and low we can come up with this type of framework so using two months of historical data and this type of framework we can for each member predict in the upcoming two months whether they will be a high potential the community star passivist or lawyer so we implemented it we tested it emerged according to two important measures detection ability and prevention capability for a detection ability we see that you we perform pretty good in eighty seven percent of the cases we are able to correctly.
predict for each member whether they will engage in low quality behavior for 71 percent of the cases we are able to correctly predict where they will engage a low quality behavior so to give you a reference point if you would invite a random stranger to the community and you would ask him to identify for a certain member where he will engage and low high performance behavior he has a fifty percent chance of being right moderator can be better than that due to expert knowledge but using as a.
tea security numbers show that you perform pretty good against the moderator and that our models are pretty trustworthy its also done an automatic way which is not done for the moderator second we have a prevention capability because we can predict member disengagement we can prevent member de jure engagement from impacting the community member is engagement can be stopped they have to read it some for a further analysis using the four quadrant framework of the previous slide that you sell for each memory can personalize the.
prevention action so what we did is we modified the content of the email or the automatic email that will be sent out to anticipate on social functional and hedonic needs to participate in a community some early analysis already show that this type of approach this type of CRM approach has a positive impact now whats the impact of this approach for the business what are the automation informational and transformational impacts first does it say with standing money yes whereas earlier the moderator would have.
spent more time on community moderator because now the moderator is automatically a relative future destructive behavior the moderator can spend more time on doing the analysis and focus on the research task at hand second informational we can see something that the moderator cant see a member will may be healthy today but not tomorrow and moderator may identify future disengagement for certain members but maybe not for everyone our prediction models automatically detect for everybody their future behavior the transformational is something we didnt do before we go from.
reactive community management were just doing damage control to proactive community management were trying to prevent an era disengagement from impacting the community so you can sustain your communities of long term now what are the process learning or our lesson that we learned in this a case study so first the database is key you need to make sure we have both enough volume in terms of measurements for the members and moderators or making sure the.
quality of all those data points is sufficiently enough so everybody is for all date and you can have seconds to increase the adoption of this types approach and the business its better to give up some predictive accuracy by choosing a different model to make sure the model gets implemented in a persons business so its better to make a model more believable and actionable for a business user you can have the best prediction model in the world with 99.
percent directive accuracy if the moderator doesnt understand it he or he will never use it so we can increase the understanding user in our third lesson using a wide box prediction models we can reveal some interesting patterns about our community we saw that if a moderator adopts a narcissistic writing style members will be more likely to engage in future design gay.
behavior so you as a moderator you should avoid a narcissistic writing style and focus on all the others so the future for community management is proactive community management now the second case study its refers to the movie her so to summarize the movie in one sentence its about a guy who falls in love with an AI system to summarize and.
to recapitalize the movie thats the mr Theodore Twombly welcome to the worlds first artificially intelligent operating system wed like to ask you a few questions okay are you social or anti-social social well how would you describe your relationship with your mother oh thank you please wait as your operating system is initiated hello Im here hi hi Im Samantha good morning Theodore morning you have a meeting in five minutes you want to try getting out of bed too funny okay good Im funny I learn everything about everything lovely you look for.
how long before you ready to do all you mean I sign your emails that you come through a breakup and youre kind of nosy okay to make the link of the movie where our case study let me again make a cheesy analogy where our lack of affection is a threat for the world jerk in Phoenix lives in a lack of using insights is a trap for inside their environments I think you all know the problem if you know an insight exists.
but you cant define it as they are hidden in all the PowerPoint presentation and all the Reapers that you have the efficiency or when you cant identify the right inside for whatever purpose that you need the effectiveness youre not really constructively activating the inside within your business so therefore if you want to really effectively use insider in your business its important to avoid.
this type of behavior and to battle Ts but what if you can adopt the ideas from the movie her what if we can also come up with some type of virtual CMI assistants who can helps us to the insight activation process well thats why we create an outside activation box Galvin Galvin is a smart assistant and the AI chatbot for market research right now its created for three purposes first to meet the consumer so AG Galvin uses some predetermined consumer segments and uses also all the inside it has available to them to impersonate add.
a consume to impersonate the consumer so hacker can have real chats and ask any questions to your consumer so it is simulated chats fire a computer its like a real chat with your consumer second show me the latest news insights Galvin is your coach when you wake up or you want a inspiration for new topics you can just ask Calvin he will.
recommend and provide you with relevant insights for any new topic that you have here is three find me some insights about specific topic Galvan is your personal when youre in a meeting or an urgent need for any new insight you can just ask Galvin Galvan will provide you the answer if it doesnt answer it directly correctly no problem he will fill up with some follow-up questions to make sure that you have your right answer so Galvin is connected to an inside database its adults AI to understand.
whatever youre asking him and has the right logic system into place to respond to you with the right answer so we implemented it and we tested it and we saw that it really eases the inside activation process it gets adopted widely why because its so simple it just it is just chat people are also really satisfied by it because of low adoption barriers its not perfect yet.
but its still way way more better to go to all the PowerPoint presentation and all the reports that you have define a specific insights what the impacts on the business was the automation informational and transformational impact automation all it saves us time because its a whole its a person digital chatbot and you dont have to look yourself for the specific insights because Galvan review the inside when you need it its save you or somebody else his management time second its.
just in your pocket its mobile you can access her inside anytime anywhere you need it so as formational because of the low barriers to use it you and it just the focus on the insight and a consumer it improves your consumer centric decision-making so what are lessons that we learned first the logic the Madame Chapman and must be intelligent but also human so in addition to providing the logic to.
respond to your research related a core question questions you also need to put in place some other logic to provide some small talk so you as a user perceived it can respond to many questions second relevance you first need to map out and define that you will be use cases so you can really look at whats possible and which use case are.
we going to focus on so we focus on three research cases to really make sure that our user really understand what hes doing and you will definitely use it adoption by associating personality with your chat box but we like we did with Galvan it will humanizes at the chat bot and it will increase chances to adopt a chat bot within your organization so whenever youre looking for another insight in your business think about Galvan I think these two case studies are just the beginning it.
are just two applications its an application in research communities where it helps us to see which members will in two months from now disengage and we can prevent it today but we can also think about AI helping us with moderation ai helping us with analysis so this is just one application in the world of research communities also if you look at Galvan the chat bot that will help you to find the insights when you need them from an insight database from all the previous research that you.
have done this is again just one application we can think about many other ways that machines are going to help managers CMI managers research managers within organizations to do their jobs and I think that will be great because we already see within insights consulting where we apply automation or artificial intelligence that its not thinking away jobs that its not lets say an alternative for.
what people are doing on a day to day basis it takes away the dirty jobs the monkey jobs the things they dont want to do or the things they dont can do and it lets them focus even more on whats really important a community manager on what are the business questions and how to translate that and do great research questions and having.
personal kind of chats with Zoomers and when it comes to the CMI manager within an organization having more chats with your internal stakeholders focus on the insights that will really transform and grow the business rather than answering emails picking up the phone with the next question do we have a little insight about this or that somewhere in a report from the past so dont be afraid of the.
future dont be afraid of artificial intelligence lets all start to experiment more thats all and I hope to see you on another conference where you share your case study so that together we can find out what ai will mean and so that we can take the future of our industry into our own hands thank you very much youre.looking forward to.
Leave a Reply