Were going to continue our brave and bold theme that weve adapted from this morning, Im being very brave Ive got new shoes on, and I forgot how brave you have to be to wear them with no socks, so my feet are actually killing. So at one point Im sure I have to roll onto this stage and roll off, but anyway we are back. Okay, weve got four great sessions for you in the run-up to lunch. Weve
Also got our brand learning program which is taking place as well, so the next session in that witches are our advertiser brand only session is happening at 12:20 with day two Zoo. So, if you’re registered to go and do that, please go there, we wont be annoyed. Just a little bit, but here we have four great sessions, so continuing our bold and brave and stretching the limitations and working into new areas with media. We wanted to look at AI and how AI is
Working with ads, so we thought who better than to get any girl, who is the CEO of T Studios onto the state to take us through it lets give a big round of applause? Yes, so welcome. Thank you. Thank You. Jerry hello, everybody, Jeremy said my name is Amy and today Im talk about AI and mainly Im, going to talk about what I think the future of AD looks like. If I click to work that will be amazing. There we go so Im going to talk about the future of apps and before
I do that I wanted to tell you a little bit about why Im here on this stage. Talking about the future of app Im a software engineer and entrepreneur, I grew up in Bucharest, Romania. Now, if you dont know where Bucharest is you’re for Bucharest, great, so its the city where Michael Jackson went in 1993 up on stage in front of 50,000 people and said hello, Budapest, we love them anyway. So I studied computer science in Romania and
Then in 2009 I moved to London to start a company called brainian. What we did at brainian is we worked with advertisers and agencies and help them create interactive, dynamic, personalized video ads. This is an example of what we did so, instead of showing a user just a standard TV ad on on digital, we added this interact to layer on top or you could click on the right-hand side there where it says, discover more and you can find out More about the brand who created an immersive, interactive experience within the video and six months ago, my company was
acquired by peds and I was really really excited to be bought by teens not just because these is an amazing company but because they acquired my company just at the time that breaks it happened so as an immigrant in the UK I was quite happy to move out to New York however I moved to New York and this happened so so the French people in the room should be really happy that Im not planning to move to Paris all right so what does the the ad of the future look like I thought.
before going and talking about the edge of the future we should go a little bit into the past and give you a brief history of the ad and buy brief I mean were going to go back to 1472 when the first ad ever in circulation is documented it was a a a parchment paper placed on church doors by a guy who had written a book and wanted to promote it a couple hundred years later the first newspaper ad was printed it was an ad in The Boston Globe I think for for real.
estate fast forward 150 years later the first advertising agency was invented fast forward even more in 1922 the first radio ad ever aired for 100 bucks and then obviously of twenty years later the first TV ad to ever air was this very creative Im sure you all agree so this adds costed nine dollars so great values for money back then fast-forward to 1990s this is the first banner that was ever placed online he was on on.
hotwire com which is why magazines website was for AT&T and it had a 44 percent click-through rate talk about performance fast verb and more to 2005 YouTube sparked the kind of online video revolution 2008 Facebook launched their social media advertising platform and part the social media revolution in 2011 tees launched out street which is these ads place in content and.
then over the past few years the the kind of mobile revolution has been happening so if you look at the past 500 years what has changed well ads have become a lot more targeted we went from placing ads on Church stores to programmatically individually targeted ads they’ve become more measurable we now to a large degree especially in digital we know what the.
performance of our ads is and they’ve become more beautiful again we went from parchment ads to Charlize Theron on HD screens which is kind of nice but to a large degree were still serving the same ads to everyone especially when it comes to video so what I want to talk about today is what I think the the ads of the future will look like and I think theyll have four characteristics.
theyll be interactive theyll be adapted to mobile theyll be data-driven and conversational that last one might be a little bit confusing so Im going to go through them interactive I started by showing you an interactive ad that we created the brain Ians a few years ago so I want to continue by showing you what I think the next wave of interactive ads will look like and this is a campaign that Im going to show you that one a can Lions last year for best use of technology and I want to show you.
what it looks like and voiceover so this is the future head its a very popular band in the UK and what we did in this case is we took the future heads and this was for Bupa which is a health insurance company in the UK and each member of the band was a part of the body and as you watch the video you can click on part of body and you got a question how much do you drink do you exercise based on your response the band.
member will have a different reaction and then at the end of the video wed create a personalized video for you based on your responses so hows your diet good bad shocking if you’re in Italy and then last question passed into exercise lets say regularly and now that Ive answered all these questions the what we do is we create and automatically create an ad for the individual based on 240 different bits of video you can turn the volume up.
so maybe up yeah so again this was created using turn 40 different bits of video and it was so successful with the advertiser bought a million impressions the impressions and they got 9 million impressions shares on social media so it was immensely successful so mobile we all talk about mobile and how important mobile is to today and the reality is that when you take a TV ads that an advertiser is created and placed it on mobile it doesnt really work because its a its a little small video in the middle of the screen so for example it.
t–s we run mobile ads right and if you look at the ad in the middle that’s the standard TV ad that’s placed on mobile it doesnt really look that great so what you want as an advertiser is you want vertical videos or square videos and so on the problem is advertisers dont really yet have these vertical videos or square videos so what we are trying to do it.
with teeth studio is were trying to make these video ads adapt them for mobile this is Im not sure how many of you have seen this ad have you an ad for Kenzo so this is a beautiful gorgeous ad was created I think last year and when you take this and place it on a big screen it looks amazing however when you go and you place it on a mobile screen you really lose much of the much of the effect so what weve.
done this studio is this is our tool you go and you take the ad and you put it in our tool and what this does is enables you to adapt the creative for mobile so you upload the video and then if you preview this video on a mobile screen or this is a desktop and if you create a mobile version which you can do easily.
using our tool and preview this creative on mobile you will look something like this in the middle so its not that exciting what you can do with our platform is you can make it vertical with one click and give it kind of a 360 effect this way you play the video fullscreen and and people can tilt their phone or swipe on the screen to see the rest of the video so creatives that are adapted to mobile I think will be very important for the future certainly data-driven we have lots and.
lots of data as advertisers as agencies and were not currently using this data – its all to potential in order to personalize the creative and a couple of weeks ago my sales team at t–s asked me to talk a little bit about how data works and I kind of drew this diagram that they all love so I placed it in this presentation it looks something like this an agency or an advertiser has a DMP within this BMP they put in.
third-party data or first party data third party data is like cookie IDs and user IDs and information that they buy from third-party data providers first party data is information that the advertiser has like CRM data and so on you then take this data put it into the DMP and most advertisers and agencies have something called a buying engine or this is kind of a a tool that is used internally to create a kind of a 360 profile of an individual you then take this data and you can do two things with.
it so you can do targeting then you can do dynamic creative targeting means targeting individuals based on that 360 profile dynamic creative means personalizing the creatives in real-time based on that data this is an example so you’re running a video and instead of just running a standard video you can place information within the video that is tailored to this individual so in this case the individual we know they.
live in Edinboro we know they travel a lot to London so were pulling in information that’s relevant to them and this is done in real time for each and every single impression that we deliver and then lastly conversational so a couple of months ago I started getting really really excited about chat box how many of you have used chat bots so far unlike Facebook and WeChat and so on so if youve used chat BOTS some of them are really really really interesting and.
amazing on Facebook on WeChat and so on so what we tried doing was taking up a chat pod experience and trying to integrate that experience into an ad and this is what we got so this is an ad for Tommy Hilfiger and you have a little button that says chat to us and if you click on that button the chat pod engages with you and goes we want to see the collection or do you want to see behind the scene videos you want to look.
at products or look if you choose that it asks you what country you’re in so it knows what currency to give you the product in you then choose products and you can buy the products within the video so its kind of taking a one-way form of communication video and turning it into an interactive experience that can actually sell product so interactive mobile data-driven conversational what is making all of this possible well I think the the one thing that is bringing all of these things together is.
AI so I wanna spend the second half of this presentation talk about what AI is how it works give you kind of like a an AI 101 in ten minutes so in order for a computer to be called an AI or a system to be called AI a needs to do two three things it needs to have the ability to learn it needs to have a lot of data to teach itself and then you need to be able to use that data to train the algorithm very cost-effectively so AI.
has existed as a concept for as long as computers have existed since the 1970s however its not until the past few years that AI has kind of really taken momentum and become more of a technology that we can actually use and that’s mainly fueled by something called deep learning deep learning is a branch of artificial intelligence what is trying to do is its trying to replicate the way the human brain works so how does the brain hooman brain work well the human brain is a bit like Manhattan.
you have the streets in Manhattan which are the connections between the neuron theyre called synapses and then you have the Skype scrapers which are essentially the neuron blocks the brain is structured in blocks of neurons so what the brain does is essentially a massive pattern recognition machine it goes and it looks at things and it recognizes patterns at the very base level of these skyscrapers it can recognize very simple patterns like a.
line or a dot as you move up into the skyscraper you can recognize more complex patterns like a rectangle if you move up for the more its in recognizing or complex patterns yet like a table its kind of like a rectangle with lines so you can recognize that and as we move up to the top of the skyscraper it could recognize spaces which is kind of the most abstract type of pattern recognition so what computers are currently doing is theyre trying to.
take this structure through this way that the brain works and replicate it into a machine so how do you do that well lets say you wanted to recognize a face or you wanted to recognize that enough of a photo has a face in it the way that would work is you first look at whats called edges in that image so that’s an image if you look at the difference of color between the hair and the face that’s an edge and what what the system does is it then takes that.
edge and says does this age combine into being something more so it creates whats called features like a nose or an eye or an eyebrow and then at the very end of this cycle where each pattern is more of an abstract pattern you have the ability to recognize faces and computers are now better than humans at recognizing objects in images and videos let me repeat that computers are better than humans at recognizing objects in images and videos this is a competition called image net where the human error is around 5% in detecting or recognizing.
objects correctly in images and videos and there’s a team at Google that did research with on image net on this competition and they had a 4 3 percent error rate so they were better than human beings at detecting objects and images so then they took those same algorithms and obviously because theyre Google they fed those algorithms 10 million YouTube videos and and then they.
asked the algorithm what they see what you see here theyre not drawing by someone they are a representation from an algorithm on what its seen in YouTube videos so it saw a face you can see there there’s a human shape and Im not sure if you can see it but its there and then obviously this is YouTube were talking about so its all a lot of cats.
and so computers can you know now take photos videos and they can recognize these objects and images really really well whats fueling most of this is something called Moores Law what Moores law says is every 18 to 24 months the number of transistors on a chip double which means you’re getting twice the amount of power and it it becomes twice as cheap so if we if this trend continues and there is no sign that it wont by 2025 we should have.
computers that are able to compute the same number of computations per second as human beings which is quite amazing because 2025 is like eight years from now so deep learning lots of data and the ability to use that data to train these computers really really cheaply so this is all great but can artificial intelligence and computers create art we have this view that art is a pretty abstract intuitive endeavor and that computers will never be able to be creative theyll just be able to compute.
calculations so a team of researchers went out to prove that that is wrong so they went and they took an image took a number of images from famous artists like they took Picasso they took van Gogh they took Kandinsky they took Turner and they trained this algorithm on how to draw like those artists and then they took an image the one on the.
top left and they said dear algorithm please draw this image as those artists and this is what the algorithm came back with which is incredible and then they took the same algorithm and they said hey dear algorithm how about you create a new piece of art and the algorithm came back with this which if you ask me its like a quite a psychedelic drawing but its pretty amazing and then another team of researches did a similar but very different experiment where they took classical music songs.
about ten million of them and they trained an algorithm on what good music sounds like and what Im going to play is not it will sound like piano its not piano its an AI looking at the Soundwave of piano and saying well the the Soundwave that sounds beautifully is this Soundwave so let me create something similar so let me play this for you so this sounds very much like.
piano but it is not deliberate chaotic but it shows you that algorithms are becoming really really good at creating things that we would consider art so this is all great right ai is creating better ads its creating self-driving cars its solving things in medicine so what should we do you know what as people working in the advertising industry for the most part agencies advertisers publishers and so on what are the things that we should do to be prepared for this future well three things one I think we should build.
teams of engineers and creatives that work together we right now you know we have startups and technology companies we have creative agencies we have media agencies we have advertisers and they work in silos I think in order to benefit from AI we need to combine teams of creative poets and engineers secondly I think advertisers and agencies but advertisers in particular issue to.
invest in a DMP strategy DMP stands for data management platform its essentially a tool that aggregates all the different data points that and interactions that an advertiser has with their audience with their consumers and stores that data in a way that it can be used in the future so your algorithm is only going to be as good as the data that you have to train it so having this.
data available is instrumental and thirdly I thought I should end on a more personal recommendation so Im going to recommend three books that are non-technical that I think anyone can understand and if you read these three books I think youll be prepared for this AI revolution number one there’s a book called the innovators dilemma Chris criticism it is not necessarily about AI its more about disruption its about how companies dont realize what the future will look like until its too late and they become disintermediated.
BlackBerry is a good example no care is a good example and there are many many many more in history so its a book about how to see trends and how to jump on trends before they become mainstream the second book is one of my favorite novels is a book called the circle about a company in the future called the circle that’s the combination of SpaceX best-loved Google Facebook Twitter maybe not Twitter and and what that impact has on the world and they’ve made a movie with like Emma Watson and stuff its not very good so read the.
book its better and then the last book is called super intelligence its written by a guy called Nick Bostrom whos a professor in Oxford its the best book on how were going to get to have general AI dont be scared about the title its not a technical book and it really really talks about how were going to get there and then lastly dont forget its bucharest not budapest thank you very much not Jeremy thank you thank you very insightful indeed lets see if the.
audience have had enough coffee in them to ensure that they maintain the momentum we had before the break where they were all in a happy question in mood also guys if you stand in the back there are a few seats if first I seriously if people have got a spare one or two seats next to them in this side and here would you just pick your hand.
in the air where these hands are there are spare seats you no longer have to stand at the back you can face your fears and come forward trust me we wont judge okay now put your hands down anyone have a question for Emmy what was he so good there we go we have one right here oh its not a question that wasnt seat oh is that a question or receipt do you have there a question perfect because ever mic in there is coming to you right now okay I knew they wouldnt disappoint us anything I knew it weve.
got a good audience today hi I came to meet chambers all right and Im just wondering artificial intelligence there’s a lot of chat going on about it in the marketplace at the moment but most small players certainly add take players I think are a bit nervous to come out and say we put a Im going to drive us forward primarily because you put that next to my BM technology and other sort of businesses that are putting things.
out there who do you think is going to be the first player to really bring this into the space in that Zac so there are a number of players that are were using more simple versions of AI to do things so give it two examples at EADS we use something called machine learning linear regression to decide what individual to show an ad to in order to optimize for the advertisers objectives so their objective is to get as many clicks as possible well look at that 360 profile that weve created for a per individual.
and decide whether theyre likely to click or not so I think the basic versions of AI are already being used in by some ad tech companies in order to optimize to the targeting part I dont think its yet used enough for the dynamic creative parts for personalizing the actual that’s something that were working on and some other companies are working on so I and the second company that I know.
does a lot of stuff around around machine learning again relatively simple models like clustering various different users in different buckets and so on is AppNexus IBM I know where after me are have a team that is working on applying machine learning to marketing so there are a lot of initiatives theyre relatively nascent in their complexity but I think over the next couple of years well see a lot more of that happening because the.
technology is there.
Leave a Reply