Episode 3 - Artificial Empathy / Simon Kemp / Founder CEO / Kepios

My interview with him for the 'Shiny New Object' podcast is now live. We talk about Simon's Shiney New Object, artificial empathy and imagination, plus how the world uses social and digital.

Episode 3_Simon Kemp-CEO-Kepios

Mon, 10/12 12:16AM • 1:07:45

This transcription of the podcast was made using 3rd party automation service so it may not be perfectly accurate.

SUMMARY KEYWORDS

people, data, report, voice, countries, bit, point, machine, internet, read, context, year, world, empathy, based, spend, number, influence, talking, business

SPEAKERS

Simon Kemp, Tom Ollerton

Tom Ollerton  00:00

Hello, and welcome to the Shiny New Object Podcast. My name is Tom Ollerton. This is a podcast about marketing technology. Each week or so, I interview one of the most interesting thinkers from the marketing and advertising industry, about their favorite new bit of marketing technology or their Shiny New Object. I asked them to go into a bit of detail about how it works, what it is, but also what are the practical applications for brands and agencies to help you make sense of whether you should take on that technology or not. This week, I interviewed Simon Kemp, who is the founder and CEO of Kepios, they are a consultancy based out of Singapore. And they, in their word say that they help brands make sense of the future. And they also do a very good job of making sense of the present, you will probably know Simon's work. From the last few years, every year, he produces a report called The Digital in 2018, Global Overview, which I think at the last count, the last deck was around 5,000 slides long. And it looks at Internet, mobile, social digital penetration across the planet from hundreds if not thousands of different data points. It's literally fascinating and unique document. And but I don't need to tell you how good it is because Simon does a very good job of that. And then once we talked about that, we will get on to Simon's Shiny New Object, which is Artificial Intelligence and Empathy. I made the mistake of going head to head with him and really sort of trying to take them on on an intellectual level, which didn't go well. And you can hear at the end of the episode, but Simon is a lovely guy. super smart, and has a lot of good things to talk about. So, let's let him crack on.  So hi, Simon.

Simon Kemp  01:58

Hey, Tom.

Tom Ollerton  01:59

I just seen you live.

Simon Kemp  02:02

Live and direct on the interwebs. 

Tom Ollerton  02:04

Yeah. So, you were doing a live talk about your global digital report? 

Simon Kemp  02:09

Correct. 

Tom Ollerton  02:09

2018? 

Simon Kemp  02:10

2018. 

Tom Ollerton  02:10

On stage, We Are Social. 

Simon Kemp  02:12

Yes. 

Tom Ollerton  02:13

And I think it's quite a remarkable performance. Because there's been more snow in London in the last 24 hours than has been in the last three decades. 

Simon Kemp  02:20

Yes. 

Tom Ollerton  02:21

But yet you still got 100 people to get on, which is incredible. 

Simon Kemp  02:25

It was good to the fact that people turned up that early in the morning, in this note to listen to me talk about data for an hour, I'm still just blown away. I don't understand people's priorities. A little bit peculiar. But hey, it was great. So, we had lots of people, lots of questions. So, we did about 20 minutes of numbers. And then we went into the trends for 2018. So, what things marketers need to know. 

Tom Ollerton  02:42

So, tell me about the report. I've been obviously, I've read every page of every letter the last three, four years. I assume people listening this podcast have absolutely no idea what it is. Can you give us a kind of sync roundup of the report? 

Simon Kemp  02:55

Yeah, so it started seven years ago, it was a way of me basically being a little bit lazy. I used to get a lot of emails from people saying, "Can you tell me how many people use Facebook and Vietnam?" And stuff like that. And I thought, "Sure." But after you've answered the 15th email about how many people use Facebook in Vietnam, I thought, there's got to be an easier way of doing this. So, I started collating those numbers into a report, which we published on SlideShare. The first report we did was 16 pages long, and that was back in 2011. And we've now got to the 2018 report. 5,000 charts. 5,000 and older manually. Yeah. And so Sarah, who is sitting with us at the moment, but hiding behind the camera. So, Sarah, and I basically put together 5,000 charts of PowerPoint numbers and Tom's face at this point, listeners, is that quite a picture? 

Tom Ollerton  03:42

Yeah. 

Simon Kemp  03:43

Tom loves PowerPoint. 

Tom Ollerton  03:46

And so what happened with that first report? So, presumably for 16 slides, how many, how many views did you get on that? 

Simon Kemp  03:52

That one, I seem to remember, we had about 20,000 views within a few months of that. So, it was really quite impressive straightaway in terms of the traction that it got. And a lot of that was people like journalists. So, you know, I put it out, there's a little bit of a way of saving myself answering inbound inquiries to the business. This is when I set up We Are Social in Singapore. But we had a lot of journalists around the region say, "This is really interesting. Do you have any numbers on the following?" "Oh, yeah, I could get that data." So, it grew organic really, really quickly. So, there were people saying, "Where will I find the following bits of information?" "I can get that for you. I'll just create a slide."

Tom Ollerton  04:24

And give me a sense of where it is now in terms of reach. So, you've got what 5,000 slides...

Simon Kemp  04:30

Yeah, so we break it down into a few different reports now, so we have 239 countries around the world, countries and territories, probably because you'll then get listeners saying "There aren't many countries," which is probably true. 239 territories, we've got quite considerable volume of data for each of those. So, we have to break it up into individual reports but across I think we had about 40 reports this year. But last year's numbers for the full year when you look at people reading previous year's reports as well as that current year's report we get 5 million people read in the report every year. 

Tom Ollerton  05:01

And who are they? Who's reading? 

Simon Kemp  05:02

It's really interesting because slides should gives you insights. Som if you've ever uploaded something to SlideShare, if you go into your analytics section, then it tells you where individual viewers have come from based on their IP address. So, a lot of that is just based on what mobile network they're on. So, that's not massively insightful, tells you what country they're from. But if they're accessing from a company that has an IP address that is easily identified, then it will tell you, so I know that people at Apple are reading it, people at Google are reading it, people at Microsoft, all these different companies. And actually, at the event this morning, we had a lady from, she's, she's asked me not to say it publicly, but it was a government from a Middle Eastern country. And she came over and said, "We've been using your reports for the last few years to guide government strategy. I'm wondering if we can have a conversation about this." Very exciting. So, all sorts of people, students, governments, companies, obviously, journalists is a big part of it. So, we'll get a huge number of inquiries from respectable outlet. So, the Wall Street Journal quite regularly, we have a conversation, there's a guy based in India who regularly wants the latest numbers on something, and yeah, so it's a broad spread. 

Tom Ollerton  06:06

And who, and who's the, who's the kind of weirdest company or business or country, or person... 

Simon Kemp  06:13

You prepped this question, didn't you? And I've had Pornhub. They didn't actually contact me directly, they wrote a very long blog post around what the reports we did a couple of years ago. And I got freaked out because, it, the blog post linked back to my blog post, and you know, I was still very nerdy at that point, I will check all my backlinks. And it said, Pornhub is linked here, was like, "Oh, my God, I've been hacked, what the hell's going on," and then realize that Pornhub has an incredible analytics. It's the A-N-A-L-etics. Sorry. Moving on, everybody's groaning, for listeners that can't hear and see the sighs, but no, they they've got this amazing analytics part of their business and they were looking at the amount of time that people spend online in different countries, and then sort of aligning that to some of their data on what people are doing on Pornhub, by country as well. 

Tom Ollerton  07:02

And so, what do you put the success of the report down to? 

Simon Kemp  07:08

Mainly giving it away for free. So, we give those 5,000 charts away completely for free  and it answers a lot of the basic questions that people need to have answers to, but don't desperately want to spend a lot of money to find. So, especially if you're a, you know, if you're a small business, you've got a marketing team of one, you need to know whether you should be investing your money in Facebook or Instagram, you just want some basic numbers that are reliable, and you can track over time. And we make that available to everybody, and especially journalists, you know, they don't want to be spending days researching what's the latest number of people using this, in this country. They just want a single source. 

Tom Ollerton  07:41

And so then how do you monetize that? So, that 5,000 slides, I know it's all third-party data mostly. And so how do you then turn that into a business? 

Simon Kemp  07:52

So, that answers all of those low level questions that I just mentioned. But with every low level question answered, you immediately move on to a higher level question. And so you know, we put our branding, obviously, solely, so We Are Social and Hootsuite branded activity, I've got my contact details at the end of that, as well as the contact details of some of the other, you know, ways of getting in touch with the businesses that are a part of the report. And we just get inquiries all the time. So, if you think about it, 5 million people reading that, out of that you only really need three or four people a year to have a need for either deeper data or to say, "Great, these are lovely numbers, what do I do with them." 

Tom Ollerton  08:25

And so, I'm assuming no one knows about Kepios. So, they go this surface level data is really interesting. I should be on this, this channel in this country at this time for this audience. So, what kind of inquiries come to you? 

Simon Kemp  08:39

So, most of the ones that you just mentioned there, where it's sort of talking about a platform or something like that, then that will go through to We Are Social or Hootsuite because that's the business that they are in. When you've got, say, a conference that wants to have somebody presenting on the future of digital in Asia, that's the sort of thing that I would do, and most of that's paid speaking engagements. So, that's kind of how I make a lot of the money is. 

Tom Ollerton  09:00

So, that's interesting. That's a regular gig for you is talking about the future, even though the product is retrospect. 

Simon Kemp  09:05

Yeah, isn't it funny, but I think the nice thing about the numbers that we've collected, because we've now done seven reports, each, you know, annual report, so over seven years, you start to see the trend. So, it's relatively not straightforward. And you can't do it with massive confidence. But you can see it's going this way, and you can see deviations and whatever else. So, a lot of the time. So, I'm noticing a lot of the underlying really small bits of data that you can then start to translate into much more interesting insights. So, last year, for example, we were the first to report on the fact that there are more 18 year olds using Facebook than there are 18 year olds in the world and then got picked up by all sorts of other companies that pretended it was their data. 

Tom Ollerton  09:43

So, talk through that. 

Simon Kemp  09:44

So, this is even more complicated this year. So, if you look at the data of Facebook's active users, there are more Facebook user accounts that report being 18 years old, then there are 18 year olds alive in the world today. 

Tom Ollerton  10:00

And by what? 20% more 18 year olds or? 

Simon Kemp  10:03

So, this year, it's 12% more, but it varies considerably by country. So, that's the global figure. But you get into certain countries. And it's we're talking quite startling figures, I'm not going to reveal them yet, because we need to sort of check that data a bit more before I'm comfortable talking about it. But we're talking double and more that. So, this is really interesting. It varies massively by culture in my country, but a lot of that is driven by things like vanity. So, me pretending that I'm 18, when I'm not, a lot of it is data privacy issues, I don't want to tell facebook how old I am. So, I'll just tell it, I'm 18. So, it's not just weird stuff that's going on, there is some weird stuff going on in there. So, there's a little bit of so there's this, I want to stress because this is probably going out publicly, I want to stress that this is not a Facebook created issue. This is dodgy third parties creating profiles for their own commercial benefits. So, you do have around the world click farms that are basically creating these profiles that then click on adverts that those click farms are selling to unsuspecting advertisers. So, they make a huge amount of money selling adverts to people that don't exist, that they've created themselves. An awful lot of those advertisers want to target the sexy young things, you know, if you're an 18 year old, then you're pretty lucrative kind of target for a lot of brands. And so, you're seeing a lot of 18 year olds being the desirable audience that these brands want to relate to. 

Tom Ollerton  11:19

So, to stitch a couple of those sentences together and tell me if I'm doing that in the wrong way. In some countries, there's almost twice as many 18 year olds on Facebook as, actually exist. And the large majority of that extra 50% of 18 year olds is coming from click farms. 

Simon Kemp  11:36

I wouldn't necessarily put my name to say the large majority of that definitely plays a factor in some countries, especially in Southeast Asia, that's a particularly important issue. And we've seen this reported in the news. So, they walk into a there was a click farm in Thailand on the border. And they had 350,000 SIM cards. Now each of those is going to be our user profile, whether that's on Facebook or another platform, you know, it's difficult to say with any confidence, but that's what that was just banks and banks of phones, individual phones, individual SIM cards, then and they've got systems that control it. So, each of those individual SIM cards has its own user profile, and that user profile is doing things on Facebook or on Twitter or whatever else it may be. So, I would imagine there's a very similar sort of process to the way that governments have been manipulating elections, for example, they are going to have large numbers of profiles that need to be involved in this to make it work at scale. And it's farms that are doing that. 

Tom Ollerton  12:36

And so, how do brands defend against that? 

Simon Kemp  12:39

You've got to buy your media from reliable sources. It's the only way of doing it. So, if you're buying it from a suspicious third party that's giving you incredibly good rates compared to what you've seen from reputable companies, if it's too good to be true, it's probably a scam. The vast majority of advertising that goes through platforms like Facebook is totally legit. So, even though there are these significant numbers of fake users, they probably account for an incredibly small part of the overall total. It's just that you need lots of them to make that business work, so... 

Tom Ollerton  13:03

So, that's a that is an amazing story. I've seen the the photos of these click farms, it's really, it's literally just a farm of phones that people use is incredible. And so what are the other anomalies or interesting quirks in the report? Like what's your kind of favorite thing that you've noticed? 

Simon Kemp  13:25

My favorite number this year, if I'm allowed to have a favorite child, and the favorite number this year is that people will spend a billion years online in 2018. Billion years, that's not, not a typo if you can speak a typo. 

Tom Ollerton  13:35

So, bear in mind is, large numbers in marketing conversations. But what does that mean? A billion? That sounds like a lot of years. 

Simon Kemp  13:44

It's a lot of years. 

Tom Ollerton  13:45

But what is so what do I do with that number? As a... 

Simon Kemp  13:47

As a marketer, you... I think as a brand, you don't do a great deal with it. As a marketer, where you still have to justify to the powers that be in your organization that they should be investing money in digital, staggers me that we still have to have these conversations, but we do. So, the average internet user spends six hours a day on the internet already. So, as you start to break that down, 4 billion internet users, six hours a day translates into a billion years of human time in 2018. Right, so you then look at it and you go, that's interesting, 4 billion people more than half the world's population. Most people now use the internet, that is a very important fact. And that's across every country in the world. And from newborn babies through to the age of 100. So, 53% of the world's population, now using the internet. The six hours is an average. So, you've got countries like Thailand, where people are spending nine and a half hours a day on average, on the internet. And then you've got countries like Japan, where they're only spending an hour and a half, two hours a day on the internet. And it varies quite considerably by culture, by age, by you name it. So, you then start to look at, "Okay, so if the average is six hours, what are they doing in those six hours?" We've got some great data on this year's report from App Annie. So, you know, most of the activity that is connected activity at the moment is through a mobile device. So, if you start looking at the apps that people are using, then it's amazing to see the stuff that they're doing. So, the app categories that are amongst the top hundred, top 200 downloaded the most use, it covers just about every aspect of life. So, you've got the things you would expect playing games, chatting with friends, watching videos, that kind of stuff. But then you've got things like finance apps, so the ability to sort of, you know, manage all of your household spending and buy stuff and whatever else e-commerce, obviously. You've got health tracking. So, you know, whether it's as simple as counting the number of steps that you've done in a day, right the way through to things like tracking your blood sugar levels for diabetes sufferers, whatever else. 

Tom Ollerton  15:32

Just to qualify one point, when of those six hours a day, does that include sleeping apps or Fitbit? 

Simon Kemp  15:41

Can do. Yeah, so you've got friends, what we talked about earlier today, as well, you've got people that are actually connected 24 hours a day. And it will be you know, I'm monitoring my sleep, and it's sending data across. So, yeah, six hours a day, let's face it is it's a significant amount of time. So, inevitably, there's going to be some passive activity in there. And there's also going to be some double activity. So, I'm watching Netflix, chatting to my friends on WhatsApp, that's two activities. I don't quite know the way that the, I don't know whether that's going to influence the amount of time because the amount of time we're reporting there, the six hours a day is based on surveys. So, global web index, the providers of that data, fantastic data, they survey 18 million people, is on their sample panel base around the world. And they go to these people and say, how much time typically do you spend on the internet a day? I believe it includes business time as well. So, at work, internet time, but still, it's not like the duplicating time is actually going to influence the overall stuff, so. 

Tom Ollerton  16:36

So, sorry, we got kind of slightly cynical here. In no way am I trying to devalue the report. Quite the opposite. But, so, people are on the internet a lot. 

Simon Kemp  16:50

No, shit, Sherlock. Yeah. 

Tom Ollerton  16:52

And yes, there's that argument that budgets should be pushed in a digital direction? Yes. It's weird that we have to argue these points. But what else is that report telling a brand in the UK that they can act upon? 

Simon Kemp  17:06

So, it's helping you make choices between particular social channels, for example, so should you be on Snapchat or Instagram, this is something that marketers seem to battle with on a day by day basis. So, we've got some numbers in there that were helped to inform that for you. It's not there to answer your questions for you. It's there to inform the answers that you draw from a variety of different sources. In actual fact, if it's doing its job better, it's giving you better questions, not simpler answers a lot of the time. So, it's got data in there about e-commerce, the fact that the UK, amongst all the countries in the world is the highest penetration for e-commerce, and it's the greatest average spend per person. So, 2,000 US dollars per person per year in the UK on e-commerce. And in fact, there's only, when you compare that then to GDP, there's only one other country where technically people are spending more now it's it's you know, it's weighted because of the GDP stuff. But only China spends more per person on a GDP average basis than the UK. 

Tom Ollerton  17:58

So, I'm going to take this radically, of course here, and how like, how do you do it technically? How do you do the report? What tools do you use? 

Simon Kemp  18:07

How do we do it? So, an awful lot of it is manual. It's going to freak people out when I say that across 25,000 data points, I do most of it manually, but it is. So, we go across a whole series of different websites. 

Tom Ollerton  18:20

So, give me a snapshot of these websites. What, what's your top? What's your top five? 

Simon Kemp  18:24

Yeah, so all of the sources are actually listed in the report itself. So, internet world stats collects internet user data for every country in the world will then send to check that across a whole series of other data points as well. So, Sarah, who is sitting behind the camera today, you can't see her. 

Tom Ollerton  18:36

You build out a spreadsheet? 

Simon Kemp  18:38

Oh, yeah, yeah. It's massive, massive Excel spreadsheet. It's huge numbers of tabs and ridiculous amounts of data. But so yeah, Sarah will go into the government site for a country in Kenya to see their telecoms agency to see what they're reporting as the number of active internet users is as well, we'll then go to the CIA factbook, and we'll get some internet user numbers from them. So, especially for Internet users, because it's so complex and time consuming to gather the data for the original data source. We report where we can four separate numbers for the internet user number, because... 

Tom Ollerton  19:12

And then you average that out in the report.

Simon Kemp  19:14

So, we choose the one that we think is most representative, but will then report the other four as well. So, in the report from most countries in the world, you'll see a series of different options. And that is deliberate. You know, a lot of people say why can't you just decide what it is, is that well, I have done this is the one that I would use, but I know that these can be politicized or you may think that they're not representative, whatever, you may just want to have a series of options to choose from. So, yeah, we collect a lot of that data manually, the Facebook stuff, the Facebook data we collect manually. No. 

Tom Ollerton  19:14

Okay. So, for 234.. 

Simon Kemp  19:45

239 countries. 

Tom Ollerton  19:46

239 countries. How many sources per country do you check in order to arrive at the number that you think is most representative?  Right. And so how long does it take you to do this? 

Simon Kemp  19:53

So, there are a couple of data sources that provide a lot of the data so between GlobalWebIndex, Statista, and Google Consumer Barometer, they provide probably three quarters of the total volume of data, that then, in total across the whole report, there are well over 100 separate sources.  Most of, well... 

Tom Ollerton  20:14

Have you got time to do this? 

Simon Kemp  20:16

Fortunately. So, we spend most of the year sort of collecting ideas of where the data is going to be, I know, when I sit down on the first day of doing the report proper, I know where I'm going to get most of that data from. So, an awful lot of it is preparing in advance, you know, you see a bit like a squirrel, you're burying your, your acorn somewhere, you know where to go and find them. But yeah, it doesn't change the fact that you then have to actually extract the data. And an awful lot of it is still manual. And that's great. And I make it sound like I'm a bit moaning about it. But the fact that a lot of these data partners make that data available for free, the fact that I have to download it manually, or copy and paste it in some cases, it's like it's a faff, but 5 million views justifies the, the effort that goes into that.

Tom Ollerton  20:58

And what, we're kind of sat here at the end of that process and this has been successful again which is fantastic. What are the kind of nauseating moments of doubt that you have for the report. 

Simon Kemp  21:11

Every year. So, you know, you go through this massive stress and pain and then you publish it, and you're exhausted. 

Tom Ollerton  21:16

Tell me what's the stress and the pain. What's difficult about it? 

Simon Kemp  21:20

So, there's one company that breaks my patience and frustrates me every single year, they will remain nameless, but they provide you most of the world's most popular business software. It crashes on a regular fucking basis, and I lose a day's worth of hard work. And it's just, it drives me mad. They also look after the hosting solution that we use. And the hosting solution changes the rules every five minutes. So, we spent days preparing the uploads to this hosting platform called SlideShare. And we had everything ready. And then the day that we're going to go and publish, they change the rules on the fact that you're not allowed to change the underlying presentation. So, you can upload a new version of the presentation. So, we spent days getting all this stuff ready. And then they just changed the rules. 

Tom Ollerton  21:59

And you were, you didn't used to be able to do that. 

Simon Kemp  22:01

Yes, you can change. And so... 

Tom Ollerton  22:04

On slideshow, you can only... 

Simon Kemp  22:05

Can't change it, you can upload it once. And that's it. You can't change it. And that's when you were saying what the nauseating pain points. If I spot a typo, because I'm really anal like that. I'm like, can't change it now. You know, all I want to do is just correct a spelling mistake in a source you know, in like, point seven font, but I've noticed that it annoys me, I've got to change it. Cannot. So, that's the nauseatingly frustrating thing. I mean, every year you get the same group of people that come back that say, "I can't find your..." It's like, fuckin' here, just read it. But you do then get a lot of people that ask... 

Tom Ollerton  22:34

Did you go and say that? 

Simon Kemp  22:35

I did. Now, I usually dont... Me? Me? Being nice to people? Sorta. Now, most of the time, I'm incredibly nice to people that sending questions, but I do lose my patience with idiots. And there are idiots and they will send you five rounds of emails, "I've got another question." I'm like, "Just read the same sources at the end of the report. It's there." But you do get some amazing questions. There's also people that are contacting you and saying, "Could you be this, could be..." The funny thing is, we spend a lot... So, Japan has been a place where we spend a lot of time working over the last few months. And interestingly, the amount of time that people spend on the internet in Japan is very low. And I've been saying for the last, you know, last couple of months, we've been touring around the world. And I don't know why it's so low. And the lady came up at the end of the presentation today. And she said, "My aunt has just moved to Japan. And she was telling me that you're not allowed to use your phone on the train." And I said, "Yeah." She was, "Yeah, and people commute for about four hours a day on trains in Japan don't they?" I was like, "Yeah." She goes, "That's probably why they don't spend as much time alone." Oh, my God, that's so obvious. It just hadn't occurred to me to put those two things together. 

Tom Ollerton  23:39

Tell me a bit more about that. You can't use your phone on a train in Japan. 

Simon Kemp  23:43

Well, you can't talk on the phone. You're not really... So, one of the things that drives me absolutely insane about society today. Let's start kicking fag packets into the gutter at this point as well. But kids that get onto public transport and watch their YouTube videos at full fucking volume. Really, at what point did that seemed like a socially acceptable thing to do? What's wrong with you? Yes, well done. You want the world to know that you're doing things. But this is not the way to do it. So, a lot of the time that in Japan just would not happen. You may get the occasional freak that does listen to something aloud. But you know, if they're going to have a phone conversation in anywhere public, they're going to a very private place and whispering into their phones sort of thing. So, they're not going to be watching videos at live volume on the train. If they do, then they've got headphones in. 

Tom Ollerton  24:25

And what other unusual trend booking digital behaviors that report on Earth. 

Simon Kemp  24:31

So, it was interesting this year that the amount of time people spend on social media has gone down slightly. I think a lot of that is that we've got slightly better at using these things. So, we were perhaps doing a little bit less of the just, aimlessly scrolling through feeds and we've gone on there with a bit more purpose to our activity. I don't think there's anything that stood out as being a dramatic kind of, "Oh my God, that's, the world has changed since last year." Mainly incremental change. 

Tom Ollerton  24:55

And so what, what if any, is the new data, either that was new data that was in this report, or that new data that's gonna be in the next three or four or five years? 

Simon Kemp  25:05

We almost doubled the number of data points in this year's report. So, there was a lot of new stuff. So, we got some great data from Localwise, who are a social listening company, they gave us insights into Facebook reach and engagement data for about 180 countries around the world. Just amazing data. Got new stuff from App Annie, so they were great. They shared some stuff with us. 

Tom Ollerton  25:28

And what I mean is, that's great. They have new data sources, but safe things like motion tracking data, or voice data, or smart self driving car data. 

Simon Kemp  25:41

That kind of data, we're not at this stage where we're able to collect it consistently for a sufficient number of countries that will report it. So, especially when it is data like, suppose it's the data that a Fitbit collects, for example, it's not the sort of thing we would include in this. So, it's very much focused on data that will help you make business decisions about using digital. So, should I invest more in promoting my ecommerce platform? Should I build an e-commerce platform? Is that kind of stuff that we'll be helping you to make decisions on? Rather than how many people are counting steps, I would like to have that kind of data. But a lot of the time, it's very difficult to get it across more than just one supplier. So, for example, Fitbit doesn't represent the entire wearables market, much as they're a big player. And to get that data across all those different players would just be close to impossible on a practical basis. 

Tom Ollerton  26:29

So, can you tell me... Without getting any confidentiality problem. This is a terrible, terrible thing. But can you tell me about a brand story where they didn't know about the report? They heard about the report, and they, you approach them and their businesses change? What are the success stories associated with a brief that you could you tell us about? 

Simon Kemp  26:56

Did I mention the Middle Eastern government already? So, they said that the end of today, they said this has changed government policy. I think, if I'm changing government policy with a report, I'm fairly pleased with that result. 

Tom Ollerton  27:07

Let's assume it's for the better. 

Simon Kemp  27:09

I would hope so well, yeah, look, I am fully, I am fully aware that there are people using this data for bad means as well, that doesn't mean that the data should be put away. It's funny that we get a lot of thank yous, but we don't get a lot of specifics. So, it's not like because of your report, I have done X, Y, and Zed, and our company has whatever. I got a lot of nice messages from NGOs. So, there was an NGO that was helping young people. So, teens in Timor-Leste understand sexual health, as a lady contacted me and saying, "I just wanted to let you know that your data has completely changed the way that we approach this. And it saved us a significant amount of money, and we've reached more people with valuable information." So, that kind of stuff, you know, when you're going through the sleepless nights trying to collect all the data, and you're fighting with the famous software vendor. Yeah, it's very satisfying to know that you will be making a difference. And a lot of the time, I am aware that it's just people sitting in middle management jobs, collecting slides to put into their own presentations, and it's not going to change the world. But there are moments when it does. So, yeah, that makes me feel slightly better. 

Tom Ollerton  28:10

That's a great story. And often, often in this industry, you can be sitting there thinking, why are we talking so ardently about big beans? 

Simon Kemp  28:21

I want to sell more. So, yeah, definitely. 

Tom Ollerton  28:23

And, so what's next? So, what, what does this report look like in five years time? 

Simon Kemp  28:27

Funny, we've already been talking about next year's report. So, we're only what? First of March today. So, we're only a month after we published the 2018 Report. We're already talking about 2019 Report. Quite like a few months off before I get involved in making it yet. But I think a big part of the report next year is going to be taking the presentations out. So, like the presentation that we did this morning that you mentioned, where we're actually translating the numbers into trends into actionable activities as well, a lot more I think of next year will be about road showing the data. So, you know that the data is there in the report as a reference when you need it. But in all honesty, like I said, it's not going to, it's there more as a way of getting the basic data that you need, rather than giving you the high level answer to the questions that you care about most. And actually, what it should be doing is asking those questions as well. So. I think a lot more of next year will be about how do we inspire better questions amongst people? 

Tom Ollerton  28:28

Do you have a plan to make the tool live? Is that possible? 

Simon Kemp  29:28

Been thinking about that a lot. There is a degree of technicality to that. But that's not a barrier. It's just a decision to do. The main reason why I've sort of held off on doing that is that people seem to prefer to have something that they can download. And the danger is if you're downloading something at any given point, in time it's going to change. It's very difficult for people to then have consistent conversations with each other. So, I think that there would be a value in doing both. But I think once you get into providing that data on a constantly updated basis, you get into some slight awkward challenges because people like GlobalWebIndex, for example, that provide commercial data to us for free, they can only do that once a year and not really compromise their business. So, things like, I might, I might get to the stage where we publish, sort of the hard, hard core, I don't mean as hardcore. But you know, the central core of the data, which are the internet users, the social media users, the mobile users. So, those are numbers that are updated, that we can get on them on a regular basis, the social media numbers are a bit of a pain together, because you have to do it manually. We could theoretically be doing on a more frequent basis, but actually having that consistent point of reference for everybody does have its advantages. So, I haven't said no yet. But I've been thinking about it. And I haven't convinced myself, it's the right thing to do. 

Tom Ollerton  30:48

Yeah, very different business, probably need some serious investment, and then you'd be managing a team of developers and data scientists. And that's a whole different... 

Simon Kemp  30:56

Yeah. And you probably want to monetize that. So, you get into advertising. And then you just get into it being a business rather than promoting the business that you want to do. So, Kepios is a business that I have sort of crafted into exactly what I want to do. So, public speaking, helping brands that care about crafting something they're committed to doing. And when you start to get into just publishing data for a living, it gets a little bit different. 

Tom Ollerton  31:23

Changing the tone of your voice there, "publishing data."

Simon Kemp  31:27

I have huge amounts of respect for the data partners that provide that data that do that day in and day out? But it's... 

Tom Ollerton  31:32

Just not you. 

Simon Kemp  31:33

It's not me, is Simon. Is not what I want to do day in, day out. 

Tom Ollerton  31:36

And so the part of your presentation that I missed, that was quite frustrating. Was your... 

Simon Kemp  31:41

...real work? 

Tom Ollerton  31:42

Yeah. Your, your vision for the future? 

Simon Kemp  31:44

Yeah. 

Tom Ollerton  31:45

We talked about a bit about the impact of voice on the next billion internet users. Touched on that last night, I thought was fascinating. So, I think people will listen to this podcast will be really interested to hear that view. 

Simon Kemp  31:56

Cool. So, yeah, voice control is one of those things that you've heard a lot about. It's been massively hyped in the 2018 trend reports and whatever else. But we're not talking about voice control in the context of something like an Amazon Echo. I mean, I think that's a great piece of tech. But that's not what's going to change the internet for everybody in the world. So,let me backtrack a little bit. The next billion, I'm sure folks listening to the podcast know what we're talking about. So, people in mainly developing nations that are not yet online, that will be coming online, in the next three to five years, a billion people will connect most likely over that time. So, we had 250 million people connect for the first time last year. So, if you look at that, whenever inevitably, it will slow down slightly. 

Tom Ollerton  32:33

But quick question, what did they do first? 

Simon Kemp  32:37

What did they do first? You know, based on an inevitable each person does something slightly different but one of the main drivers of getting people online is still social media, they want to be able to chat to their friends. So, first thing they do is they sign up to a Facebook. And that's the first thing and then once they've... 

Tom Ollerton  32:52

Specifically or? 

Simon Kemp  32:54

Facebook is still the biggest. Facebook still grew significantly last year, and we were talking about I don't remember the exact number, but I think it was double digit. It may even have been 15% growth year on year in terms of active users. Now, some of that may be duplication. But inevitably, there's still real people significantly, being the big part of that. So, yeah, I think they go online, they set up a Facebook account. And as soon as they're on Facebook, they see content that their friends have shared, and then off, they go from there very quickly, you spend six hours a day on the internet reading about nonsense. So, when I say nonsense, I mean that in the nicest possible way. But you know, the amount of time I spend watching BuzzFeed Tasty videos, for example, that there's no justification for me watching somebody making an ebi avocado burger five times and yet I just I can't stop myself. I must watch this for 20 minutes. 

Tom Ollerton  33:41

So, that's interesting. The first, the first act of people coming onto the internet, for the first time is to, is to connect. 

Simon Kemp  33:48

For many people, yeah. So, they want to be able to chat with friends and family. And social media is a very practical way of doing that. I think it's, if you think about your own loved ones, the fact that you just want to be able to drop them a message on WhatsApp, so even you and I, when we're coordinating and doing this recording, it's like, "I'll be over in 10 minutes." "Oh, I'm sorry, I'm slightly late." And it's it's a very, very utilitarian, practical thing, but it has so much value. You remember before the days of mobile phones? When you're arranging to meet with your friends, and you'd meet them at the Clock Tower at two o'clock, and at 2:30, they still weren't there. And you're like, "How long do I wait to realize whether you're not coming?" Right? So, I think it was just the fundamental changes that basic things like our WhatsApp or a WeChat, or whatever else is introduced into just the way we are. We were talking about this the other day about the fact that kids no longer actually make plans to meet each other. They just say, "Oh, yeah, I'll see you." And then it's like, "Where are you? What are you doing? Let's meet up." It's not like, "Let's catch up on Saturday." Well, not actually make a specific plan with a time to, I, just yeah, I'm obsessive compulsive. I have to have details. I can't deal with it. 

Tom Ollerton  34:48

So, stepping back to the to the voice piece. You've got billions of people coming online over the next three to five years and they're not all gonna buy an Amazon Echo first, which is... Unfortunately for Jeff. 

Simon Kemp  35:03

He's okay with that... 

Tom Ollerton  35:05

And so what? What do you mean by the, we'll be using voice on the internet first. 

Simon Kemp  35:10

So, you've got a few challenges, which have stopped people coming online already. So, those next billion, it's not that they don't want to come online for the most part is that there has been some kind of challenge or barrier that stopped them. So, you've got basic things like technology infrastructure. So, is there even mobile data in my area? Can I afford to buy a phone? Do I have access to electricity, so completely fundamentally basic challenges, you've then got things like literacy, this is one of the biggest issues that I think most people in the West have not got their heads round. When we talk about the next billion, it's not just a case of "Oh, once they've got data, then off we go." It's like in parts of Africa, their literacy rates, even in audiences aged 15, and above, less than 50% of the audience can read and write to a proficient level that they can make sense of written text. The vast majority of content on the internet is still text based. So, yes, you and I watch a lot of YouTube, I'm sure. But if you think about how many news articles you read, your WhatsApp messages are all text, when you think about it. And you know, if you think about it carefully, the vast majority of that content is still text based. So, if you're illiterate, you can't read it, and you can't type stuff in. So, you can't use WhatsApp either from that perspective. So, literacy is the biggest challenge to getting people on the internet. But the good thing is we can, almost all of us can have conversations, even if we can't read and write, we still got language. So, you and I could quite happily have a conversation, even if we weren't able to type stuff down. And the thing is, as soon as I get into a voice control world, I can start to say, "Okay, Google, take me to this website, or I'm looking for the nearest shop around me that sells potatoes," whatever it may be. You can speak that into a device, and it will take you to the website, it will then read that website back to you. So, you've got these great text to voice converters. So, when we're talking about voice control, it's not clever, snazzy shit of add chips to my car, Alexa, it's much more I can now access content, and I can read it. And so how do you think that content is going to be served? Because you've, you've applied an old model to a new paradigm, right? Which is, "Hey, Google, tell me where the local shop is that sells potatoes," or whatever it is. And then you made the assumption, rightly or wrongly that, this voice tool is going to read back a website, but in actual fact, it probably won't work like that. Probably much more based around...  Probably sayings, right? So, you obviously walk down the road, 100 meters and turn left, and you'll be there. It will be a lot more practical than I think, it's... 

Tom Ollerton  37:42

What I mean is, is it going to be a, is it going to be a Google service, or a Facebook service? Or an Amazon service? Or whoever it is? Reading back websites? Or will, in order for a voice service to work, will it need a much more recognizable interface, which is just a person like a character and an assistant? Because if the voice assistant said something like, "I'm going to read you from the website, XYZ," I think, what's a website? 

Simon Kemp  38:10

Yeah. 

Tom Ollerton  38:11

So, we'll there, will then need to be some form of character or assistant or an Alexa or Siri or Google Assistant to perform that function? 

Simon Kemp  38:23

Doesn't need to be. There's a big business opportunity in creating one. So, if you look at Siri, Siri isn't just about functionality. There's a massive, massive revenue opportunity for Apple Siri. But I don't think that it's a necessity to make it work. So, I think you will probably find that the phone device that you have will have an inbuilt function in the same way that it's got ways for you to just type stuff into a browser, I think you'll probably find that there's a basic voice tool that's in there, it's still going to be brand centric. So, it'll still be an Apple tool versus a Samsung tool versus whatever else, you'll still see the Android versus iOS dichotomy in there. But I don't think it's to the level that I think you're alluding to of, will this be a completely different platform? I think in the short term, at least, we will probably see voices just being a way of accessing existing content, it will evolve over that. The sensible content producers will realize that for the voice world, they can do things very differently. You don't need to make a website if you've got voice. 

Tom Ollerton  39:20

They had an interesting story the other day. Have you ever used Motoread? 

Simon Kemp  39:24

No. 

Tom Ollerton  39:25

It's an, it's an app and Chrome plugin where if you find, if someone sends me an article during the day, I can't read it. Because my brain is just not in like, sit down, spend 25 minutes or so, we can't do it. 

Simon Kemp  39:35

Yeah. 

Tom Ollerton  39:36

So, what you do is you click on the Motoread icon and it reads that page and dumps it all into your app as an audio file. So, then you can listen back to it. And so what I'll do view some in article, Simon I'm gonna read that now. Click on the article, goes into the apps, and the following morning, on the tube on the way in, you can listen to it, right? And you can, but then what you can do, this is where it starts to get quite weird, is you can turn up the speed and you train your brain to listen to articles at three times a speed. 

Simon Kemp  40:06

It's like listening to me talking normal, isn't it? Yeah. 

Tom Ollerton  40:11

And so, telling someone this story and they said, apparently, blind people read the internet at five times speed. So, when they're looking at a website, they'll just suck down out information, five times speed and everything else seems like really laboriously slowly. So, if that, if you extrapolate out from there, then this next billion people are going to listen to the internet... 

Simon Kemp  40:35

At five times speed. 

Tom Ollerton  40:36

Which means they can get through information five times quicker, which means that potentially five times more efficient and effective, so all this next billion people can be operating on the internet five times more effectively than we are. 

Simon Kemp  40:48

I think we'll, hopefully all of us will get to this stage where we're more efficient, but in the same way that a lot of these people in developing countries leapfrogged from nothing to mobile devices without going through desktops, that you'll probably find that they are much quicker to adapt to those new technologies. So, voice search has been around for a long time. It's not like it's a new development. And you and I still don't use it as much as we should. Whereas these guys are going to jump straight into using voice search. And it's going to be their primary, even if you go to India now, when people are perfectly literate, they're still using voice search a lot better, because it's a bit of a hassle to type in Hindi characters compared to English, whatever else, it's just actually, it's an awful lot more straightforward to ask. You know, just just hold it up and go, "Okay, Google." 

Tom Ollerton  41:27

So, how would the, so how are the social network giants gonna deal with a voice-driven internet? So, if you've got an, a billion people who don't have a screen, they just have a speaker and a microphone, and they're talking to the internet. And yes, they're going to call their friends, you can understand how a service like WhatsApp might work there or, or a WeChat. But then, you know, these these are services or Skype, they're predominantly voice systems. What, what's Instagram going to do for a billion people when they have only got a voice and microphone? Are they going to say you're looking at a square picture? Of a plate of food in Hoxton. 

Simon Kemp  42:05

Screens are going to disappear anytime soon. But the amount of time we spend on screens as part of our digital experience will drop quite considerably. But I'm not going to stop watching movies just because there's a voice world. So much, as much as I love podcasts. And the fact that we're doing one as a certain irony, I'm still going to want to watch epic content, I'm still going to want to watch videos, I'm still going to even watch videos without the sound on. So, this just provides you with more options, especially when it comes down to an efficiency and convenience perspective. So, if I want direction, so we were talking about the fact that it's been snowing crazily in London today, I don't want to take my phone out and take my gloves off to try and open the map to find my way to Tom's office. I just wanted to speak to me in my ear and say, "Turn left here." 

Tom Ollerton  42:42

So, I get that but that, you switched the conversation into like, a Western viewpoint. I'm just focusing on this next billion people. 

Simon Kemp  42:50

Okay. 

Tom Ollerton  42:50

Because it's gonna be easy. It's gonna be cheaper to serve the billion people with an Internet with a microphone and the speaker. 

Simon Kemp  42:55

Yeah.

Tom Ollerton  42:56

They used to have a microphone, a speaker, and the screen. 

Simon Kemp  42:57

Yeah. Correct. I think it's unlikely that you would, I don't think the majority people will have a device that doesn't include the screen, it will be an option, but I think you'll probably find that most of them want to watch videos. So, Facebook, very clear that video is the future for both Messenger as well as the vanilla platform. WhatsApp, if you open up WhatsApp now, ironically, it's a text based platform, but the camera button is at the center of the interface. It's just that we're not using it that way. Yeah, but what's that will look very similar to Skype very soon, it will just be you and I are having video chats. So, I think even for people in that part of the world, and if you look at the way that they're using this, now, they are kind of doing that they're having video chats more frequently. Because you know, they if they can't tie messages, then they are going somewhere like China, for example, you look at WeChat. The way that people use WeChat is that they don't type anything anymore. They speak. So, it's audio based. Ironically, they still have a screen that they tap to then trigger that, but it's listening and doing this kind of weird stuff. 

Tom Ollerton  43:52

So, the the future of the internet for these next billion people will be massively video centric. Voice driven. 

Simon Kemp  44:00

Correct. 

Tom Ollerton  44:01

So, looking forward again, what are the trends and technologies that are making you feel a bit confused and weird? 

Simon Kemp  44:09

Confused? 

Tom Ollerton  44:10

What challenges you? And you think, "I don't understand this. Why is this happening?" 

Simon Kemp  44:13

It's more people's behaviors that I don't understand. Why do people still take photos of their lunch every day and upload them to Instagram? Or I'm sorry, if I've missed the video, I don't think there's anything confused. There's things that I look at, I get very excited about that. I don't know as much as I would like about. So, blockchain, it's a totally overhyped thing. But I don't know nearly enough about how the technology itself works. I can see all the benefits of it. And I can see how it's going to change the world. But it's not like I feel I'm a massive expert in these kinds of things. Artificial Intelligence, one of our favorite stories. I kind of look at that. And I think, again, there's an awful lot of hype that goes around it. But there are elements of it that I look at and think, "No, this this really is going to change the world very quickly, in ways that we can't even imagine." So, we were speaking yesterday as part of a separate conversation about creativity and artificial intelligence. And I sort of look at that. And I think this is staggering, because you've got artificial intelligence, which an awful lot of the time at the moment is logic based. But when you start to get out of that into more imaginative, emotional territory. It starts to get really, really interesting, because a lot of creativity is, it's very much about that right brain activity in a human. And the funny thing is, you can do right brain thinking for devices as well. It's about experience, it's about understanding, what stimulates a certain kind of response. And a lot of this comes down to the idea of machine empathy. I know that an awful lot people are going to be listening to this that, "What? How can a machine be empathetic." But the weirdest thing about empathy is that it varies by culture, which suggests that it's totally learned. And anything that's learned, you can encode and if you can encode it, it becomes an algorithm, and therefore, you can feed it into some kind of artificial intelligence system as well, technically, if you ask me to write a code on that, I have no idea where to start. But, you know, I started thinking about the applications of that, once it's possible, it becomes absolutely fascinating, because if you've got a machine that is genuinely empathetic to other machines, you remember, did you ever watch "WarGames," the movie with Matthew Broderick back in the day, yes, back in the day. So for those viewers and listeners who have not watched this movie, basically, I'm gonna have to give you a spoiler to ruin the story. But nonetheless, you've got these war, war simulators that basically end up almost triggering a nuclear war. But just before they actually launched the real missiles, they realized that nuclear war is pointless, because they've run through every possible scenario. And every possible scenario is that everybody dies, an empathetic system is going to come to that conclusion incredibly quickly, it's going to run through all of this, I don't understand the benefit of annihilating everybody. And the funny thing is that an artificially empathetic system can be a lot more rational, if it chooses to be or it can dial up the level of emotion to feed into that if it wants to. So, it becomes a really sort of, I can see you've got lots of questions. So,I'll stop. 

Tom Ollerton  46:28

No, just tell me more about how you think that a can actually be empathetic. I can understand that it can, it can synthesize different responses, depending on how people have responded emotionally, but it's not actually going to empathize, there will be no empathy, it will be creating the illusion of empathy. 

Simon Kemp  47:35

Yeah, the illusion of, but if anything is like that empathy has two elements to it, they're the most sort of common definition of empathy is the ability to identify, understand, and sort of internalize other people's feelings. So,I can see that you're sad. And I relate to that, because I've been sad myself, and therefore I know what you're going through. But the important bit about empathy, in all honesty, the fact that I've identified and understood that doesn't mean anything to you. It's only when I change my behavior with you that the benefits of empathy come through, knowing that you feel what I feel, and vice versa is important. But it's, it's what that then does for our relationship. And you can see how systems can empathize with us by change... 

Tom Ollerton  48:14

...behave empathetically, they can't empathize. 

Simon Kemp  48:17

No. 

Tom Ollerton  48:18

I was reading this interesting article the other day, and it was in WIRED News, basically saying that Google have won advertising without understanding it. So, they said this website is popular, because all these people have linked to it, they know what's on it, don't really care, buy some ads. So, they're, they're the most powerful, one of the most powerful forces in advertising without having any, any understanding of it. So, I think it's conceivable that machines will be able to do a similar job with emotion, yeah, but they will never feel anything.  They need to. I still get the benefit of... 

Simon Kemp  48:51

But there's still no empathy, if they don't feel. But it is because if it's changing the way that it treats me, based on my emotional state, and my emotional reactions, then that is the benefit of empathy. So, the way that you react to me when I'm emotional versus the way the machine reacts to me, when I'm emotional, I get the same benefit from both of those things. What's the difference? I know, instinctively that it's different. But this, this is a little bit like AI and art, right? So, if you look at any kind of AI creative output at the moment, you give it to a critic, and they're like, oh, it lacks soul and blah, blah, blah. If you don't tell them that it's created by a machine, most of them have no idea. And they just analyze it like any other music or piece of art or whatever else it may be. A lot of this is our reaction to the fact that it's a robot. We're not empathizing with the machines. And I know that sounds ridiculous, but it will come to the stage where we need to empathize with them as much as they need to empathize with us because we rely on them for so much stuff. 

Tom Ollerton  49:42

So, yeah, got me, my favorite favorite subject here. Really, my whole view about creativity and machines is that a lot of creativity is down to context. So, there's a Professor of Computational Creativity. A guy called Simon Colton, presented it, a little while ago. And he talks about a poem and he reads out a poem, however long this poem is. And he said that, you know, that poem was written by into a poet, whatever. And then he says, right, well, actually, that poem was written by a woman, not a man, and, then puts a slightly different change on it. And actually, that read, that poem was written by a child, and once again, like, "What, really?" And then he goes, actually, that poems written by a pedophile, and then your brain spit, and then eventually says, that poem was written by a computer. And the only thing that's changed is you, is the context, that the thing itself has, has stayed the same. And I think that, in that, I think I like the argument that a computer can't create art, because so much of the art is the context, right? 

Simon Kemp  50:50

So, I'm going totally flip that on its head. 

Tom Ollerton  50:52

...that it can take what, and what a computer can do is create content, it can create media, but an art depends on the context. So,look at and you know, the Duchamp's toilet bowl where he's, you know, write a name on it, right? It's all context. It's, it's, it's conceptual. So, if even if a machine only seen IBM do it, then they rewrote like a Beatle song, something, but it didn't really. They sort of came up with a chord progression and someone else wrote a song. Let's, let's assume, for the, for the moment that they did actually create that thing. That if, it was completely nothing like a Beatle song because the context of the Beatles doing it was completely removed. It just it was media that was representative of 105 tracks that they released. 

Simon Kemp  51:39

But you've got things like the next Rembrandt, so they analyzed 30, 40 different Rembrandt paintings and how he structured those and his choice of colors and poses and everything. And they created a brand new painting. 

Tom Ollerton  51:50

But the context was the case study and the technology. 

Simon Kemp  51:54

Correct. 

Tom Ollerton  51:54

The actual, the thing in itself was unremarkable and fundamentally useless and unsellable. Because it's not a Rembrandt. It's It has nothing to do with Rembrandt. But then what makes it interesting is the story. 

Simon Kemp  52:07

Yeah. But it's exactly that. Everything in art is a story that we tell ourselves as the, so you've got beauty in the eye of the beholder, but value is in the heart of the beholder. So, you look at, if you look Banksy, I mean, I love Banksy's work, right, but it's a stencil that he's taken, and he sprayed it onto a wall. What makes that better than somebody who scribbled their name on the wall? It's totally the story that we're telling ourselves. When you get to a Jackson Pollock piece of art, right? I mean, I really like his work. But seriously, I've taken squiggles of paint, and that's worth millions of dollars. And yet a kid at nursery school could do exactly the same thing. It's not worth million dollars because why? It's the why, the why is the story. 

Tom Ollerton  52:47

Context. Because the kid doing it is the context is well that's a child messing around with paint. The context of Jackson Pollock doing is a completely different thing, because he was the first or one of the first to innovate in that space. So, that's, so that's where I, part of my role is to understand the relationship between artificial intelligence and art. So, brands can make use of use of that. And I think we're we're lucky as advertisers and brands and agencies is because we control the context. So, if we go right, we're going to show a rectangle to a person on a mobile onibus in Chelmsford, on a Saturday, as long as that results in the desired action, then the context is justified. 

Simon Kemp  53:32

Yes. 

Tom Ollerton  53:33

For a machine to create art, without any context, other than it being a machine, I think is it's very difficult thing to get around. 

Simon Kemp  53:40

But it's what you said, it's the story that you tell that makes that difference between content and art, right? And most of that... 

Tom Ollerton  53:47

Wow, we're getting deep. 

Simon Kemp  53:49

We are. But the really interesting thing is that an awful lot, so an awful lot of that story is about empathy. It's understanding your emotional reaction to it. So, music is all about emotional reaction. Art is all about emotional reaction. 

Tom Ollerton  54:02

Gonna have to pull you up on that. I think music is is all about context. 

Simon Kemp  54:07

Yeah. But it's an emotional reaction to If I hear a sad song on a happy day, versus if I hear the same song on a sad day, I have a very different response to it, I then have an experience and the number of times I hear it, it changes what it means to me as well. It's it's a very, it's very subjective response, right? You and I will respond to exactly the same song slightly differently or dramatically differently. So, you like Indie, I love Techno. If I play do some of my favorite Techno tracks, you be like, "This is just noise." And yet for me, it genuinely does touch emotions that nothing else can reach. And that's purely context. And it's a very subjective and emotional thing. And it's is actually the same bits of, sort of hormone and brain biological processing as empathy. It's the way that I internalize an external stimulus, and then respond to it. So, I think we're in violent agreement here. But what's interesting is that there is nothing to stop a machine that becomes empathetic being able to take something that is merely content and tell a story around it that makes it into art in the same way that Jackson Pollock or whoever else it may be has done. So, there's there's no reason why a machine wouldn't be able to... See, if you've watched "Her," the movie, he falls in love with his personal assistant, where we've got to the stage where we have that level of relationship, if that personal assistant does start telling us about the fact that they absolutely love this new font that they've seen. And you have such a relationship with them that they influence the way that you perceive that font, you can very easily see how these things progress. Now, whether it whether the machine loves the font or not is a different question, because it doesn't feel emotions in the way that we do. But if it knows that, I like these things... 

Tom Ollerton  55:40

What you've done with that example is you've created a context within which the creativity will be valuable to you. So, you by responding to the "Her" AI will tell, you will tell that AI a lot about yourself, and it will learn about you. And so you will justify to yourself that AI is, is useful for you. Som anything that AI says, by a green jumper, then the context is thatm that choice is the right one. But it's all it's ever really do is reflecting your own choice. So, in actual fact, you're making the choice. 

Simon Kemp  56:18

Yeah, but then... 

Tom Ollerton  56:19

And that AI would be a kind of proxy version of your own choices anyway. 

Simon Kemp  56:24

It may not be quite as filter bubbles, that there's it may go out and look at lots of other people as well. So, it may know that I have an incredible like for the style of Ryan Gosling. And it analyzes his thing, and it recommends to me... 

Tom Ollerton  56:34

The style of Ryan Gosling...

Simon Kemp  56:37

But this happens in real life, right? So, if I think about my fashion... 

Tom Ollerton  56:42

The light's going off, by the way... 

Simon Kemp  56:45

Sorry, folks, we've run out of battery on the line. I think about my fashion choices since I got married, my wife has had a significant, my wife has had a significant influence on choices that I make from a fashion perspective, because she tells me things that I cared about the way she feels about it. If you think about that, it's very easy to see how you'd move into a situation where we depend on the advice of a personal artificial intelligence system. And that then influences our choices. Now it might be out there, it might be scanning hundreds of thousands of millions of images across the internet, to help me identify images or styles that I want to achieve and then recommending based on that. So, it's not just that it learns what I like, it learns what I like in a broader context, it looks at what my friends are sharing. And ideally, in the best of all possible worlds. All assistants are talking to other assistants. And you end up in this perfect understanding of a... 

Tom Ollerton  57:35

Huge, huge gulf missing there, because what, it's a good way to describe it is a joke that Alex always tells at our event. He says, "A machine learning algorithm walks into the bar. And the barman says, 'What are you having?' And he says, 'What's everyone else having?' Which I think is a really nice way saying like, you can only choose based on what are their, like internal or external data that it can get its hands on, but what a machine can't do. And I haven't even seen one instance of this is making a conceptual link between two unrelated things. So, I mean, is it Move 37 and AlphaGo? In like the, in the first game where it was, I don't know anything about Go but watch the documentary on Netflix. And it basically remove this, this piece on the board. And I was like, "Wow, my god like that's, that's, that's an incredible move." That was created. And I obviously, I wouldn't get in the door at DeepMind, let alone get a job there. But that wasn't great. If it was just based on previous data, what it wasn't doing was making a contextual link between two things that weren't related. And that is it, the core of what creativity is, is the ability to link two separate ideas and think, "Right, okay, that now becomes my idea whether it's a joke, or a melody, or a song or rhythm or an advertising campaign or a strap line," you need to join two previously unrelated things. And that... 

Simon Kemp  59:05

Imagination, right? 

Tom Ollerton  59:05

Yeah. 

Simon Kemp  59:06

Innovation, intentiveness. 

Tom Ollerton  59:08

And a machine can't do that. 

Simon Kemp  59:09

I think, I'm not gonna disagree with that. But I think we are giving ourselves way too much credit as humans, tell me the number of human beings that you know, that walk into a bar and don't choose based on what other people are having. Every time, you go into the bar downstairs now and you watch how many of them gone in with a totally original thought of, I want blackberry juice mixed with whiskey, please because I'm an innovative thinker, no bullshit, we go in and we follow what marketing has told us what our friends told us what our experience and our heritage told us. And even when you look at the ways that we recombine two completely unrelated things, that is a totally cultural aspect. If you look at the way that people in the Middle East versus the way the people in the Far East do it, it's totally different, because there is such a strong degree of experience and cultural influence on that. So, I'm not saying that we don't do it, but I think we've missed the point that it's not like it is 100% isolated and separate when we come up with these ideas. If you look at a child learning how to put a look together, I'm gonna back to my Ryan Gosling style here. If you think about children, and the way that they learn to put things together, they sometimes put the most ridiculous items together and their parents will they go, "You're not going out of the house looking like that. It's interesting, but no." And they learn over time that certain combinations are better than others. And whether it's fashion, whether it's music, or a child in front of a piano, but it's like making noise, I'm happy, but a dog could do the same thing. And over time, we learn what other people like, what we like, and there are the genuinely incredibly creative, innovative people that come out of nowhere was a completely unusual science. I still remember the first time I heard Bjork, for example, and just going, "What the fuck? This is so different to what I'm used to." And yet, listen, listen back to it now. And it's like, because she's influenced so many things later, you kinda go, "Yeah, it wasn't that original really, was it?" No, it was, it's just that it then influenced everybody else. And her way of being original, influenced other people being original as well. And I think if you listen to, if you speak to Bjork, and you ask her, what influenced her, she will say, actually, all these things influenced my originality and creativity as well. There is no such thing as a totally unique original thought. It's always two things that have existed already, you just put them together in a slightly new way. 

Tom Ollerton  1:01:20

I'm not arguing that. I think that's, that's really succinct way of putting it. But a machine can't do that. 

Simon Kemp  1:01:26

But it can, it's just that we don't see the value of it. I think this is where we're gonna get to really interesting state in creativity for AI is we, as the human observer, judge all of those outputs based on our collective experience on what we think it should look like, based on our needs. But what happens when an AI critic judges an AI piece of output? Because let's face it, a critic is just another human being that is subjective. And sure, they may have experience with judging thousands of bits of art. But when somebody says, "Jackson Pollock is the next big thing." That critic has a massive influence on it, but it's still just fucking squiggles of paint with no disrespect to Jackson Pollock. Somebody somewhere decided that this was our versus some kid that spilled paint on a piece of canvas. And I think when you get to the stage where artificial intelligence judges other artificial intelligence and decides, "Is this delivering the outcome that the original brief was there to deliver?" It doesn't matter whether I like it as a human being, if it's a better way of delivering the outcome I wanted, I just have to accept that that's a better way of doing things. 

Tom Ollerton  1:02:27

So, what's the impact that's gonna have on social. So, if you... 

Simon Kemp  1:02:31

You came back to social... 

Tom Ollerton  1:02:33

It was Selena's point. I think it's really interesting that if you've got AIs that can produce acceptable copy, acceptable video, acceptable images, music and whatever. And then you've got the whole social listening piece of the internet. So, you can have AIs, listening to AIs, you can have... 

Simon Kemp  1:02:54

Totally. 

Tom Ollerton  1:02:55

You already, you've got AI influences. I mean, pretty niche at the minute, but at some point, you're gonna have, you're gonna have digital celebrities that have never existed that will... 

Simon Kemp  1:03:07

Like Tay. 

Tom Ollerton  1:03:11

What we do in.... 

Simon Kemp  1:03:14

Still do driving, driverless cars. We've done the next Rembrandt as well. 

Tom Ollerton  1:03:18

What? Yeah, bonus point there. So how, so? And to tie it back into your, into your report? 

Simon Kemp  1:03:25

Tenuous, but okay. Bring it back on. 

Tom Ollerton  1:03:27

So, you're providing a report that tracks the activity of digital behavior in 239 markets around the world? So, how are you and other people in the research and inside profession going to be able to tell what is a robot? And what isn't? 

Simon Kemp  1:03:47

I don't think that matters as much as we want it to. I think what matters is the outcome that it delivers. If it's an AI influencer, that changes my behavior versus a human influence, does it matter? The result? So, I'm going to be quite idealistic about black and white... 

Tom Ollerton  1:04:01

Going back to your point about the server farms. So, presumably, those, those will get, be made obsolete by AIs that can just like whether it's a robot, it's up in front of 1000 little phones, like clicking 'like' on all these different images. So, that's gonna, that will, that system will get automated as well, presumably, 

Simon Kemp  1:04:17

I think it is already. Yeah. But I think when you get to a situation where we become a lot more effective and efficient at influencing whether it's influencing me directly or influencing my personal artificial intelligence system assistant, that then influences my behavior, I think we're going to get to a stage is probably a decade at least off so anybody that says anything beyond four years as a prediction means that it's probably more magic than science. But I think you are in a situation now where I'm not convinced that a lot of these distinctions matter as much as we want them to as human beings, but it's like watching "The Matrix" and saying, "Oh, we're gonna fight to be the vast majority of human beings who would prefer to be sitting in that comfortable little test tube, just existing and pretending that are having a happy life," because we pretend we have a happy life when we're miserable at the moment, most of us. So, I think that whole idea of, "Does it matter whether a machine has changed my perspective versus a human being?" 

Tom Ollerton  1:05:14

So, that's an interesting point. So, I get that, that yeah. Personally, I don't really care as long as the output is good. 

Simon Kemp  1:05:22

As long as I'm happy, and nobody else has suffered. 

Tom Ollerton  1:05:24

Yeah, that's fine. But in terms of, you know, reporting on human behavior, like it, is it going to put the research and insight profession into a bit down? Because we're not gonna be able to work out whether we already said it today that there's a number of 18 year olds that don't actually exist? 

Simon Kemp  1:05:43

Yeah. 

Tom Ollerton  1:05:43

Will we get to a point where an, an AI can create a fiction far quicker than we can spot it? 

Simon Kemp  1:05:49

Yeah. And I think that's the point well, as an industry research, and insight needs to realize that there's a difference between measuring the means, and measuring the end, and far too much of the time, we're measuring intermediate steps. So, measuring the number of likes on a Facebook page has diddly squat value, most of the time. It's did, somebody's behavior change. So, when we're looking at measuring what matters, we need to measure changes in behavior, it probably measure changes in perceptions and attitudes as well. But ultimately, what influences those, you work backwards. So, was it a machine versus was it a human? You then have that, you know, bit existential question of does it matter, you might want to know which one has a greater impact, but I think it's much more that research and insight then needs to focus on the outcome measurement, and not the intermediate step measurement, we're gonna have to get a lot better understanding what real value looks like, instead of just having metrics that we can count and therefore thinking because we can count them, they count, they don't. The number of likes I have on my Facebook page really doesn't matter that much until it translates into, in value. 

Tom Ollerton  1:06:48

So, really wrap up here. 

Simon Kemp  1:06:51

We could go on all day. 

Tom Ollerton  1:06:53

And so, just to look back with talks about the report where it came from. Some of the interesting story, some of the actual positive benefits that had directly on them, sometimes policy, and sometimes education in different markets. It's brilliant. And then we got massively sidetracked into computational creativity, which is very much close to my home. So, I really appreciate your time. That was fantastic. So... 

Simon Kemp  1:07:17

Thank you for having me. 

Tom Ollerton  1:07:17

Let's do this again next year. 

Simon Kemp  1:07:19

Yes. 

Tom Ollerton  1:07:20

And see, see what's changed. 

Simon Kemp  1:07:23

And we'll have to data analyze the evolution over the year and see whether machines have changed the way we think. 

Tom Ollerton  1:07:29

So, just to finish off, if you had to describe the next year in Global Digital, in one word, what would it be? 

Simon Kemp  1:07:38

Growth. 

Tom Ollerton  1:07:40

Let's see with that. Thanks. 

Simon Kemp  1:07:42

Thanks, Tom.

Subscribe to our Newsletter

Watch ‘Advertisers Watching Ads’

Check out our Blog

See videos from our Events

Get in touch with Automated Creative