Dan: Welcome to today’s Boss To Boss Podcast. In our interviews, we feature remarkable sales and marketing minds doing imaginative things in often unimaginative markets, usually from the world of B2B.
Kerry Harrison is the Co-Founder of the award-winning Tiny Giant, an agency launched in 2018 to help forward-thinking businesses harness the creative power of AI and other emerging technologies.
Kerry is joining us today to share her thoughts on what I think are some of the most interesting aspects of marketing and business today – at least I’m assuming they’re her thoughts and not those of a machine she’s pre-programmed this morning.
On the agenda, we have as follows, the reason why confusion still seemingly reigns on matters concerning AI and machine learning in business, how Kerry and her colleagues are breaking down those perceptual barriers, one mojito at a time, why digital transformation projects should bring with a creative session that demonstrates what AI is really all about and why humans and machines both need each other if they’re ever to achieve their full potential.
Kerry, thank you so much for joining us.
Transcript
Dan: So just to kick off then, many people have no real grasp – in my opinion – of what AI and machine learning really even mean. On the other hand, you have lots of people who are seemingly terrified of it, fearing that it is going to steal their jobs and remove any sense of meaning from their lives. Why do you feel there’s so much confusion surrounding these topics?
Kerry: I think one of the key reasons that people get confused or afraid is probably because of the way that we’ve been presented with AI through our lives. You know, a lot of that comes from movies and books and things like The Terminator, which always feel really quite scary.
I think one of the key reasons that people get confused or afraid is probably because of the way that we’ve been presented with AI through our lives.
I think also – and one of my bugbears, I guess – is the idea that the media do like a bit of scaremongering. I think there’s often those headlines about AI coming to take our jobs or AI being better than humans in certain roles. I remember, I think it was last year or maybe slightly before that, where GPT-3 the open AI text generation platform, which is a huge AI platform, came out, and the Guardian wrote an article and the headline on it was: “A Robot Wrote This Entire Article, Are You Scared Yet, Human?” and I thought it’s interesting because we see those kinds of headlines a lot because it grabs our attention and it gets us reading and it gets us a bit scared, we’re more likely to share it and get involved in that kind of article. So that doesn’t help either.
We’re also not necessarily educated on AI. I mean I’ve obviously gone through the school system and no one has ever taught me about AI or machine learning, and obviously, it’s a new and developing thing, but hopefully it’s on the school curriculum at some point now. But we’ve not really talked about it, so everything that I’ve learned in the last 3 years around AI, since I’ve been involved in space, has very much been because I’ve wanted to go out and learn about it. So I’ve read books and articles and I’ve been to events and I’ve spoken to people and I’ve got involved in the whole process and with the developers and that’s the only way that I’ve learned to know about AI and what it all means and what machine learning is. And I suppose unless you actually take that initiative then really, we can’t get clear answers. And I think that’s one of the things that I’m quite passionate about is getting out and educating and teaching people about AI and machine learning because I think it’s quite important that we eradicate some of that confusion, because as long as people are confused or afraid, we don’t necessarily engage in the necessary conversations around AI, machine learning and and what it it can do to our society. So for me, that’s quite an important thing.
I think it’s quite important that we eradicate some of that confusion, because as long as people are confused or afraid, we don’t necessarily engage in the necessary conversations around AI, machine learning and and what it it can do to our society.
So I do lots of talks and I go and deliver workshops wherever I can, because at the moment, AI and machine learning feels like it’s very much owned by those big tech giants and the average Joe doesn’t necessarily know a lot about it and if we don’t know a lot, we can’t question or hold people accountable.
But yeah, there’s lots of confusion because I think it is quite a new thing and it’s always evolving and all those headlines and The Terminator robots don’t really help our case very much. So, yeah.
Dan: So carrying on from that then, it strikes me that some of the work that you do – and you’ve just alluded to it there – could act as a really powerful mechanism for overcoming some of those perceptual barriers, particularly within I suspect large organisations.
So I just wonder, would you be able to give some examples of the kinds of AI and machine learning activities that you deliver and the way that they can bring to life those otherwise abstract and mysterious concepts?
Kerry: Yeah, sure. So and as I said over the last 3 years I’ve been working in the world of AI and creativity. That’s my sort of sector as it were. I mean, AI is being used for so many things in so many different sectors but I’m really particularly interested in the way that it can be used to augment human creativity and how we can use it as a starting point for our creative thinking.
I’m really particularly interested in the way that it can be used to augment human creativity and how we can use it as a starting point for our creative thinking.
I’ll go over maybe 3 projects that we did at Tiny Giant. So, I Co-Founded a company called Tiny Giant and AI creativity’s been a really big part of that business. One of the first projects that we did back in 2018 was to create the world’s first Ai curator for a science festival. So it was Cheltenham Science Festival in the UK, and every year they have human curators who pull together some content for the festival and the coordinator of that festival messaged us and just said: “I’ve had this crazy idea and I feel like you guys might be able to help me – I’d really like to have an AI curator this year who would sit alongside the human curators for the festival.” This was like a dream project for us.
So one of the things the curators do each year is to come up with talk suggestions for the festival, so we needed to use an AI algorithm to help come up with some talk suggestions. So in terms of what we did, we basically got every single talk that’d ever been delivered at the Cheltenham Science Festival and myself and my co-founder went through every every brochure from those ten years and stripped out all of the talk titles and a little detail about about the talk and we used that and we trained a neural network, so a form of machine learning algorithm, on all of those talks so it was able to generate its own talks. And so a range of talks were generated for the festival which were very weird and wonderful. And we also had to come up with a look for our algorithm because obviously a bunch of code doesn’t really sit very nicely in the front of a brochure alongside 2 human curators. So we created this kind of modern day version of Ada Lovelace and we called her Ada, obviously after Ada Lovelace.
And one of those machine learning talks was actually delivered at the festival. And there were lots of talks generated – I mean hundreds of talks. And they couldn’t decide which one to go with so we put it out on Twitter and one was chosen, and it was delivered by a natural history presenter at the festival. So it was a talk generated by AI but delivered by a human and then I sat on a panel at that event as well and we talked about how AI was being used in creativity and also how it was infiltrating into our culture. We even took Ada onto BBC radio on the morning show and gave her an AI generated voice so she could answer questions, and you know, it was a great project.
So that was our sort of first project in that area and then alongside that, the 2 other projects I’ve really loved as part of our business are both alcohol related. So not really sure what that says about me. But the first one is AI cocktails, and this was a project that we did for fun really just because we were really interested in how AI could support or help someone who was very creative. So how it could work alongside a really great human creative mind. And so again, we trained a neural network on hundreds and hundreds of cocktail recipes, and then the neural network was able to generate new cocktail recipes and they were pretty cool. But rather than just going, “Yeah, these are cool”, we took them to a mixologist and said: “Could you bring some of them to life for us?” And the amazing thing about the human mixologist is he’s got an amazing palate. He understands what flavours go together and so we just chose 3 different cocktails which he made up and turned into a reality which is a great project. We got to try lots of different cocktails.
It was interesting because that was just a fun project to see how it would inspire him to think of new ways of working or you know whether it could take his thinking in new directions. But the cocktails got quite a lot of coverage interestingly and people seem to really love them and we’ve since served them up at tech events all over the world. I think one of the reasons, and you were saying earlier about that sort of bringing it to life a bit, is that it does bring it to life because AI is something that’s not very tangible and to actually hold a cocktail that machine learning has been a part of is a great conversation starter.
The AI cocktails got quite a lot of coverage and people seem to really love them… I think one of the reasons is that it does bring it to life because AI is something that’s not very tangible and to actually hold a cocktail that machine learning has been a part of is a great conversation starter.
We did a similar thing with the gin as well. We created the world’s first AI gin. Same kind of process, we worked with a little distillery in Bristol and they brought it to life. We also used AI to name it as well. So we trained a neural network on hundreds of gin names, we also put some beer names in there and then I also threw in some star constellations and some mythical creatures to see what would come out and yeah, again hundreds of names generated but they chose one called Monker’s Garkel, which sounded slightly gin-like but also slightly offbeat.
So the AI-generated gin went into the world and sold out. We did a run of a thousand and it was very cool and very nice. It actually won taste awards and all sorts, so we were quite pleased about that. But yeah, that’s the kind of stuff I do.
I think having something tangible – whether it’s a cocktail or making some art or drinking gin or whatever – it might be those things that are a really great way of making things feel a bit more real and understanding that process.
AI and machine learning are also brilliant for artmaking, so you can create some really wonderful pieces and go through the whole process of machine learning, creating a data set, training your network and then generating an output. I found this to be a really great vehicle for opening conversations around AI. I did a session a couple of weeks ago at a university and some great stuff came out of it, but actually, it was the conversations around it that were really important, just the questions people were asking. Do you know how important it is that we understand it? Or does it just come into our lives and we just accept it for what it is? And I think having something tangible – whether it’s a cocktail or making some art or drinking gin or whatever – it might be those things that are a really great way of making things feel a bit more real and understanding that process.
Dan: Presumably from a corporate perspective, almost irrespective of what the transformation project is, I mean I guess it wouldn’t even necessarily have to be AI orientated, just as a way of breaking the ice and getting people to think a bit creatively, getting people to swap that fear in their minds for a bit of excitement and a bit of positivity. Presumably that’s such a powerful initial mechanism because so many of these projects have failed before they’ve even begun, right? Because they just don’t have the buy-in from people at that emotional level. So presumably, this is such a powerful mechanism for just getting that initial process underway?
Kerry: Yeah, absolutely, it’s a great way to kick off some kind of digital transformation project and you know if you’re using AI it’s absolutely fantastic because it gives people a basic understanding, and like you say, digital transformation projects are hard to make successful if the team aren’t on board or if people don’t understand. It’s vital that you have people with you so things like this do really help and as I mentioned with the art making particularly, I can go into a business and people can physically go through the machine learning process as in we would go out and take some photographs to create a dataset we would then train the neural network and we would then generate some pieces of art. So you know, you go through the process and you understand some of the basics around how AI machine learning actually works, and then obviously we can have wider conversations. And you don’t have to be able to code. When I go into businesses or universities or even schools, I use programmes or platforms that don’t require code so people can experiment and explore and play with. And obviously they need to understand the processes but you don’t have to be able to code or be some kind of amazing developer to do it. You can play and work in the space without having to do that.
AI it’s absolutely fantastic because it gives people a basic understanding, and like you say, digital transformation projects are hard to make successful if the team aren’t on board or if people don’t understand.
After those sessions, they don’t need to be that long, people will have a much better understanding which means then if you start implementing AI machine learning people kind of go, “Okay, I get, I understand it and therefore I know what it can do and I know what the potentials are and I also know what the dangers are.” Then people have less of that fear associated with it because they understand what they’re playing with and what you as a business might be playing with so it’s just much easier to get on board.
Dan: Yeah, and presumably one of the things it does in that context is demonstrate the value of human input as it’s a collaborative thing and it could absolutely be done entirely by the machine but presumably you can see a contrast between that?
Kerry: Yeah I mean absolutely, I think the thing is, we’re not at the point of AI like in the movies, ultimately, anything AI now is machine learning and a human has to be involved. No AI is without human input at the moment. Right now, we as human beings are making these decisions around AI and it’s important to understand no machine learning algorithm is starting on its own. We are starting it and then yes, for sure, it goes off once it’s been trained. It can then make predictions and go off and do its its own thing. But as you said, a lot of the projects that I’ve done that I’m particularly interested in are that sort of working alongside it because I’m really interested in what we can do alongside and how it can make us think differently or how it can augment our own creative skills.
I think the thing is, we’re not at the point of AI like in the movies, ultimately, anything AI now is machine learning and a human has to be involved. No AI is without human input at the moment.
So I think that’s that’s interesting and you know it’s also interesting because going back to the cocktails, when we first started working with the mixologist I remember talking to him over the phone and going to see him and when we got there I talked him through the recipes and he was just like: “Oh, I don’t want to do it, AI can be loads better than me. It’s gonna come and take my job.” So we talked quite a lot about it and just said: “We use these things as a starting point, they’re not necessarily what you’ve seen before but you have this amazing palette and this real depth of understanding, and we’d really love you to to use it almost like a brainstorming session.”
It took him a while but once he got into it, he actually really liked it. Interestingly, when we went back into his bar he had this little storage cabinet of all these draws and all these strange things, and that’s because he’s started to play with all these new ingredients that he’d never tried before.
So if AI can help us to do that, it will give us more time to do more strategic thinking and more interesting stuff.
It’s interesting watching the transition between someone being terrified and then once they get involved and realise, actually, as a human I still have a big part to play as a creative mind I can do something different here. I Love the idea of us working together and I think going forward that will become a much bigger thing in terms of the business side of things as well. We will use AI to do some of the inspiring stuff like I’ve just spoken about but also some of the more base work that we might not necessarily really love doing at the moment. So if AI can help us to do that, it will give us more time to do more strategic thinking and more interesting stuff.
Dan: And are you aware of many large companies using the kind of activities that you’re using? It and it seems to me that there’s basically 2 obstacles, invariably. One is a lack of data and the other is humans just not understanding it. Not sort of having that kind of initial buy-in and then working against it.
It just strikes me that you have, maybe not a complete solution, but a partial solution, to the second one of those and there are so many of these projects taking place every day around the world, but without that human buy-in it’s so hard to succeed.
So from your experience, other than the kind of things that you’re doing, are you aware of other companies doing this?
Kerry: Not that I’ve come across. I mean sure, AI is being used in a lot of businesses and in a lot of ways, and I know in things like in newspapers and publishing, people are starting to use AI to do some of the base work there. In terms of marketing, there are big copywriting AI tools that don’t necessarily work alongside you but they can create things like subject lines and sms notifications and basic web copy and things like that.
So I guess some of them can do the groundwork and then leave you to do some of the more interesting stuff. As a copywriter for 18 years, writing subject lines wasn’t always always the nicest job to be doing, so if an AI could come and help me to do that or to do that on my behalf, then that would be great. I don’t know how many businesses are tapping into them from a creative point of view, but yeah, that idea of working alongside I think is becoming a bit more commonplace.
I don’t know how many businesses are tapping into them from a creative point of view, but yeah, that idea of working alongside I think is becoming a bit more commonplace.
In terms of the creativity, I don’t really know. I’d love to if there’s anyone out there that is – I’d love to talk to them because I do sometimes think it’d be really nice to work with other people who are working in a similar space just so we could chat and bounce off each other and see how people are using it creatively, but it it still feels like it’s a bit of a new thing really, the creative side.
Dan: I’m particularly interested in the implications of this stuff from a sales and marketing perspective. Over the last few years, there’s been a huge emphasis, in all industries but in B2B in particular, on really nailing the customer journey. I just wonder if you’ve encountered any kind of creative ways that the kind of tech that you’re using could be leveraged to enhance the customer experience? Or elevate brands in other ways?
Kerry: Yeah, I come from a marketing background so before setting up Tiny Giant 3 years ago, I worked as a copywriter and creative head and managed creative teams in advertising and marketing. So although I don’t necessarily work in the marketing space, that feels slightly different from where I sit, I’m still obviously interested in what’s going on in that world.
So those kinds of things are a very common place in our life but we might not necessarily think that’s AI or that’s machine learning. It’s just so normal now to us.
So yeah, I mean AI machine learning is being used a lot, or starting to be used, in the world of marketing and I think one of the most commonplace areas that’s being used is Spotify and Netflix and Amazon, those kinds of product recommendations that we see, that’s all machine learning. So those kinds of things are a very common place in our life but we might not necessarily think that’s AI or that’s machine learning. It’s just so normal now to us.
And things like customer segmentation work very well. Traditionally, in terms of segmentation, we tend to group things as humans into large relatively generic groups, you know around age or whether you bought before and where you’re based, that kind of thing. Obviously, with machine learning, it’s a whole different ball game. You can very quickly analyse data and create much more targeted segments and you can then also even go further than that by automatically adjusting campaigns to be more personalised for each segment. And the thing with machine learning is that it can detect an unlimited number of things. So it can find patterns that maybe we wouldn’t even think about or we wouldn’t even spot.
And the thing with machine learning is that it can detect an unlimited number of things. So it can find patterns that maybe we wouldn’t even think about or we wouldn’t even spot.
And then there are a couple of creative uses. So Lexus did an ad which was the first AI written tv ad and I’m not sure if you remember that but they used AI to generate a script for their latest tv ad and they also used sentiment analysis. So they used as a data set, hundreds and hundreds of award-winning scripts, so you know, it was already based on something that they knew worked well already. They use sentiment analysis to see which parts of those scripts resonated most with the audience and then they created this AI generated script. But going back to the human side of things, it wasn’t just like, “Oh, let’s just take this script.” It was like let’s give it to Kevin Mcdonald who’s an award-winning film director and he can then take this script and turn it into this ad. I’d love to have seen that process of how much that film director added to that actual process. So it launched. I have to say. it’s probably not the best I’ve ever seen, but I really admired their innovative approach.
There’s also a really nice campaign that was done by a charity called Malaria No More. It’s worth looking up. It’s got David Beckham in it and he is using a range of machine learning tools that enabled him to speak multiple languages. So he delivered this talk at a kitchen table and just very seamlessly moved between so many different languages and it was all all about Malaria and combating Malaria, so it’s quite an important message to deliver. That video reached something like 1.6 billion people worldwide. So sometimes I suppose using machine learning in a creative way definitely gets you noticed, and that video got shared a lot but also I suppose because you could seamlessly translate. It was kind of using that Deepfake technology, which is always associated with really negative things. So he delivered it and it looked really amazing. But obviously because he could speak in multiple languages he was able to resonate with more people.
So it is being used a fair bit but just still feels like it’s still a long way to go like there’s so much potential and it would just be great to see all people getting involved and doing interesting things and I suppose it was the same with.
Dan: So from a kind of B2B perspective, because generally the B2B world tends to follow on a few years after the consumer world, I just wonder are there any creative applications of Ai from a user experience perspective that you’re particularly aware of?
Kerry: Not necessarily. I think the closest would be in terms of the Ada project I’ve worked on, because now Ada is a representative for science. So we’re working with other businesses so that they can sponsor her or can go and visit their labs or whatever and we can work like that. But yeah, I still think it’s very early days really and I suppose you know the big advertisers and B2C are going to be the ones that adopt it initially.
Dan: And I mean often a conversation I have with people about the stuff – and I’m by no means expert in any of this stuff – but it always strikes me that a common theme in all those conversations is that people will say it’s still kind of very early on we’re still not quite at that point. A Lot of the things that we refer to as things machine learning are very often fairly rudimentary forms of robotics and technology that may have existed for quite a long time but have been rebundled, repackaged and resold to the market with a slightly different value proposition. But this conversation is giving me a sense that maybe we’re not as far away as sometimes I think we are from maybe reaching that inflection point where actually all industries look at this stuff as a fundamental part of their toolkit.
Kerry: Yeah, I feel like that and I think that will become the case. Even in the last 3 years I’ve noticed that machine learning tools have become much more democratised and accessible. 3 years ago we were working very closely with developers to help us to bring anything to life and actually now I am learning to code but I’m not a developer by any means and it’s accessible for me. I can go in and I can make art with machine learning. And here are things you can use as a business or even as an individual and go and explore it. And I think as that happens, as more people start making tools which are accessible to non-coders or to smaller businesses, more people will be able to harness machine learning and AI for themselves to save themselves time and effort or even to stimulate new creative thoughts or ideas to help them to innovate and create new products and things like that.
So yeah, I do think I would just say it is still very early days. But even in the 3 years I’ve been working here, there’s been such a huge development in that time and I can imagine that in the next 3 there’ll be even more and actually it’ll be quite normal. But yeah, it will be interesting.
Dan: So thanks ever so much for sharing all of your experiences and ideas. Really, really interesting stuff.
Kerry: Pleasure. Thank you so much for having me.
