Ep. 91 | Human-centered AI with Jen Stirrup of Data Relish
This week Jen Stirrup, founder of Data Relish joins Allison Hartsoe in the Accelerator. If you don’t know Jen’s work in the AI space, you should. She is an advocate for human-centered artificial intelligence which is better explained as augmented intelligence. The question we explore this week is how can AI augment our own roles such that we behave more intelligently in customer interactions.
Please help us spread the word about building your business’ customer equity through effective customer analytics. Rate and review the podcast on Apple Podcast, Stitcher, Google Play, Alexa’s TuneIn, iHeartRadio or Spotify. And do tell us what you think by writing Allison at info@ambitiondata.com or ambitiondata.com. Thanks for listening! Tell a friend!
Read Full Transcript
Allison Hartsoe: 00:01 This is the customer equity accelerator. If you are a marketing executive who wants to deliver bottom line impact by identifying and connecting with revenue generating customers, then this is the show for you. I’m your host Allison Hartsoe, CEO of ambition data. Each week I bring you the leaders behind the customer-centric revolution who share their expert advice. Are you ready to accelerate? Then let’s go. Welcome everyone. Today’s show is about human-centered AI and to help me discuss this topic all the way from London is Jen Stirrup. Jen is the author of two books as well as the CEO of data relish, which is a boutique data science and business intelligence consultancy, but you may have also seen her on some of these amazing lists, the top 17 influencers in big data and analytics, the top 20 women working wonders in AI, ML and more and my favorite, the top bad-ass women working in data and that I have to say as a list I want to be on. Jen, this is awesome. Welcome to the show.
Jennifer Stirrup: 01:13 Thank you so much for having me along today Allison. I’m really excited to be here and this is one of my favorite topics.
Allison Hartsoe: 01:19 Excellent. Tell us a little bit about your background. I mean, to my knowledge, most people don’t start out as a kid saying I’m going to be an AI. I mean maybe that’s happening today, but how did you get here?
Jennifer Stirrup: 01:30 I just found I was fascinated by all things data and I was really interested in how we apply technology with human nature and how we could be human-centered using technology to really improve people’s lives.
Allison Hartsoe: 01:44 That makes sense, but when we think about artificial intelligence and everybody is so excited about this, but for many folks, the words bring up visuals from minority report and Terminator. Can you tell us a little bit about how you think about AI and what makes it human-centered?
Jennifer Stirrup: 02:02 I think what makes it really human-centered is it really allows us to automate processes and it allows us to meet the most of our human abilities. I’ll give you some examples, so if you use that facial intelligence to try and read handwriting, then that means that people can be freed up to do other things. Now, it doesn’t mean that we take the human ITV completely. I don’t know about you, but my handwriting’s terrible.
Allison Hartsoe: 02:28 As is mine.
Jennifer Stirrup: 02:30 I spent so much time typing and I don’t write anymore and to read my handwriting, you would definitely need human intervention in order to do that because I think it really helps to make things faster and better, particularly when we think about customers and their expectations about the services that we supply to them.
Allison Hartsoe: 02:48 So handwriting makes sense and sometimes when I think about automating these processes and then helping people more or less smooth out their lives, it reminds me a little bit of where we were in the email paradigm or the fax paradigm ages ago, and it didn’t actually give us more leisure time. It is actually kind of made work blend into our personal time more. As we’re looking at human-centered AI, are we really just looking at a world where we’re always kind of blending the work and the personal life because everything is so seamless?
Jennifer Stirrup: 03:23 I think actually our special intelligence will in some ways become boarding and it will become part of our everyday and say for example your smartphone, your smartphone already has the basic AI entity already and the best technologies that technology that we don’t notice, I think it will have a social impact. It will impact say people’s jobs, I think, it will make things in some ways harder because it will automate some processes and people’s jobs will change. Now as humans we don’t like change and we do not like to be changed, and I think that might be difficult for some people and I do understand that. Because I really think the social aspect is an important piece.
Allison Hartsoe: 04:03 What you’re talking about with humans that don’t like change and if we go back to the business paradigm, I think this is one of the areas where, and what does it mean to put AI leadership into my company, how does that change the staffing? How does that change the company? How do you help executives think about that?
Jennifer Stirrup: 04:22 Well, first of all, what we do try to do is to encourage them to make AI part of the vision. As I believe really strongly that the most successful companies are the ones who put the customer at the heart and center of their vision for the company.
Allison Hartsoe: 04:37 I love that.
Jennifer Stirrup: 04:38 Yes, it’s absolutely true and if you do that, you don’t have to really watch so much what the competition are doing and you can use AI as part of a wider strategy to make your customers more happy. The way that you can do that depends on how much you want to involve AI in your business. So you may want to start off simply by using AI to automate a process or make something smoother. So it use handwriting example. Maybe that handwritten form is a part of the way that the customer interacts with you and if they get a faster service, they’ll be happier.
Jennifer Stirrup: 05:12 Everyone expects everything right now.
Allison Hartsoe: 05:14 That makes sense.
Jennifer Stirrup: 05:15 Absolutely. All you can make AI very much a different incentive of the customer engagement strategy. So you could use chatbots, or you could even use speech recognition. Speech analytics is a very important technology at the moment. Take, for example, you have Google and Alexa and you interact with these technologies using your speech and AI is very much a part of that. So that’s a stretch school for many organizations. So to get started, I think organizations have to choose an actionable business problem that fits in with the overall strategy and their vision for the organization. When they do that, they can usually solve a problem and that helps them to articulate the benefits of using AI. And that could be something like we want to increase customer satisfaction, we are going to implement a process such as a chatbot to help our customers engage with us better.
Jennifer Stirrup: 06:08 So once everyone in the team understands the benefit, we can start to de-risk the project. So we start to take some of the risks out of it so we can see, well we’re going to do plenty of testing and we’re going to develop. Obviously I will test internally first. And then that allows the projects to become almost like a beacon across the organization that AI is coming and it’s in place. What we don’t want to do is delay the decision to use AI cause that’s really kicking the con for the general rule to someone else to pick up. And I really think the smart organizations want to use any tool possible in order to be very customer-centered. And AI is just one of those tools. I think so
Allison Hartsoe: 06:48 It’s interesting when we talk about being customer-centric and AI. And then the first example was about the handwriting and then you mentioned the chatbots. Is there something particular about customer data that makes it better for AI or allows it to be a better target?
Jennifer Stirrup: 07:07 I think that’s a really good question because we start to think about how bias and the data can actually influence results. So, for example, we know in some instances some recruitment AI programs have inadvertently promoted men over women. And the reason for that is the AI has got more data about men in it. So it chooses basically what it knows. And that introduces a bias. So what we have to do is recognize as humans and an AI as well, that these biases exist, but it may also work the other way perhaps. And you do see sometimes if customer interactions can be influenced by the individual’s particular biases. Um, so I know I’ve experienced this myself sometimes and I’m fairly small and softly spoken, and sometimes then it’s harder to fight to get attention if you feel aggrieved in some way as a customer. But if you’re dealing with an AI, like a chatbot or a customer interaction in that way, that really is removing some of the bias which maybe someone else might have against me. So I don’t think the issue of bias is very clear cut because I do think the customer data is crucial to really offering that really tailored and good service the customers never expect.
Allison Hartsoe: 08:26 That’s a really interesting example because I think about the American airlines number of Eagles example that they arm their front line with, and I’ve never seen this in person, but the story goes something like this where they weigh in not just your customer value, but how much pain you’ve been through recently and then they arm the front line with a sense of which person should I upgrade first second and third. Whereas before perhaps it was left to gate agent and the salesperson who’s chatting up the gate agent might get more preference than someone who actually has been through hell and maybe doesn’t have as high of status, and would that be a good example of a way that human bias can be overcome using artificial intelligence?
Jennifer Stirrup: 09:13 I agree with that, I really do think. However, we need a strong moral compass when we talk about technology that can impact all sorts of things to make sure that customers get an equitable treatment. Talking about airlines, I saw an example really it’s actually set up some algorithms to work out which families would be most likely to pay to set together. And the airline actually splits up these families based on the algorithms because they felt it would make them more revenue. So, which is just shocking. And for me, that’s a real bias because nobody around that table, nobody who thought up the idea, nobody who developed it. No one who tested it ever thought, how am I gonna manage? Is this the right thing to do? Because something is technically correct doesn’t mean that it’s morally correct and I don’t know about you but, I know that when I traveled with my son, I would really prefer to sit with him and yes I would pay to do that.
Jennifer Stirrup: 10:11 And trust me, you want me to pay to sit next to my son as well because he’s going to be pestering you for sweets the whole way. So other people that have a vested interest as well, it helps the job. But fortunately someone somewhere put the brakes on the idea, but it did make it through to the headlines. And I really think that we have to think very carefully when we use technology and what it does I think is make visible is biases, which may be people did but never articulated before and maybe as a whole it will help improve service across the industry.
Allison Hartsoe: 10:43 I think we would be shocked in general, we tend to think that we are not very biased, but there was the book that came out, I think it’s called everybody lies and he used the Google search results as evidence of how biased we actually are in a variety of ways across the United States in different segments and the things that we tended to search and I wonder if this is really if human-centered AI is really starting to shine a light or could shine a light on what’s correct to do with the technology. Like you said about the moral compass coming into the data. Is this an activity where a person might need a certain amount of ethics training to go alongside AI and perhaps like a new job or a new category?
Jennifer Stirrup: 11:27 I do agree that actually, that should happen even if it’s not happening right now. I know many organizations have an induction process and that ethics could be embedded and is part of that process, but I do think the whole story about ethics really means that it should be an ongoing conversation and not something that we ever should stop talking about and being customer-centered really means that we have to think about the ethics as we go along. Especially if we see something in the news and you think actually that’s not a good way to treat people, and we can see how your customers respond to that. For example, if you do the over the airline example really makes people thought more carefully and to think we have to try and use technology for good purposes and it can so easily be used the wrong purposes. One example is Deepfake. Deepfake technology is all about see, for example, taking someone’s face and embedding it on a film. And there’s been some really good examples, if you want to call it that, involving Barack Obama for example. The fundamental truth is these videos. These programs are ultimately fake. And I think we have to think very carefully with this technology being so easy to use, how it’s used and how we can stop it as well. Here we can identify a deepfake as opposed to a real video.
Allison Hartsoe: 12:48 And does that take human judgment to create the model that identifies it or is it a combination of the AI and the human tuning the model that makes it come through?
Jennifer Stirrup: 12:59 I think it’s a combination. If you look at, say Facebook, for example, or Twitter, they receive lots of reports for many different in violations of conduct online. And some of those will be genuine violations and some may just be, and somebody attacking somebody online, which is not good. Or it could just be something malicious. So, but what you do find is that these organizations are still recruiting people to look at the human aspect and make a human judgment, even though they’re using technology as well. So I do think that we should always call it augmented intelligence in a way because it augments what we’re doing, it makes things faster. And hopefully better for people.
Allison Hartsoe: 13:42 I really like that instead of artificial intelligence calling it augmented intelligence because that keeps the human in the mix and everything I read about AI seems to come back to the idea of judgment that it’s almost impossible to program or even come up with human judgment because it is cultural and changing. It’s not very easy to embed.
Jennifer Stirrup: 14:05 That’s right. The example I like to think about as the example that Steve was New York gifts.
Allison Hartsoe: 14:10 Apple’s founder.
Jennifer Stirrup: 14:11 Yeah, that’s right. He gives an amazing example of which I’m going to call the Wozniak test. There Wozniak test is all about getting an AI to make you a cup of coffee. Oh, that sounds really easy. But actually when you think about the AI coming in to your house, going through your cupboard, identifying the coffee, identifying the kettle and the water and giving you the right amount of ingredients and the right order at the right time, it involves an incredible amount of general knowledge and that’s what’s difficult to in view when you look at AI. So one definition I like of AI is actually to look at it and how it’s being used. So we have artificial general intelligence and that’s what you alluded to earlier. Maybe Minority Reports or Terminator for example, as a an AI which has good general knowledge and expertise, common sense as it were, but we also have artificial and narrow intelligence.
Jennifer Stirrup: 15:08 ANI artificial narrow intelligence is what you have in Google or Alexa. When you speak to the app, it’s human-inspired and human-centered, but it’s only very good at one thing, which is speech recognition is not doing everything. And to think that distinction allows businesses to really focus on AI leadership because they can start to narrow it down and see, well how are we going to use AI? And that the best thing they can do is dismiss the notion of artificial general intelligence and pick up an example of artificial narrow intelligence that the faint achievable and do it in small steps with achievable business schools. And when they narrow the scope down, that makes it more successful then. It’s about proving success and proving it in a number of series, a series of small successive steps and when you prove the AI in that way, it’s all leads up to a bigger success in the long term.
Allison Hartsoe: 16:01 I love that idea and when we talk about success, there was an example you shared with me before and I wonder if there’s a role that companies can play to get beyond just product pushing and even get beyond customer satisfaction to be more centered in how do I improve my community or how do I improve the lives of the people that I serve. And the example you were talking about was about homelessness. Would you mind sharing that example again because I think it’s so interesting, although I don’t know that it directly connects to a company. Maybe it could,
Jennifer Stirrup: 16:31 Yes. The example they try to engage was to try and analyze and predict who would be most likely to become homeless. Now, that is a huge impact on savings because if you can intervene and get people to basically to catch them before the fall, then that is very human-centered. But it also means you can have a real impact for that family. And they are a parallel was even though that was faced much, a case of social impact is still analyzing people’s behavior, and you are still trying to identify and protect an outcome and for businesses, I think that’s really what they want to do is analyze and understand the people that they’re dealing with and then have a good impact for that person. So what you can do is think about their shopping habits for example.
Allison Hartsoe: 17:17 I was just thinking about Facebook and their ability to identify people who are being bullied too much online and try to catch them before there’s a negative outcome. But your example works too.
Jennifer Stirrup: 17:27 Yes, I love that example and I wish there was more online opportunities for people to be caught bullying or to be caught and any sort of a bad online behavior has coming back to the moral compass, I do think some things, that’s why Facebook is seen less adoption with younger people now because it had a really bad experience and they do go back online after that. And I think that’s a real shame in a way because the technology can be used for good. I’m on Facebook and I use it to connect with family members all over the world, so it kind of a good use. For businesses, it does mean they can connect well with their customers as well, and it’s nice when that happens, which they really do think that should be more efforts. Pretend you’re trying to catch people who behave badly online.
Allison Hartsoe: 18:12 Do you think AI can help businesses develop, let’s call it a U-shaped curve. In general, a business wants to go up into the right, but in some cases that straight line of going up into the right makes them worse and worse at something, whether it’s propagating bad media or a troll get more pudding. Or in the case of gambling, there’s a big fish case where they were constantly getting more people to gamble and spend more money. There’s a point where that’s not a great business outcome because they’re feeding a gambling addiction. So could AI help us create these u-shaped curves and ultimately be, like you said, more human-centered?
Jennifer Stirrup: 18:51 I do really hope so. I think what we need to do is identify really good test cases. We also don’t want to end up with a Cobra affects will be end up with very unintended consequences, and I think that’s what happens with Facebook in some ways as well. They laid out a lot of data, they didn’t think about it carefully, and then they ended up, they are really ammo and I think it’s all a bit testing. I do think diversity and inclusion are crucial to these debates and discussions as well. Diversity and inclusion brings a different lens to the discussion, but I think also it brings something deeper than that. It means that people question each other and themselves more and hold each other to account more. And when that happens, I think you have a higher bar of success and the project then you might reach.
Jennifer Stirrup: 19:36 Otherwise if people gave themselves a very low bar, if success in a project and the ethics can be missed out and we don’t want that. So I think that’s why having people of different backgrounds, different ages and different ethnicities, different genders, having a great mix of people is part of the team is so important to the success of the technology. We don’t want the future of technology to be held in the hands of a particular group of people and we’d encourage and leaders who are listening to this podcast to think about the advanced, the mix and why that’s important, and it means a higher bar of success for the organization and also for the customer.
Allison Hartsoe: 20:15 I love that. Well, Jen, if people want to reach you, what is the best way for them to get in touch?
Jennifer Stirrup: 20:20 You can reach me on LinkedIn. I’m very active on LinkedIn. I also blog over at jennstirrup.com and my company websites where I also blog is datarelish.com.
Allison Hartsoe: 20:31 This is a great site to check out just to keep up with what’s going on. I enjoyed reading through your blogs myself.
Jennifer Stirrup: 20:38 Thank you. Thank you very much.
Allison Hartsoe: 20:40 As always, links to everything we discussed, including Jen’s site and blog are going to be at ambition data.com/podcast. Jen, thank you so much for joining us today. I’ve really enjoyed our conversation.
Jennifer Stirrup: 20:52 Thank you very much, Allison, and thank you to you in the ambition data team for having me along today.
Allison Hartsoe: 20:57 Thank you. Remember everyone when you use your data effectively, you can build the customer equity. It is not magic. It’s just a very specific journey that you can follow to get results.
Allison Hartsoe: 21:10 Thank you for joining today’s show. This is your host, Allison Hartsoe and I have two gifts for you. First, I’ve written a guide for the customer centric CMO, which contains some of the best ideas from this podcast and you can receive it right now. Simply text, ambitiondata, one word, to three one nine nine six and after you get that white paper, you’ll have the option for the second gift, which is to receive the signal. Once a month. I put together a list of three to five things I’ve seen that represent customer equity signal, not noise, and believe me, there’s a lot of noise out there. Things I include could be smart tools I’ve run across, articles I’ve shared, cool statistics or people and companies I think are making amazing progress as they build customer equity. I hope you enjoy the CMO guide and the signal. See you next week on the customer equity accelerator.