Artificial Intelligence and Business Strategy
In collaboration withBCG
After earning her undergraduate degree in philosophy, political science, and ethics, with aspirations to become a lawyer, Sidney Madison Prescott was drawn to technology jobs that specifically emphasized data quality and governance. In 2020, she joined music streaming service Spotify as the global head of intelligent process automation, where she uses robotic process automation to automate tasks and free up workers to focus on higher-value-added and more creative work. For Sidney and her team at Spotify, AI and machine learning are not tools to replace jobs; they enable humans and machines to work together for increased efficiency and productivity.
In the final episode of Season 3 of the Me, Myself, and AI podcast, Sidney joins hosts Sam Ransbotham and Shervin Khodabandeh to share her views on automation, augmentation, and fostering engineering talent.
Read more about our show and follow along with the series at https://sloanreview.mit.edu/aipodcast.
Sam Ransbotham: Engineers are not stereotypically known for their dancing prowess, but increasingly, organizations need to choreograph how humans and machines work together. In this episode, we chat with Sidney Madison Prescott, global head of intelligent automation at Spotify, about the dance between humans and machines that improves business processes.
Welcome to Me, Myself, and AI, a podcast on artificial intelligence in business. Each episode, we introduce you to someone innovating with AI. I’m Sam Ransbotham, professor of information systems at Boston College. I’m also the guest editor for the AI and Business Strategy Big Ideas program at MIT Sloan Management Review.
Shervin Khodabandeh: And I’m Shervin Khodabandeh, senior partner with BCG, and I colead BCG’s AI practice in North America. Together, MIT SMR and BCG have been researching AI for five years, interviewing hundreds of practitioners and surveying thousands of companies on what it takes to build and to deploy and scale AI capabilities and really transform the way organizations operate.
Sam Ransbotham: Today we’re talking with Sidney Madison Prescott. She’s the global head of intelligent automation at Spotify. Sidney, great to have you. Thanks for joining us.
Sidney Madison Prescott: Great to be here.
Sam Ransbotham: Well, the title of our podcast is Me, Myself, and AI, and we tend to focus on you as an individual, so maybe let’s start there. Can you tell us a little bit about your current role at Spotify?
Sidney Madison Prescott: I currently head up a global team. We are primarily within the U.S. and the U.K./European areas, and we are working toward modernizing and really improving the efficiency of the workforce within Spotify. So this is looking specifically at all of the processes across many cross-functional teams to ensure that, whether we’re working with the UI of the application, whether we are talking about the finance function, we are being as efficient as possible in the ways that we go about our daily routines as Spotifiers. And really, we’re taking that on by looking very closely at ways to leverage intelligent automation. This is, in essence, a tool stack where we combine robotic process automation, artificial intelligence, and machine learning in order to really facilitate that digital transformation. And our team is working very closely primarily right now within the finance function. However, we have branched out to our ads business operations as well, really looking at, again, that digital transformation and how we can really unify the front-office and back-office processes that our Spotifiers work on today.
Sam Ransbotham: So your degree is … you have a major in philosophy with a double minor in political science and ethics from Georgia State University — go Panthers! That’s certainly not a typical connection to artificial intelligence. Can you make a connection between your background and your path to get from there to your current role?
Sidney Madison Prescott: I agree; it’s not the traditional trajectory of someone who would work specifically within these really kind of emerging and innovative technologies.
I would say my degree really prepared me to think critically about various tasks, various responsibilities, and to identify a business problem [and] potentially pitch a solution to that business problem that was technical in nature. And I really do rest very firmly on the critical thinking and the logic skills that I received out of my philosophy degree. Those really helped facilitate my growth and really amplified my strengths in terms of process reengineering and really being able to think outside of the box when it comes to how we are translating business problems into business efficiencies through technology.
My career path really began with an internship, and I was specifically working with configuration and asset management. And so I was actually a pre-law concentration, and I wanted to become a lawyer. I began working on software contracts, and that really was my first foray into technology. I began to understand business law and business contracts; however, I really started to see the amazing contrast between how we think about what technologists do and the reality of the work that is done around the world.
Once I started diving into the software contract and the true-up aspect of the business — and that’s really where I moved very firmly into becoming a technologist — once I made that transition, from there I was really diving very heavily into the connections between the disparate systems and the challenges that we have with those configurations and the outputs or the quality of the data that comes from those various configurations and integrations with different systems.
And so through that, I really started to home in on a passion for data quality and governance, and really looking at data as an asset. And this is where I began to firmly understand the correlation between process efficiency, process automation, and [how] moving toward a more automated business provides you with greater data insights and better … executive decisions that you can make based off of those data attributes.
I started that journey within Fiserv, and then from there, I had an opportunity to dive into a proof of concept on robotic process automation. And this was relatively early on, so I would call myself basically an early evangelist of that technology. This was back in maybe 2015. And so through that proof of concept, I began to understand: What is robotic process automation? Why is it something that can be used to facilitate a digital transformation at an enterprise level? I also started diving at that point into intelligent automation, which is that combination of robotic process automation, artificial intelligence and machine learning, and sometimes optical character recognition as well.
So it was very much a transition from software contract legalese, to configuration and asset management, to data quality and governance, and then that segued into more of the emerging technologies, such as artificial intelligence and robotic process automation.
Sam Ransbotham: That’s particularly interesting, because most of the people, I think they come to this awareness of the importance of data governance and quality after that foray into machine learning and artificial intelligence, when they find how much depends on data. And then they investigate why everything’s falling apart, because of ugly data, and then they get interested in governance. And you’ve swapped that around to come from the other direction, then. I don’t want to take this too much in a governance angle, but I find it difficult to talk about. So one of the things I talk about in class is the importance of data quality and governance. I just feel so boring as I say it. How do you get over that sort of, away from —
Shervin Khodabandeh: How do you make it interesting?
Sam Ransbotham: Yeah, how do I make that interesting? How do you get people fired up about those aspects?
Sidney Madison Prescott: Data quality and governance are often seen as an afterthought, meaning it’s just not something to focus on, right? You’re more focused on accelerating business growth or solving different business challenges. But from my standpoint and how I speak to my stakeholders about this is, the acceleration of your processes and the automation of your processes, that is only one piece of the bigger puzzle. And so if you are automating processes, but the outputs or the outcomes of those processes are data elements that are not reliable, in essence you are defeating the purpose of that automation. I mean, automation is only as good as the outputs that it provides back to us, right? Whether it’s the engineering team, whether it’s business stakeholders in finance, etc.
And particularly when you look at the decisions that we have to make based off of data outputs, that is where it really becomes critical. Because when we look at risk and control, when we look at SOX [Sarbanes-Oxley Act] processes specifically — so, those that impact the financial health of a company — this is where data is absolutely essential. It’s essential that we get it right, that we can rely on that data. And I think really what I bring attention to is a deep desire to mitigate the risk that is created by bad data in your environment and the decisions that you make based off of that incorrect data.
Sam Ransbotham: As part of your role, your team goes in and increases automation in areas. How do people react to that? Are they thinking, “Oh my gosh, there goes my job? I used to do these tasks, and now they’re going away?” Or are people happier? Do they feel threatened? What’s the reaction to the sorts of processes that your team brings in?
Sidney Madison Prescott: It has been very interesting. Initially, there is a thought that, “Oh, it’s going to be scary.” But the great thing about Spotify is it’s a very dynamic and agile environment, and a very creative environment. And so, because of that, it was very conducive to more excitement and curiosity about the technology. It’s been primarily really excited. People are happier. They enjoy the bots. They nickname their bots. So it’s been primarily, I would say, very positive, for the most part.
And one thing I’ll say as well — I think that is a result of both the culture within Spotify and also the fact that we are very much focused on making the Spotifiers’ engagement within the company more efficient and hopefully facilitating less attrition as a result because of that increase in employee satisfaction with their role.
Shervin Khodabandeh: Sidney, I’d like to pivot on something you said. You’ve been talking about automation and various ways of automation, intelligent automation. What are your thoughts on other ways that AI can create value? Because automation is sort of at the far end of the spectrum, right, where AI decides and executes, and completely replaces [a] human? And we’re seeing a lot of interest and attention recently in where AI and human work together to do something that neither one of them could do alone as effectively. What are some examples of that, and what are your thoughts on more, I would say, integrated human-AI ways of working? Does that make sense?
Sidney Madison Prescott: It does. And this is actually a subject I’m very passionate about, because I think this is the future of what we are going to see, not only as technologists, but really as a society in terms of incorporating machines that are capable of, I’ll say, a version of cognitive functionality, with that intuitive cognitive nature that humans have. The merging of those two? I think that is an extremely propelling value proposition, specifically in relation to the ways that we can continue to evolve technology. And I would agree: I think what we are seeing now is less of a desire to say that humans step away from a particular process and the machine takes over, and more of a, I’ll say, almost a dance — a collaborative dance, between the machine and the human.
And I think this is wonderful, because what it enables us to do is look at the ways that humans flourish in relation to their cognitive capabilities, and then the ways that machines can basically help enable humans in the areas where we are not as strong, with large populations of data being able to quickly assess deltas in that data — being able to see patterns in millions of rows of data. These are the areas where AI can really flourish for us, AI/machine learning. The more cognitive functions where maybe a decision needs to be made off of that data, that’s where the humans can come back in and pick up that work. I call it human, really, augmentation of humans, at the end of the day. And the goal would be maximizing our potential as humans to really —
Shervin Khodabandeh: Digital dancing.
Sidney Madison Prescott: Yes. Exactly, exactly. To really facilitate better outputs, and to also allow us to really … it’s almost as if we amplify our own abilities with these machines, to the point that we can really say that we are operating at an optimal level, whether we’re [a] business, whether we are engineers. And then, of course, those outputs would naturally benefit the enterprise at the end of the day.
Shervin Khodabandeh: So this notion of the digital tango — and, Sam, I have to give you credit it for that because this is something you’ve been talking about — it’s actually quite interesting and it feels quite underappreciated: that it’s either human solo or AI solo, and the middle ground, where [a] human has to do things differently because of AI, but AI sort of morphs around what humans [are] good at and adapts, and [the] human does a bit of the same thing. It feels completely underexplored, partly because it feels a little scary, right? You have to rethink processes and existing ways of working, and existing norms of job descriptions and things like that, and as you said, this is sort of the future. What are your thoughts or ideas on how this could become more accepted — that evolution, that opening of people’s eyes to, it’s not all or nothing, but it’s actually in the middle ground, that there’s so much opportunity for us.
Sam Ransbotham: Any analogy for dancing is going to terrify me right off the bat.
Shervin Khodabandeh: Well, you came up with it.
Sidney Madison Prescott: It is a very interesting question. And I think that the hesitancy to really embrace this kind of human-in-the-loop machine and human partnership comes from the human resistance to change. And we all know that this is something that we really struggle with — change — as humans. We believe we are quick to adapt, but in reality, we don’t adapt quite as quickly as we’d like. And I think the fact that we have not yet embraced the digital-human workforce, and really combining those two together in almost a seamless workforce, is because we are so reliant on our prior understanding of what work — and I’ll put that in air quotes — what “work” is, right? The fundamental element that is going to change this is redefining what we mean as humans when we say work. What does work mean? Whether it is an accountant, whether it is an engineer, whether it is an engineering manager, an executive in the C-suite, what do we mean by work?
And we’ve started to see a bit of this transition within the pandemic, right? We’ve gone into the virtual world. We’ve heard pros and cons about this. And we started to see pushback on some of the companies that are attempting to embrace that more virtual workforce concept. And this, I believe, is the playing ground for the future, which is, can we embrace a virtual workforce and a reality in which workers around the world are able to work in a virtual manner to facilitate the growth of the business?
And the next piece of that is, in that virtual work, can we move away from thinking of these silos of human versus machine-relegated tasks? And I believe if we can merge those together — the virtual workforce combined with a redefining of what it means to work as a human, and almost think of ourselves as “What skills do the future accountants need? What skills do the future back-end engineers need? And how are those skills going to play seamlessly with these emerging technologies?” — that, I believe, is where we’re going to hit the sweet spot where we no longer have this almost tension between, “OK, well only humans can do this.” I hear and see this a lot in my work, which is, “Oh, a human has to do this.” I’m like, “Well, humans are prone to error, so maybe we don’t always need to rely solely on ourselves,” and thinking less of the machine as an adversary and more as a partner and enabler in that business process.
Sam Ransbotham: It sounds like … I know you have this citizen data scientist program. That sounds like exactly what that program is all about, teaching those basic dance steps.
Sidney Madison Prescott: Yes, very much so. So within Spotify, we’re very focused on our citizen developer community, which is what we call it. And this is where we are enabling our business stakeholders, through upskilling, various boot camps and trainings internally, to really enhance their understanding of these emerging technologies, and we are really encouraging our business stakeholders, specifically, to embrace the technology in a way that really resonates with their job responsibilities, currently and in the future as we continue to grow the business.
Sam Ransbotham: Sidney, we’re recording this episode right around the holiday season, when Wrapped has just come out. So in this Wrapped product, Spotify members get a deep dive into their most memorable listening moments of the year, including podcasts, and we hope this podcast, I guess. Can you perhaps use that to illustrate how you’re doing some of these things?
Sidney Madison Prescott: Absolutely. So there is a heavy amount of data mining/machine learning that’s being used in order to facilitate kind of the presentation of, “Here’s your top artist. Here’s how many times you listened to them. Here are the different genres that you’ve visited.” It’s a great example of, again, leveraging data mining in such a way to create almost this visualization, if you will, of what your listening landscape has been as a customer of Spotify.
And we even see this in a lot of our back-end processes as well, where we are really looking at ways to leverage … we have massive amounts of data, and we’re really looking at, “OK, what are the insights? How do we actually use this to do some predictive analytics?” whether it’s the ways our systems are interacting, whether it’s actually understanding process inefficiencies. So we’re looking very closely at monitoring deviations in our various process flows in order to better identify where we have areas of inefficiency that we can zero in on and tackle those to make the overall workflows of our employees faster and more efficient.
So it really is about mining that data, but we’re mining it for critical information that can help us to be more proactive about inefficiencies, more proactive about making better decisions, predictive analytics on maybe challenges that we might foresee coming down the road for a particular business process, even predicting increases in volume and how that will translate throughout the environment. So it’s a really great use of, I think, data that otherwise would … if you don’t use that data, it’s almost a waste, because it’s like you can just glean so many insights from it.
Shervin Khodabandeh: Sidney, there’s a war for talent in this space, and many companies, digital natives and otherwise, are struggling to attract, and retain, and cultivate the top talent here. What are your thoughts on that, and is there something you could share about how Spotify goes about doing that?
Sidney Madison Prescott: Yes. So this is a really interesting topic, because it really comes down to the pipeline — the pipeline from all the way up to elementary school, down to whether or not it’s technical college, four-year college, etc., and into the professional workforce. And the question becomes kind of almost back to what we initially spoke about: Are we facilitating an understanding of what the future roles within the enterprise will look like and the skill sets that will be needed? And are we enabling individuals throughout the course of their academic or technical lifetimes to better engage with these different technologies at an early age?
And I think what we’re seeing is, there’s a lack of understanding of what it means to be a technologist. I think there’s a lot of misconceptions floating out there, specifically within, I would say, the younger generation. And I think we need to do a better job of really teaching younger students — and even preteens — what a career in technology looks like and, more importantly, what it looks like 10 years from now, 15 years from now, when they’ll be entering the workforce.
And then you back in from the enterprise perspective, and you start reaching out to those college students, you start reaching out to those high schoolers, and you help them to see a day in the life. We do a lot of “a day in the life of an engineer,” “a day in the life of an engineering leader,” and kind of showing it’s not someone in a hoodie that’s hacking on a computer, which is the movie element, right? It’s a lot more than that.
And even myself, when I moved into technology, I found it quite fascinating to see the day to day, because I myself had no idea. I had no foundation to understand what it meant to be an engineer. And I think that piece is where we can really tap into a future workforce, helping them to understand the nuances of today’s engineer but also tomorrow’s engineer, and what it takes to actually succeed in a career in technology.
Shervin Khodabandeh: The distinction between today and tomorrow that you rightly point out is actually quite important, because I could imagine if my son asked me, but for the fact that I spent time in this area professionally, I would be quite biased about what engineering was like in my era. And it feels like the rate of change is so fast that prior experience and advice becomes very quickly irrelevant. And so this notion of … if you go back 20, 30 years ago and you looked up at somebody in a STEM program, or would be looking up to folks, trying to understand what it would be like to step in their shoes, it feels like those steps are going to be completely different in the future.
And so this willingness to really be open-minded, and also the exemplars of, what [does] tomorrow’s engineer look like? Because it feels like the old approaches are not going to work anymore. I mean, folks have to be much more exposed to what’s happening now versus what happened 10 years ago.
Sidney Madison Prescott: Agreed, and this also leads back to that need to really ensure that as engineers not only move into the workforce but as they mature out their career footprint, that they are consistently remaining abreast of the changes and the ever evolving ways that technology is manifesting in their field.
Because even the engineers today will tell you, getting into engineering is very different, even from sometimes what they are taught in school. And so that really is the disconnect: Are we teaching the right things? And then also, are we continuing the upskilling, right? As technology evolves, are we still moving midcareer engineers into a position where they feel empowered to continue learning about new technologies and being open to new opportunities to actually leverage those innovative tools?
Shervin Khodabandeh: Very well said.
Sam Ransbotham: What are you excited about next? What’s coming next for Spotify? What’s fun for you? What’s big on the horizon that we should be looking forward to?
Sidney Madison Prescott: Oh, I’m really excited about the, again, that human augmentation, looking for ways, whether it is through artificial intelligence, whether it’s through chatbots, whether it’s through optical character recognition, of really creating a seamless experience — I’ll say a seamless work experience — between the human and the machine. I fundamentally believe that that next frontier is a frontier where humans almost seamlessly interact with the different tools in their environment in a way to better facilitate faster outcomes. And we’re really looking, specifically at Spotify within my team, we’re looking at ways to leverage both machine learning and artificial intelligence in order to really enable the business to move faster and to focus on more value-added tasks. And so we’re looking at the amplification of our existing machine learning footprint — specifically, looking at ways that we can take data that is currently not machine readable, translate that into machine readable data, pass that off to a bot, and then potentially pass that off to a human. So again, building that tool stack, and really a very almost intuitive workflow between the machine, between the human, and the outputs that are received out of that.
And we’re also looking at ways that we can start to continue to mature out the tool stack, whether we’re looking at workflow automation. We’re starting to look at the dynamic between having a front end, which might be some sort of user interface with a chatbot, and then have a back end where it’s actually a robotic process automation workflow built into it, to then trigger some different workflows. So creating more, I’ll say, complex and nuanced ways to leverage these tools rather than siloing them off and saying, “Oh, this process is only robotic process automation,” or, “This process is only a chatbot.” So, breaking down those barriers, and then, of course, continuing to break down the two barriers that I feel truly exist within most enterprise environments and also lead to the most blockers, which are the silo between the business and the technology side, and then also the silos that exist between the front-office and the back-office functions.
Sam Ransbotham: Sidney, all quite fascinating. I think I particularly enjoyed discussing breaking these important barriers, like the barriers between business and technology, and between the front office and back office. And I think that analogy to digital dance has a lot of potential. Thanks for taking the time to talk with us today. I’ve really enjoyed it.
Sidney Madison Prescott: Absolutely. This has been a great conversation. And again, this is ever evolving, this entire industry, and it’s very exciting to be a part of [it], so I would definitely encourage listeners to dive in and start really learning about all of this, because it’s the future, but it’s also now, so it’s really important for all of us to get engaged and get excited about what this means for each and every single one of us in society.
Shervin Khodabandeh: Thank you, Sidney. You’ve been very inspiring.
Sam Ransbotham: Thanks for joining us for Season 3 of Me, Myself, and AI. We’ll be back in two weeks to kick off Season 4 with a lively conversation about innovation — and Star Wars — with Mark Maybury, chief technology officer of Stanley Black & Decker. In the meantime, please remember to subscribe and rate and review our show on Apple podcasts and Spotify.
Allison Ryder: Thanks for listening to Me, Myself, and AI. We believe, like you, that the conversation about AI implementation doesn’t start and stop with this podcast. That’s why we’ve created a group on LinkedIn, specifically for leaders like you. It’s called AI for Leaders, and if you join us, you can chat with show creators and hosts, ask your own questions, share insights, and gain access to valuable resources about AI implementation from MIT SMR and BCG. You can access it by visiting mitsmr.com/AIforLeaders. We’ll put that link in the show notes, and we hope to see you there.