Spotify and the future of intelligent automation
On this episode of TELUS International Studios, we're joined by Sidney Madison Prescott, global head of intelligent automation at Spotify. Sidney shares her take on the future of artificial intelligence, robotic process automation and machine learning as well as some of the best practices used at Spotify to drive efficiencies.
Who are the largest players in AI? Does reinforcement learning work? And how far away are we from roads filled with self-driving cars? For answers to these questions and a lot more, be sure to tune into the episode.
To learn more about TELUS International and our digital CX solutions, contact us today.
Intro: TELUS International Studios where customer experience meets digital transformation.
Patrick Haughey (PH): You are very welcome back to TELUS International Studios. I'm your host, Patrick Haughey, and this is a particularly exciting episode of the show because we are taking a look under the bonnet of one of the most interesting and fast growing companies in the world right now. That is Spotify. Most of us are familiar with Spotify's platform and how the company leverages digital technologies to deliver a really strong user experience. But what about the technology that Spotify applies to make the company itself run more efficiently to automate certain tasks and processes to ensure that Spotify team members are doing only the most high value tasks possible, all of which in turn feeds into an enhanced user experience. Well, that is what I find out all about in this episode of TELUS International Studios. My guest is Sidney Madison Prescott, Spotify's global head of intelligent automation. Sidney is such a fascinating person. The road she took to the world of technology was very different to what you might expect, and we hear all about that story. She has also had some very positive experiences with mentors throughout her career, and she explains why. And Sidney also shares her thoughts on the future of technologies like AI, machine learning, automation, quantum computing and lots more. If you have been following the series so far, and like what you hear, please hit follow on Spotify or subscribe on Apple Podcasts and leave a rating or review. So let's meet our guest.
PH: Sidney Madison Prescott, Global Head of Intelligent Automation at Spotify Welcome to TELUS International Studios.
Sidney Madison Prescott (SMP): Thank you. I am excited to be here today.
PH: I'm looking forward to hearing all about how you got into technology, and I know you're an incredibly passionate advocate for technology, but I know it wasn't what you started at, and I want to find out how you got into technology from where you started. But first of all, let's just start off where you are today. I'm fascinated by your title, as I'm sure many people are. What does the global head of intelligent automation at Spotify do?
SMP: Great question. So I am leading a team of engineers and also solution architects on a global scale to build out an intelligent automation footprint at Spotify. And what that entails is looking at several different technologies. So we leverage robotic process automation, machine learning and artificial intelligence in order to build efficiencies into the - all of the different business processes that we do today within Spotify. And so my team sits within the financial engineering organization, but we have a reach to all of the different business teams at Spotify, and we're focused on looking at what are all of the very manual, repetitive, tedious tasks that we do as Spotify-ers and how can we automate those tasks to drive greater efficiency, greater data quality and really ensure that our employees can focus on more value added task at the end of the day as a whole around the organization?
PH: Ok, so this is basically streamlining and being very smart about how Spotify is run as an organization as opposed to anything to do with the platform itself that we would pick up and use.
SMP: Absolutely. So it is from an organizational standpoint, how do we streamline and make the, I'll say, the back end of the work that we do as efficient as possible, but in that we are also enabling faster time to market and things of that nature which do come out on the application side.
PH: Absolutely. Maybe to just sort of illustrate everything that you just said there, is there an example you could give us about your job at work? Is there something - Is there something that people were doing in Spotify that you recognise that there was an opportunity for automation here, giving them higher value tasks in that case. Do you have an example?
SMP: I do, yes. So we are working today as a team with our ad operations employees. And what we're really looking at is how can we take the task that the team does today and how can we make those more efficient? One example is when we go into the Spotify application and we look at all of our different audio ads that run within the application, we have to test those ads to ensure that they - the integrity, the audio integrity is there. And so this is a process where we take each ad and we test to make sure the audio, the visual that everything is up to spec for that particular advertisement. This is a manual process, but we have been able to automate this process and leverage robots to actually check each advertisement to make sure that it's running properly, that it is showing the right or the correct specs in terms of the visuals and that the audio levels are appropriate. So this is a really perfect example of something that you wouldn't think that humans are doing today, but in reality, it's a very integral part of the responsibilities of the team.
PH: Yeah, very interesting. And it must be hugely fulfilling to be able to take a sort of a wide view of everything that everybody in Spotify is doing and spot the opportunities to actually make it smarter I guess.
SMP: It is. It's really fascinating to weave together all of the different business processes that really make up how we do work at Spotify and how we deliver on our product. And so one of the greatest aspects of this role is getting a high level overview of all of the different business units, their responsibilities and how those are interconnected and then taking a look at ways that we can optimise the processes of those teams, both from an upstream level and also downstream to really make again efficiency the key, I would say, value proposition of the deliverables for my team.
PH: Ok. In a couple of moments, I'd like to dive a little bit more into some of the technology that you use so the technology at play in your role because it's so fascinating. But first of all, tell me how you got into technology. I know you didn't start in tech and I think, am I right, it was an internship that changed everything for you.
SMP: It did, yes. So I started out as a philosophy major when I was in college and I was looking to go to law school and I happened to have a very fortuitous moment where one of my peers and in university pointed out to me an opportunity for an internship. And that is kind of, I would say, where the tide turned. I took an internship within configuration and asset management, and I quickly was fascinated by all of the different nuances of the role and what it means to work as a technologist on a global scale. And within that role, it was a really wonderful learning experience because I was able to help facilitate work for different teams and really again understand how those teams work together at an enterprise level to really produce deliverables. And that is where I really fell in love with technology and also with really the processes that we do to really innovate and redefine how we can deliver value at a firm level.
PH: So also in that internship in that firm, were there certain types of techs or at play with certain platforms that were being used to achieve the goals?
SMP: Yes. So I was working specifically with disparate databases and really looking at what is the integrity of the data within these specific databases. And then more importantly, how are these databases communicating with one another? And can we trust that the way that the data is extracted, transformed, loaded, can we trust that that is accurate, that it's timely? And so one of the things facilitating this work that I quickly began to realise is that we had a lot of systems that were supposed to be integrated together and it was assumed that all of the timing of those aforementioned aspects were correct and they really were not. So we went about a really, first a fact finding mission and then really from understanding the attributes that we should have been collecting within those different systems and understanding the ways that the timing was off. We then started to think through, OK, how do we really facilitate a better connection? How do we facilitate better integrations between these different systems? And so that was really interesting to me because it taught me a lot about the assumptions that we have as technologists about pre-existing systems or systems that are part of the enterprise before we particularly work for a company. And it also taught me a lot about change management and about really thinking through how you can bring others along on this kind of journey of digital transformation within a firm.
PH: And I know one of your mentors through your career. I think you pointed out in saying that he taught you how to tell a compelling story about your team's work, and we might come back to them a few minutes because I think no matter what job industry role you were in being able to tell a compelling story about what you were doing, what your team is doing is such a valuable quality. So we'll come back to that in a couple of moments time. So from that internship, then you had sort of found your path. So where did you go from there?
SMP: So from there, I worked really focusing on data quality and the integrity of data, and it gave me a passion that I still have today in relation to data visualizations. How do we create data that will be a valuable asset to our business stakeholders as technologists? And fundamentally reminding ourselves that at the end of the day, we really need to be able to have tangible evidence of the value that we're creating with all of our different automated solutions. And so as I honed that perspective within data quality and governance, I had an opportunity to then complete a proof of concept on a very new technology, which was robotic process automation. And that is where I began to get into intelligent automation. So through running that proof of concept, I learned about robotic process automation, I began to learn about all of the different tools that make up intelligent automation as an industry. And I also specked out the tools that existed already around the world in relation to driving efficiencies, driving productivity within teams. And that really led me into my current role, really standing up and scaling out and maturing intelligent automation centres of excellence.
PH: Well, one of the words that come up and really define what you've done is this word intelligence and intelligent automation. How do you define intelligence? What does it mean in this context, in your own view?
SMP: Great question. So there are a lot of different definitions for intelligence. From my perspective, in terms of the team that we have today and the technology that we are leveraging, we really are looking at the distinction between: is a particular automation cognitive, in that it can facilitate different ways of learning quote-unquote? And is it able to recognise different characters? Does it have spatial recognition, things of that nature. Now there's a separate part of the intelligent automation piece, which is also the robotic process automation. And those robots are not cognitive. Meaning that they cannot execute anything that has not been predefined in their developer build workflows, and so where intelligent automation really comes together is we start to think about the robots that can only execute on their predefined workflows and combining those with artificial intelligence or with machine learning engines and that combination of tool sets and different ways, ways of basically computing. That is how we get to the intelligent automation piece. So it really is the amalgamation of these different tools coming together to be able to facilitate a specific set of business outcomes, which we would typically relegate to, let's say, human intelligence rather than a machine.
PH: Well, one of the questions that I know it kind of comes up and when we talk with this field is when it comes to machine learning algorithms, are they really learning or are they just memorizing? And I guess, can we tell the difference between the two, really?
SMP: I think that's a great question. So if we think about it, it really comes down to how you really break out the different categories of kind of machine learning, deep learning, and artificial intelligence. So if you think about machine learning, it really is a subset of artificial intelligence. And so it's basically creating systems that can have the ability to learn and improve over time with experience. Now, machine learning models need a large amount of data very, very much like the human brain. We need a lot of data continuously over a lifetime in order to build up our understanding of the nuances of, say, a chair or the concept of education, the concept of computers. And so machines are very much the same. We are able to basically build up their understanding of a particular concept through the amount of data and really also curating the ability for the machine to curate that data is very important. I think one of the other things I would say is in terms of learning versus memorization. I believe it's a bit of both. And we do this as humans as well whereby we learn quite often by memorizing. And it's very interesting, even if we think that we aren't memorizing something, we are. So, for example, memorizing where a particular Starbucks is located on your corner or, you know, memorizing the way to get to a specific location - that is a form of learning. And I believe that machine learning can basically facilitate the same levels of understanding, whereby it's a component of a little bit of memorization, but that memorization actually enables the machine to then learn different patterns, learn different potential outcomes that will help to solve problems. And so the last part of that is when we start moving into another subset of machine learning, which is really your deep learning. And that's really where we are attempting to create a very similar computer network that mimics the neural networks of humans. And so that deep learning, this is where we get into the potential ability to delve deeper into a little bit less memorization and a little bit more learning of patterns again from a wide variety of data that is fed into that particular neural learning network.
PH: It's such a fascinating area, and I know I said a few minutes ago, you kind of you're now very far from where you started out, but hearing you speak there - I do get the whole philosophy major thing because this the type of thinking, the structure of thinking, I would guess, is quite similar. There's patterns there.
SMP: There are, yes. One of the most fascinating things that I found out as I moved from being a philosopher and I would still consider myself a philosopher, but a philosopher slash technologist is really the focus on logic. It's very, very fascinating. So the logic aspect is extremely similar between the humanities and technology, specifically with the logic that we actually use to code. So we use if then statements, which is something that philosophers base a lot of their logic and argumentation off of. So there's a lot of similarities between the fields. And I think that is also why we have a lot of ethicists and philosophers that are moving closer into the world of technology.
PH: Yeah, absolutely. I guess they were the original technologists to in some ways, I guess way back when. I got a couple of questions in from the team who know a lot more about this stuff than I do, and I think you're going to appreciate some of the questions. So first one is; Most AI applications to date have focused on software products like image recognition or targeting advertising or content evaluation, and for that reason, AI's presence in the physical world remains limited. That being said, how far are we from being able to really change the way people do things such as self-driving cars, robots, construction, etc.?
SMP: That's a great question. I believe that we are still - I think that we're making strides, but I think that we are still a ways off and that is primarily because of the challenge that we face as humans when we start to attempt to build something in our image, so to speak. And we have this inherent problem, which is; we are limited by the fact that we don't know what we don't know, which is a kind of a logical conundrum, even when we think about the way that today human beings understand the brain, right? How it works. We do understand it to some extent. But when you talk to neurologists, you talk to doctors. You quite often will hear a disclaimer that you know, we don't really - we aren't exactly sure why, you know, x y z works like this. So we have an understanding, but we are very aware of the things that we currently don't know or we aren't sure of, but we do not have visibility into understanding what we don't - that we don't know what we don't actually know. And that makes sense. And when we start to think about that, that does start to define how we build out these self-driving cars, how we build out autonomous robots. I believe as we continue to evolve our knowledge level of understanding that we have, specifically about human neurological capabilities. And doing that through scientific research evaluation, I believe that that research will fundamentally benefit the innovations on the technology side. So I think that challenge is present where as our knowledge builds, we then build that into how we conceptualise these different tools. But it's still, it's ever growing. And the progress that we're going to make is that it's always going to be slightly offset by that. Those additional layers of understanding that we just don't have in relation to human cognition and how we can actually build a computer system that in essence mimics human aptitude for learning and adapting in a given environment.
PH: Yeah, it's such a fascinating time. Another question is it a problem that the largest players in AI are primarily software and advertising companies, rather than companies that manufacture things? Or are these tasks inherently harder?
SMP: No, I do not believe it's a problem. I actually believe we need synergy between both those kinds of, I'll say, software companies and the manufacturing firms. I believe that synergy is what's going to allow us to achieve the next iteration of artificial intelligence improvements. And when we think about it, it's been quite a conversation. I would say over the last few years about the importance of every company being at the forefront of technology, and almost every company must be a technology company if they hope to remain an industry leader in their given area. And so part of that, I think, is whether it's a software company, whether it's a manufacturing company, how do we actually bring those two together because they're both sides of a different coin? They are both attempting to solve complex business problems through the use of different innovative technologies. And so I believe it's less of a question of which tasks are harder, whether it's software or manufacturing and more of a question of what is the best combination of the two that can help us successfully remediate business problems.
PH: Yeah, that's a really good answer, actually. What work is being done to ensure that AI is more inclusive of things like different accents and dialects and cultures?
SMP: There is a lot of work that's being done on a technological feasibility and execution standpoint. But I believe that there is a fundamental change outside of that enablement that we need to do, and that is really continuing to drive the diversification in this particular area of technology. When we drive diversity in technology, so diversity of age, ethnicity, race, gender, sexual orientation, et cetera, having those voices and those experiences from that diverse set of engineers in the room when the decisions are being made as to what is going to be, let's say, the default language for a particular application or the default, maybe graphics skin tone setting for a particular application. Having a diverse set of engineers in that conversation is the way I truly believe we can ensure that all demographics are included. So by putting a diverse cohort of individuals into the design process, the software development lifecycle, we can help better enable that inclusivity that we are looking for in terms of the ways that we design out these artificial intelligent systems.
PH: Yeah, just throughout your own - you know, we started kind of by talking about, you're here today because of an internship, really, that was your sort of, the moment that brought you down this path. And of course, with internships comes mentorship. What impact has mentorship had on your career and I guess particularly as a woman in tech?
SMP: This is a question that I'm asked a lot, and I love this question because mentorship, I believe, has given me such a great understanding of the unspoken nuances of the corporate world. And then even more specifically, the unspoken nuances of the technology sector. And so making the transition from a university and more academically focused career over to technology and senior leadership. I did not have that understanding, that built up understanding of those unspoken nuances of how to navigate the space and to your point, specifically as a woman, as a woman of color. And so mentors, my mentors really helped me, I would say fast track my understanding of all of the smaller but extremely valuable points of reference that I did not have as a, you know, someone new in their career, just starting out in a particular industry. And also, that mentorship helped me to build up my confidence, which I think is absolutely essential as a woman in technology, having a level of confidence about not only the knowledge, but also your ability to think critically about different scenarios, whether it's from a leadership aspect or development aspect, I think is absolutely critical to success and to the ability to retain not only employment, but also to move up the corporate ladder to move into the C-suite. So I think that a focus on having a mentor really does help you from a confidence perspective. It's helped me quite often in my career to have my mentors as a sounding board about different business challenges that I was facing, or even just management challenges that I was facing in relation to being a people manager within the technology sector, which is a very nuanced area to be in. So I believe that mentorship helps us in so many different ways, particularly if you are within an underrepresented demographic within the technology sector.
PH: And are you a mentor to anyone or, in the future, is it important to you to become somebody's mentor?
SMP: Oh, yes, absolutely. So I have, I've been a mentor to several new, I would say, young professionals who are just starting out in the technology sector and it is something I'm very passionate about because I do - I would say quite a bit of my success is built on that foundation of the fact that I was a mentee as I moved through my career. So yes, I do mentor and I do plan on continuing to mentor and continuing to speak about the relevance of mentorship specifically for women in STEM fields.
PH: The way you speak about mentorship and your team, etc. I think, you know, you come across like a great team leader and one of those things about being a great team leader is being able to, as we mentioned earlier, tell a compelling story about the work that your team is doing. So, any tips for us? It's a great skill to have.
SMP: Yes. So telling a compelling story is also, I believe, a key to the success that I've had thus far in my career. And it really starts with clearly defining the business problem you are trying to solve. Clearly defining the stakeholders that are that will be involved will be key players in the way that you actually transition a particular problem into a solution. And then also really identifying what are the ways that we can define success. And these should be tangible metrics that you can measure and that you can present to your senior leaders, to your executive sponsorship in relation to the work that your team is doing. So a lot of the time I find as technologists, we put our heads down and we come up with an incredible solution and quite often it's lost in translation the value that that solution provides back to the business, so it's very important to define out the ways that you can communicate what the value your team is delivering and also really to find out how you can better serve your stakeholders. This is a big piece as well, bringing the stakeholder along on the journey rather than having kind of heads down and so focused on the technological solution, building that empathy for your business stakeholder in relation to the pain point that you are attempting to automate - that goes a long way also in building that compelling story because your stakeholder then has a key, you have their buy in and they become really an advocate for the work that you and your team are doing.
PH: No, excellent, yeah. Excellent answer and excellent advice. And final question, and thank you so much for the time and you've given us and the story you've told expertly. The final question when I have somebody as knowledgeable and as passionate about technology as yourself on the podcast, we do like to look into a little bit of what's exciting you about the future of technology. One of the questions that did come in that maybe you'd like to touch on finally is your thoughts on quantum computers. Next generation technologies like quantum computers. Is this one of the big things that you're watching at the moment and excited about?**
SMP: Yes, but there are some challenges there. So for anyone who's listening just in terms of quantum computers, it's really you can think of it as a machine that is relying on the properties of quantum physics. And this is specifically in relation to the way that the computer stores data and how it performs from a computational perspective. And so the way that quantum computers perform is drastically accelerated by the fact that it really is leveraging more of a physics based approach to the storage of data, right, and performance capacity. And we've been starting to become more excited about this because what it means is that we can basically have more computing power on surface areas that are as small as, let's say, an atom, which we would never be able to do with traditional microchips. Now the challenge we have very powerful quantum computers today, and we have quantum computers that have increased processing capacity, but they also have very high error rates. And so there is a fundamental challenge that we're so facing, which is OK, we've drastically increased the processing capacity. We've drastically increased our ability to store data and to actually have computers running on surfaces that are just undetectable to the human eye. But the outputs are not reliable or I'll say they're not as reliable as we'd like them to be. And so the question then becomes, how do we continuously improve these computers so that they can become more reliable? And then at that point, I do believe there would be even greater excitement about the potential of quantum computers throughout our lives, just in terms of how we incorporate them, even from a manufacturing perspective.
PH: Yeah. As they say, watch this space. Sidney Madison Prescott, global head of intelligent automation at Spotify. It's been really fascinating speaking with you. Thank you so much for giving us the time to chat to us here on TELUS international studios today.**
SMP: Absolutely. It was truly a pleasure.
PH: What an interesting person to have on this episode of TELUS International Studios, I hope you enjoyed that conversation as much as I did. Thanks to Sidney Madison Prescott for taking the time to join us today. Thanks to you for listening. As ever, we will be back very soon with another episode of TELUS International Studios. Please do check out previous episodes. There are some real gems in there. And again, if you are enjoying the series, if you like what you hear. Please hit Follow on Spotify or subscribe on Apple Podcasts. I hope you can join us next time.