Artificial Intelligence — Opportunities Ahead

Narrator

The future of project management is changing fast. On Projectified with PMI, we'll help you stay ahead of the trends as we talk about what that means for the industry and for everyone involved.

Stephen W. Maye

I'm Stephen W. Maye for Projectifed with PMI. For an easy way to stay up to date on Projectified with PMI, go to iTunes, Stitcher, Google Play Music and PMI.org/podcast. In this episode we meet Michael Chui at the McKinsey Global Institute. Michael shares research on the impact that Artificial Intelligence will have on the workforce, where AI presents new opportunities in the project world, and how project professionals can prepare themselves to thrive in an emergent, exciting and uncertain landscape. Michael, I've been very interested in your work. I've had a chance to read some things that you've written, I've watched videos that you've produced, so talks that you've given in other formats, and have been just fascinated by the research and by what you have taken from that, and really looking forward to talking with you today. So thank you for being here.

Michael Chui

Stephen, it's a great pleasure and thanks so much for the kind words.

Stephen W. Maye

Well they're all true. So let's jump right in. You know, you have done - obviously not alone, but we want to hear your perspective on this - you've been involved in fascinating research around automation, Artificial Intelligence, machine-learning, robotics, digitization and really that as a bundle, and how all of that affects work and the potential future of work. So if we're going to use as kind of a shorthand here 'the research', which I realize there's a lot of pieces, give us a quick overview of what the research was about or has been about, and what for you has been the biggest surprise coming out of that.

Michael Chui

Well I mean, I think a couple of things. First of all, thanks for acknowledging it. Everything that we do at McKinsey and the McKinsey Global Institute really is a team sport and so I'm representing the work of a terrific set of folks that we've done this research with. The first thing is, why are we looking at this at all? I think one of the things that we have observed as we've, you know, looked at the world - and I'm a technologist by training, a computer scientist, cognitive scientist - is some perhaps surprising advances in the technologies themselves around Artificial Intelligence and robotics. You know, as a former Artificial Intelligence researcher I would have thought just a few years ago that the task of creating a self-driving car would be years away, just a very difficult set of engineering challenges, etcetera. And yet here where I live in San Francisco, nowadays if I drive on the streets, you know, it's more likely than not you'll run across one of these self-driving cars in test mode. And so there's that, there's a set of things that have happened in the cognitive realm as well, you know, not only beating the world champion in chess two decades ago but now the world champion in Go, which is a game that's many times more complex and difficult than chess; but then the ability for machines to read lips better than deaf people do. And so again, to a certain extent the technology is surprising us. And then when you think about what the implications for these technologies are in the world of work, and the degree to which things that we thought only humans might be able to do effectively, now the machines are doing things as well or even better. You know, there's a natural series of questions about what does this mean for work? What does this mean for employment? You know, are we going to have mass unemployment going forward because the robots and AI will take over everyone's job? And so that was really the motivation for the research. So what did we find and how did we think about it? I think one of the things that we looked at doing as we looked at jobs, you know while others have studied jobs themselves, we actually thought that if you look at any individual occupation that's the wrong level of granularity to examine this phenomenon. Because everyone in their job, everyone in their occupation does multiple different activities, each of which has a different propensity for machines to automate. And so rather than looking only at 900 or so jobs that the US Bureau of Labor Statistics catalogs, we looked at all of the constituent activities, over 2000 of them in total. And for each of those we scored them against 18 different capabilities which potentially could be automated. So everything from cognitive tasks, recognizing known patterns or developing novel patterns, doing logical problem-solving; some physical things such as fine motor skills, gross motor skills, navigating the physical world; linguistic things, understanding when someone's talking, being able to process it, being able to respond in a natural language such as English; and even some social and emotional capabilities, the ability to read or recognize human emotion in another person, process that and then respond in an appropriate way. So against 18 of these different capabilities we tried to understand what level is necessary in order to accomplish all of the activities that we pay people to do in the economy. And then at the same time we looked at to what extent could technology actually accomplish different capabilities of those 18 different capabilities, and just compared. And you ask what are some of the surprising findings? Again, to a certain extent they're not surprising in retrospect but at least at a topline, if we just look at technical potential, what we found was about half of all the activities we pay people to do in the global workforce could be automated by adapting currently demonstrated technologies; not even requiring some sort of breakthrough in technology by some of these Artificial Intelligence and robotics researchers. Literally by adapting currently demonstrated technologies, half of the things we pay people to do in the global economy could be automated. Now that's a huge number. The other thing that we tried to understand was, well look if that's theoretically possible from a technology standpoint, you know, how long and how fast might that actually happen? And this is where we tried to understand, look what are the things that have to happen in order for technology to actually be adopted? And number one, we have to solve a bunch of technology problems. So while I said theoretically half of the activities could be automated by adapting currently demonstrated technologies, that adaptation mostly hasn't been done. So the technologists actually have to spend money and time integrating those technologies together and adapting them for individual activities. That takes some time. In many cases, you know, millions of dollars and years to do. So, you know, in order to get things to actually become automated and adopted, the technology problems have to be solved. You need to have a positive business case, so you need to compare the price of the technology versus the cost of human labor, and then also net in all of the other potential benefits of automation such as reduced variability, increase through-put, decreased errors, greater safety. And then even when you have a positive business case, even when that all nets out positively, there's a natural adoption curve, and we've looked at dozens of different technologies and across all those technologies, between the time of their commercial adoption and their eventual plateau in adoption, the time of commercial availability and the eventual plateau in adoption, takes in the neighborhood of eight to 28 years. And so when you net all of those things together, the point at which half of today's activities might be automated, the middle of all of our scenarios is 2055. And so that's quite some time. Now we modeled a scenario that's 20 years later and 20 years earlier, so again we're humble enough to know we can't predict the future and there's lots of variability here. But again, what we think is that things will happen somewhat slow in macro - that is, it takes a long time for half of the world's activities to be automated - but faster in micro in the sense that if you're the individual who is affected by these technologies, or if you're a company that has to compete on the basis of these technologies, this might happen quite quickly for you. If you think about business, the hard work is usually not the development of the technology itself or even the acquisition and procurement; it's all of the project management, to be honest, the change management that has to happen.

Stephen W. Maye

Yeah.

Michael Chui

And so that's what really is the long pole in the tent.

Stephen W. Maye

Interesting. So if you apply this to the project world, and let's go broad first to understanding that there are projects of some type in essentially every industry, but if you think about that idea of work defined in projects and fast forward a bit, so we're gonna allow for some of this adoption curve to have taken place, where do you see the significant differences in the project world?

Michael Chui

I think there are two broad categories of potential implications of the adoption of these technologies for people in the project management space. Number one, I think it will change the activities that a project manager does in terms of just the way that they execute their job, and then it will change the way that the projects that they manage also will get done. So let me just comment on each of those in turn. In terms of the way that projects get done, and as you mention there's a huge variety in terms of the actual projects that are being done across sectors and functions, etcetera, but there are three broad categories of activities which we found to have the highest susceptibility to technical automation. Number one is a little bit unsurprising, they are physical activities in predictable environments. You know, think about an assembly line in a factory for instance, that's a classic example. But there are a number of different other sorts of physical activities and predictable environments which I think will be increasingly automated. You know, this includes agriculture for instance, you know, what happens within a barn. So there's still a lot of opportunity there even though we sort of understand how this might work within a factory. But I think there are two other categories of types of activities which again have high susceptibility of automation and we'll find more and more of the projects that project managers are managing increasingly they'll need to manage not only the people but the machines that do these things, and they are number one: collecting information, and number two: processing information. I think two things are interesting about that: one is that they're very widespread that in many cases 30, 40, 50% of the activities that are being done might fit in these two categories, but also interestingly, while a lot of these types of activities are done by relatively low wage or front-line labor - so for instance, you know, whether or not it's processing a financial transaction or what have you - a lot of these activities are actually done by, in many cases, high wage, high skill people - whether they're attorneys, physicians, managers and executives themselves.

Stephen W. Maye

Yeah.

Michael Chui

And so I think what we're gonna find is, you know, the projects that are being managed, the mix of work is going to change over time. So less of those sorts of things and then more of the sorts of things which are harder to automate, at least currently, for machines to do, which are things like exhibiting creativity, exhibiting emotional intelligence, leading and managing and developing other people. So I think that's going to happen on terms of the projects that people manage. And then in terms of project managers themselves, again the same kinds of things. There's a lot of time the project managers spend collecting information and processing information [INTERRUPTION] and while we have some tools to do resource leveling and those sorts of things, I think we're going to start to see those things happen in steroids. I strongly suspect that a lot of the work, the actual time that some of your listeners are spending doing things like cutting and pasting from one system to another, I think some of the technologies we can buy off the shelf now make those activities much easier to do.

Stephen W. Maye

Yeah. And I think there's some good news in there. When we think about project professionals - and I'm allowing for this people that are in leadership roles in PMOs, people that are executives but leading significant initiatives or significant projects, all the way through to someone who's quite early in her career and perhaps in her first project management role. I think there's good news in what you're saying in that the trend within that space is that more and more emphasis is being placed on things like leadership, on things like managing change, on things like coaching and developing people and providing an environment where people can make decisions and flourish and do their best work versus something that's a kind of command and control, track and report. Those things have to be there in some form, but that doesn't feel to me like the future of project management.

Michael Chui

I think that sounds exactly right. And there's a glass half-empty view on this, which again if you look at the work that many people do an unfortunately small amount of it requires the ability to connect with people on an emotional level to motivate them, to exhibit leadership, to be creative. And so that's the glass half-empty. The glass half-full version is, but going forward maybe more and more of the work we do, more and more of the work that a project manager does will perhaps be these things that we think of as being more human, that are perhaps more interesting, exciting, creative and all those sorts of things. So I do think that there's definitely an optimistic story about this in that you can get the machines to do "the boring stuff" and perhaps the less interesting stuff, then people are freed up to do the things that are perhaps more engaging, more interesting.

Stephen W. Maye

Well I certainly prefer that version of the story, so for the moment let's go with that.

Michael Chui

Okay [LAUGHS], very good. Happy to do that.

Stephen W. Maye

So where do you anticipate the most significant impacts from automation - and I'm using that broadly. So Artificial Intelligence, machine-learning, digitization, other broad areas of automation including robotics, where do you expect to see the most significant impacts here in the near future?

Michael Chui

Well I mean, I want to preface my comments by saying the headline should be, 'In fact this is going to affect all of us'. You know, regardless of industry. I mean, we looked at all of the occupations in the economy; over 60% of the occupations we looked at had over 30% of their activities which had the potential to be automated by adapting currently demonstrated technologies. In other words, all of us are going to have our jobs change over time. So, you know, I think that's the number one headline which is that this isn't something that will just affect one or two industries. I think it's quite complicated, and again being able to be very granular about what individual activities are there, what the costs of automating them are and then what the business case is for doing it, you know, all that has to happen. And the other thing that I point out is, while our research was done on, you know, the sort of atomized view of this activity, that activity, in practice - and certainly in my firm when we've been helping clients, whether it's in healthcare or energy or in financial services - think about how to use these technologies to improve their performance. They typically don't just look at an individual activity and say, okay we can automate that activity; rather completely re-examining the entire process and how can you re-think or transform that process by using these technologies is really in practice what happens. And so that's part of the reason why it takes some time. It's a project indeed to [INTERRUPTION] create a new process. But that's [INTERRUPTION] really where the benefit comes.

Stephen W. Maye

Where do you think we are from a social acceptance perspective? And is that changing?

Michael Chui

Well I think it's a very complex question. First of all, different people have different interests or likelihood or trust in machines, etcetera, to do different tasks. And I think that even for an individual but particularly for a population that changes over time. Certainly right now most patients want to talk, particularly if it's an important discussion - whether it's a serious disease or a set of questions about what treatment should someone go through - most people, many people I would say, would feel more comfortable talking with a medical professional about that decision, and hopefully someone who's empathetic and understanding and can help people make difficult trade-offs. At the same time I've observed, certainly in other fields, you know over time we get much more comfortable with machines being helpful there. Even today, you know, these electronic health records, one of the things that the physician diagnostic assistant systems do is pop up a warning: do you know that this patient is also taking this other drug and what you just prescribed might create an interaction. That's very, very helpful and maybe the patient isn't even aware that that thing happens. But that's a bit like the computer being an external brain or helper for a physician. Now I think we'll see more and more of that. And certainly in the surgical realm, you know, we do see these 'robots' being used more and more, and in many ways they're quite helpful. And so I think over time we'll see more and more acceptance and interest. You know, medical decisions have extremely high stakes - it's not a perfect analogy but there was a time when people said, look if I'm going to go to a bank I want to talk to a person, look them in the eye, because this is important, this is a financial transaction. And now we have lots of people who would rather not talk to a teller, would rather either transact on their phone or an ATM, etcetera. And not because they don't like people but they just find that transaction to go faster, maybe more predictable and you could do it late at night and all those sorts of things. So I think the way that we view these machines changes over time based on familiarity, based on understanding the risks and just experience. And I think we're in the very midst of that when it comes to self-driving cars. Because we are starting to see them more and more on the roads, we're starting to see that as a feature on existing cars more and more and I think people likely will become more and more comfortable with that given more experience.

Stephen W. Maye

Do you expect to see significant differences in adoption across the country and across socioeconomic levels and so forth?

Michael Chui

Well let me contrast two things. You know, we talked about things that surprised me, I've been in some senses pleasantly surprised at the degree to which...if you asked truckers, for instance, there was a piece in the New York Times, for instance, about their views on self-driving trucks; I think the awareness is very, very high. If you ask a factory worker about robots, I think in many cases they're quite well aware of what's coming down the pike from a technology standpoint. You know, technologists and others will disagree about the pace at which it might happen and how effective these systems will be, but I don't think that only technologists understand what's coming. So I think at least certainly from an awareness standpoint I do think that in fact thanks to the proliferation of communications technologies, people sort of get what's coming. Now there might be quite differential impact based on, you know, when a technology comes into play. If it automates a factory worker's work or a truck driver's work, that might have a different implication in terms of just, first of all it's just extremely painful to lose one's job and very difficult in many cases to retrain to do other things. And I think those are some of the societal challenges that we're all going to have to grapple with going forward. Now as someone who lives in California in a sunny, optimistic place, we've seen this happen before, we've seen a double digit percentage of what in the US workforce is done change over decades. For example, 40% of the US workforce were involved in agriculture around 1900 and 70 years later somewhere in the neighborhood of 2%, so that's a huge change in less than a [INTERRUPTION] century.

Stephen W. Maye

Yeah, that's a massive change. Sure.

Michael Chui

So we've seen it happen before. My only other point there is that it doesn't happen by accident. I mean, during that time period we had a high school movement, we had a society, government and private sector come together and agree that we should have roughly universal secondary education, a high school basically. And so I think going forward, to have good outcomes we're going to have to make good decisions, but if we make good decisions I am optimistic. Even though it's not easy for people in many cases when this impacts them, but broadly speaking over time I'm optimistic that we'll be able to have people transition to other jobs as they're displaced from activities that they previously had to do.

Stephen W. Maye

Think about the project professionals again as we go back to this broad community of people that spend a significant part of their day moving projects forward. Whether they're at the executive level, whether they're in the PMO, whether they are young, upcoming project managers. How do they prepare for a world that is increasingly smart - and I mean that in the sense of smart technology - a world that is increasingly automated? How do they prepare for that? How do they prepare to continue to bring value in that future world?

Michael Chui

Yeah, I don't know if I have any secret ideas that no one has thought of before, but maybe a few principles that might be interesting. One is I think increasingly, and even more so than previously, a project management professional needs to be able to understand these technologies, understand the art of the possible, and try to stay at least abreast if not ahead of what these technologies can do; because they can both affect and improve the work that any professional does as well as the projects that they're trying to manage. So I think that's at least one thing that's very useful. Secondly, some of our other research has led us to the conclusion that the effective use of data and analytics is changing the game. Now I think being able to become more familiar with statistics, experimental design and these other sorts of disciplines where rather than just business judgment and experience, being able to apply numerical quantitative statistical analyses to decision-making will be very, very important. And then finally, if we go back to previously in the conversation, some of the skills, some of the capabilities which will be most difficult for the machines to do - whether it's motivating others, whether it's emotional intelligence, whether it's being creative - and then being flexible and agile. Because as I said, all of these technologies are going to affect all of us and we're going to change what we do day to day. And so how can we be resilient, how can we learn how to learn? I think all of those sorts of soft skills, mindset, capabilities which are in those realms are going to be increasingly important going forward and to be able to cultivate those, practice those - and in fact, in certain cases I think you can learn those over time - will also stand people in good stead as these technologies continue to wash over the work that we all do.

Stephen W. Maye

And I think that brings us back in some ways to what we talked about earlier, which is that the good news for this space, if we look at this broad space of project management and project leadership and managing large portfolios of projects and so forth, is that I believe what you're describing is consistent with where people in this space are focusing. I'm involved in a number of events throughout the year and I've watched that conversation over the last decade move more toward what you're describing; particularly when you start thinking about those kind of leadership skills and skills around judgment, and skills around motivation and so forth, then I think there's a lot of consistency - maybe not 100% - I think there's a lot of consistency with the direction that people within the industry are moving in their focus and what you're describing as what's likely to be required and what's likely to be the areas of skill and expertise that are gonna help people really prosper and flourish as they move forward.

Michael Chui

I think that makes a tremendous amount of sense. I mean, if I were to be most provocative, some of the pure skills involved in project management increasingly will be done by the machines: resource leveling, for instance, or [INTERRUPUTION] all that kind of stuff, the software we'll start to do. And then, as you said, the thing that will be most valuable and engaging is how do you motivate people, how do you lead a group, how do you get people to do the things that need to be done?

Stephen W. Maye

Well Michael, you've brought us something really fascinating today, I appreciate it. When we think about not only this audience, this community of people that we really speak to in terms of those that some significant part of their work week is dedicated to constructing and leading and guiding projects forward, but really the broad population, I think you're involved in something that is relevant to all of us and something we need to be thinking about, and is a fascinating, fascinating space. I'm sure a lot of people are a little envious of the work that you do. I'm sure that you get that at cocktail parties, right?

Michael Chui

Well I do feel lucky to be able to do what I get to do. So it's been a pleasure to talk with you and an honor to be able to be heard by your listeners.

Stephen W. Maye

Absolutely. The honor's ours. Thanks again Michael Chui, you've been a wonderful guest and we look forward to talking with you again.

Michael Chui

Well thank you. Thanks so much.

Stephen W. Maye

For an easy way to stay up to date on Projectified with PMI, go to iTunes, Stitcher, Google Play Music and PMI.org/podcast.