HBR IdeaCast: How AI Can Improve How We Work

Paul Daugherty and James Wilson, senior technology leaders at Accenture, argue that robots and smarter computers aren’t coming for our jobs. They talk about companies that are already giving employees access to artificial intelligence to strengthen their skills. They also give examples of new roles for people in an AI workplace. Daugherty and Wilson are the authors of the new book Human + Machine: Reimagining Work in the Age of AI.

SARAH GREEN CARMICHAEL: Welcome to the HBR IdeaCast, from Harvard Business Review. I’m Sarah Green Carmichael.

By now, I’m used to the idea that machines are going to be an ever more present part of work.

But if I try to imagine what those machines will be doing or what they’re gonna look like, that’s when the picture starts to get a little fuzzy.

Our guests today say machines will be doing all sorts of functions. And they stress that if we humans develop and deploy AI responsibly, the technology will take us to new levels of productivity.

Paul Daugherty is Accenture’s chief technology & innovation officer. James Wilson is the consulting firm’s managing director of IT and business research.

Together they’re the authors of the book Human + Machine: Reimagining Work in the Age of AI. And they’re here to talk with us about the impact this emerging technology is having on people and organizations, and the roles and skills we will all need in the future.

Thank you both for taking the time.

JAMES WILSON: It’s great to be here. Thanks.

PAUL DAUGHERTY: Great to be here, Sarah,

SARAH GREEN CARMICHAEL: So, there’s been a lot of, oh my God, robots are coming for your job hype in the media about the future of human and machine collaboration. Do you buy that hype?

PAUL DAUGHERTY: No, you know, I think — if you look at it, it’s the 50th anniversary of 2001: A Space Odyssey 50 years ago, last week, and that set in motion, this whole narrative about, you know, machines taking over the human race. And the other narrative that started is, you know, the machines are coming from our jobs, and then we had this whole thing about, you know, the machine, the machine is beating us at chess and checkers and Go and all these games, and we really think that’s all misplaced. And, of course, technology always does some things better than people can do. That’s what technology is. And that’s been the history of technology. But we wrote the book about is the fact that really, you know, AI, artificial intelligence, you know, robotics and the machines we’re talking about, like any other technology, really helps us as people, as humans, do things more effectively. And hence the title, Human + Machine.

SARAH GREEN CARMICHAEL: Yeah. So, one of the things I think that worries people is that, you know, in the first stage of companies adopting some of these technologies, they do use them to replace some people. But why, in your view, does a company that does that, that uses machines to replace people, eventually stall out?

JAMES WILSON: Well, in our research, we have really seen that there is this early-stage shift from a automation focused with artificial intelligence to an imagination and re-imagination focus. And we’re seeing that companies that focus on imagination and re-imagination are able to do a lot more and to do things differently than the companies that are just focusing on automating the old ways of doing things. So, if you think about typical process design, there might be 12 steps in a process, and if you come in and you say, all right, we’re gonna automate six of these 12 steps, you’re basically all you’re doing is speeding up an old way of doing things. You’re putting a new catalytic converter into a Model T, for example, as opposed to really rethinking your way of getting around.

SARAH GREEN CARMICHAEL: So, what have you figured out about which tasks people do best and which kinds of things machines do best?

PAUL DAUGHERTY: Yeah, people are good at emotive capability, communication, improvisation, generalization, things like that. And the machines are good at, you know, memorization, transactions, prediction, and one thing we don’t think has been looked at enough that we — was really the core focus of our research was what happens when you combine those two, and what’s, what’s that middle ground of collaboration? So, in the book, you know, in our work, we actually talked about collaborative intelligence, which is what happens when you put the strengths of the human together with the strength of the machine, and that’s where you found these, these new categories of jobs being created that we call the missing middle because, you know, missing because there really hasn’t been as much focus on it. We’ve seen this for binary focus of how can I replace people with the machines rather than thinking about how can I make people more effective and productive? And really, give people superpowers using this new technology to do their jobs much more effectively or as citizens or consumers, you know, live, live our lives more effectively.

JAMES WILSON: One thing that we have seen is that leading companies are really setting a precedent for creating unprecedented new types of jobs. So, we’re, we’re starting to see the emergence of new job categories that we haven’t seen before in this missing middle, in between kind of the human side and the machine side of work.

SARAH GREEN CARMICHAEL: Yeah. So, give us an example of what some of those jobs might be and how they might be shaped by this human plus machine collaboration.

JAMES WILSON: Well, there are two kind of big buckets. One bucket of jobs are where people help build and manage smart machines, and the other bucket are where people are helped by machines. And within the bucket where people are building machines, we see kind of three clusters of jobs that companies can kind of predictably put into the organization and really need to start thinking about today. One of those jobs is called a trainer, another, an explainer, and the third category that we talk about are called sustainers. But within that trainer category, there are interesting new types of jobs that we’re seeing emerging like personality trainers who use natural language processing and work to build intelligent agents and chatbots. And those don’t necessarily require a background in software engineering. It might require, you know, having a background in psychology or background in drama.

That second category that we talk about is called an explainer. And these are roles that where people explain why machines are doing what they do. So, for instance, at one organization, uh, we interviewed a guy who’s basically an AI detective. And so, when his company’s pricing model starts doing things that are unexpected, he has to go and explain to colleagues why it’s behaving in a certain way. We’re going to see a huge need for this explainer role for any company that’s operating in Europe these days with the GDPR that’s coming into effect quite soon. I saw one piece of research that estimated companies, global companies, are going to need about 75,000 new data compliance officers in their organization to explain to customers algorithmic decisions, say, if a customer calls in and say in a bank and that sort of thing.

And then the third category are what we call sustainers, and sustainers really manage the tradeoffs between what’s good for the business and what’s good for society and really use a kind of an ethical and responsible AI lens when making decisions. So, one type of role that we’ve seen there are called AI safety engineers. One of the things they think about are unintended consequences. You know, so what happens if this robot is hacked, you know, even though it’s a consumer robot, but if it’s a robot, you know, if it’s being used at say an airport or for industrial reasons and that sort of thing, what happens if the robot hacks itself for some reason trying to get a more efficient result.

SARAH GREEN CARMICHAEL: So, in this world where we are working with more different kinds of machines and algorithms and bots that can do certain tasks very easily, what are some of the skills you think that people will need to develop either to sort of remain employable or to get more out of these machines to make sure that they are really using them to give them superpowers?

JAMES WILSON: Well, I think that the roles that and the people that are most vulnerable to displacement from artificial intelligence are the ones that aren’t using artificial intelligence. So, how do you use AI as quickly as possible on the job? How can you start learning AI tomorrow on the job? So, I think there are two that executives really need to think about. Uh, the first might be a bit more obvious: doubling down on training. And it might be obvious, but our research really shows that companies still aren’t investing in retraining and reskilling at the level that they need to be.

The other thing that executives need to start to think about is really lowering the barrier to using AI. I think a lot of people think AI is rocket science, and in fact in a lot of ways it is. It requires really high-end math or stats skills, so the, you know the question facing a lot of workers today is, well, yeah, I’d like to be able to use it, but how do I use it? There are a lot of companies these days that are beginning to say, well, let’s democratize AI. Let’s, let’s make it as easy for a salesperson to use as their Excel spreadsheet or as PowerPoint. A one company that we’ve looked at, for instance that’s doing this is AT&T, and they’re putting point-and-click types of AI tools onto the desktops of about 50,000 people. So, if I’m in a call center, if I’m a salesperson opening up an account, I can interact with AI, can upload a data set, link it to natural language processing, just point-and-click approaches and kind of give myself superpowers in that customer interaction in ways that weren’t possible before. But the key point there is that happened as a result of executive intervention.

PAUL DAUGHERTY: Broadly speaking, when you think about the skills that people need, I think there’s two broad categories.

One is the skills you need for the people who will do AI. Those are, you know, the machine learning experts, data experts in a lot of STEM technology, coding types of experts and clearly we need more of those and there’s a lot of focus on building those skills, and a lot of companies are focusing on acquiring those skills. But relatively speaking that’s a lot smaller number than the other category, which is the people that use AI which will be almost every profession, and I when we look at that category of people who need to use AI to do their job, to do these new missing middle jobs that we talk about, that’s where I think we need to think about skills. And Jim talked about some good examples there. And in that category I think we need to focus on the one hand on building, building the more human-like skills. Because you know, AI technology will continue to do the things that machines are good at. That’s just been the history of how we apply technology to do anything to automate anything that, uh, that we can through throughout civilization. So, what we really need to focus on is what human skills can we really accentuate using the technology.

Also then focused on getting people more facile and more familiar and more comfortable with using technology. So, the last chapter of our book, we talk about eight fusion skills, which are the new fused, you know, human plus AI capabilities. For example, one is called a judgment integration, which is how do you make a decision combining your human judgment with the judgment from an algorithm? And you have to think about things a little differently. You have to apply and use the technology differently. And an example of this being used as some new wealth management approaches that the banks are using using AI to, to give agents better tools to recommend products to their customers and better make judgments about how to advise their customers. And that’s the kind of fusion scale that we think is important going forward. But it’s going to be a combination of that human skill and the technology or the AI skill. You know, one of the predictions that I would make is that when we look several years out right now, we bemoan the shortage of STEM skills and decoding skills and the AI skills.

I think, you know, somewhere around five years out in the future, I think we’ll be talking about the lack of more human-oriented skulls, humanities types of skills that can design the experiences and manage the experiences that we’re creating using AI as we reimagine businesses, as we reshape products to use technology to interact with humans in a more human-like way, we’re going to find there’s a dramatic need for many, many more professionals who can bring in that kind of ability to shape our human experiences using technology. And those will be some of the softer skills applied rather than a hardcore, you know, AI or tech or coding skills.

SARAH GREEN CARMICHAEL: You know, often in these conversations we’re focused on customer interactions, analyzing customer data, or managing customer relationships. But in the book, you have some examples of how companies are also using some of these tools to manage their own internal processes. And I wanted to ask you about how Unilever for example, is using AI in hiring.

JAMES WILSON: I think we usually think about the use of advanced analytics in HR in terms of crunching through resumes, but what we’re actually seeing is that AI is moving into the interviewing process as well, and that’s really transforming recruiter’s ability to interview more candidates. At Unilever, for instance, they’ve incorporated AI into kind of the first two rounds of interviews. The first round they play online games, and that gives the company a sense of the skill set of the person, kind of their behaviors. Maybe there’s a job posting that maps to their skill set that the person wasn’t aware of. The second round of interview actually is done with a video analysis system that can evaluate the person’s comfort level with certain types of questions, their gestures, their facial cues, and that sort of thing. So, it really allows the company to talk to three or four more times candidates than before.

And then, by the way, that third and final round of interview, the candidates are actually talking to the human recruiters. But by using that approach, the company has been able to expand diversity fourfold. They’ve been able to get candidates into interviews from many more universities, I think about three or four times more universities. And maybe most importantly they’ve been able to drastically reduce that cumbersome recruiting process that can drag on for months. At Unilever, I think before they brought in this solution, it was about four months long from that first interview to the final decision. They’ve been able to reduce that to about four weeks. So that’s a huge improvement both for the company but also for the candidate and their experience interacting with the company.

PAUL DAUGHERTY: Yeah. There’s another example that I would give in the HR area is something that we’re doing within our own company, within Accenture. We have over 400,000 employees, and we’re using AI in a creative way, still at an experimental level, but we’re using machine learning and AI to, based on, you know, a person’s profile, to understand their current job, experience, their resume, what they’ve done, their assignments. And based on the changes in technology, it’ll learn and, uh, recommend to how soon that individual might need to change their profession because what they’re doing now may become obsolete. So, in other words, how long will your skills will be relevant. But then it goes further and says, OK, based on what, you know, what should you start learning that based on what you know, that’ll help you be relevant and effective as you know, as technology changes. So, that’s still a little bit of an experiment, but I think it points the way to how we can use AI itself to help with this, you know, jobs and reskilling issue that we’re talking about earlier.

SARAH GREEN CARMICHAEL: In your view, how many industries will be affected by this? Because occasionally you’ll hear, everything’s going to be affected. This will affect everyone. And then other times people are like, no, really won’t. What’s your view on that?

JAMES WILSON: Well, you know, we don’t have a stock answer for that, and I think you know, uh, it’s going to vary by region, by country. France, for example, is going to focus on healthcare, mobility, and transportation. I think in regulated industries there’s going to be a bit slower adoption rate. Though certain types of algorithms, you know, high-speed trading algorithms and that sort of thing have been around for a long time in Wall Street. But when you start using a more sophisticated deep learning techniques in situations where you wouldn’t be able to explain to the customer or to the institutional investor why a certain transaction was made — those might take quite a while to be adopted if they’re adopted it at all.

PAUL DAUGHERTY: On the bigger landscape, we do believe that AI is right now really what I would say is the alpha trend driving other trends that we see in the market, and it will impact every industry. And we just talked about some specific industries that it’s going to impact more quickly, and I think it’s, it’s those industries that are very data intensive, those industries that have a lot of human interaction, those industries that have a lot of compliance and regulatory implications, we’re finding a lot of applicability. Those industries that have a lot of supply chain logistical components to them because those are the kinds of problems that we’re, we’re finding AI can solve very well. And then broadly speaking, health generally, there’s so much opportunity to improve wellness and health outcomes by applying AI that that generally speaking will be a big benefit that we’ll continue to see as AI is applied.

SARAH GREEN CARMICHAEL: Well, listen, it’s been really fun talking with you guys today about all of this. Thank you both so much for sharing your time with us.

PAUL DAUGHERTY: Thank you, Sarah.

JAMES WILSON: Yeah, thanks, Sarah. It’s been a great conversation.

SARAH GREEN CARMICHAEL That’s James Wilson and Paul Doherty. They’re the authors of Human + Machine: Reimagining Work in the Age of AI. You can find it at HBR.org.

Thanks for listening to the HBR IdeaCast. I’m Sarah Green Carmichael.

Copyright 2018 H. James Wilson | All Rights Reserved.