We believe that big data is good. Good for our economy; good for consumers and good for society.
In a recent Experian #DataTalk, we had a chance to talk with Tom Davenport about the future of work with A.I. and other smart technologies.
Here is a full transcript of the interview:
Mike Delgado: Hello, and welcome to Experian’s Weekly Data Talk, a show featuring some of the smartest people working in data science. Today we’re excited to feature Tom Davenport, who serves as the President’s Distinguished Professor of Information Technology in Management at Babson College. He’s also a fellow of the MIT Initiative on the digital economy, and senior advisor to Deloitte Analytics. He lectures about big data analytics at Harvard Business School, MIT Sloane School, and Boston University. Aside from his busy teaching schedule, and strategic advisory roles, Tom has written or edited 18 books about data science, and over 100 articles for Harvard Business Review, Sloan Management Review, and the Financial Times.
Today we’re talking about the future of work in the age of artificial intelligence and smart machines, which is a topic of his latest book called, “Only Humans Need Apply: Winners and Losers in the Age of Smart Machines.” Tom, it’s an honor to have you in our chat today. I thought it’d be helpful to start this discussion talking about how this new age of AI and smart machines is different from the previous technological advances in our history.
Tom Davenport: Sure, Mike. I just finished an update of my first book on analytics, which is called “Competing on Analytics.” It’s 10 years old, and I talk that we had a pretty stable period with analytics from the mid-70’s to the mid-2000’s or so, and then everything sort of exploded with big data, AI, not necessarily being created for the first time, but certainly coming back in a much bigger way than anybody had seen before. I think these last few years have been quite striking in terms of how much change there’s been in the technology, and in terms of what people are trying to do with data and kind of unfortunate. It made my book obsolete more than I had hoped.
Mike Delgado: I read through your book over this last week, and found it really helpful. It calmed me down a little bit because you have a very optimistic view of the future, as opposed to some of the doom and gloom that I’ve been hearing a lot about. You focus more on looking at smart machines and technology to augment our work, rather than replace our work and make us redundant. You talk a lot about how some of the tasks that we do can be automated, and that could be a good thing because it can help us focus on more important things, more strategic things. I was wondering if you could talk about maybe some examples of what you mean by machines augmenting our work, rather than making us redundant?
Tom Davenport: Well, since the book, I’ve been even more confirmed in my belief that augmentation is a more likely future, and obviously, a better future for us humans and more and more people seem to be kind of leaning in that direction. But some examples might be instead of e-discovery technology eliminating lawyers, there may be a few changes on the margins of people in the legal profession. Most of the negative impacts tend to occur at the entry level. I think for the most part, lawyers just end up using this technology like any other.
You see it with underwriters in insurance, which I end up talking about a lot in the book because it’s not the most exciting job, or the most exciting industry necessarily. But it was one of the earliest adopters of artificial intelligence in the form of rule based, expert systems. There were a few underwriters who lost their jobs, but for the most part, underwriters either started working directly alongside these machines as if they were colleagues, or went into kind of ancillary jobs related to underwriting like communicating with customers and agents about their insurance policies.
By the way, this is not just an AI phenomenon. We’ve been supposedly automating jobs for a long time. Spreadsheets came along and not too many accountants, I think, lost their jobs. They just ended up doing a lot more analysis with spreadsheets. We’re always very concerned about this issue of automation, but it never seems to happen in quite as dastardly a fashion as we might expect.
Mike Delgado: I thought it was interesting because whenever I’ve heard about automation, especially in the news, I hear about AI replacing a lot of tasks done by what’s referred to as blue collar work, not so much the strategic thought leader work. You talk about how AI is very much capable of taking over some of the smart thought leadership roles. Anything that could be codified.
Tom Davenport: Yeah. I think that’s one of the scarier things about it for a lot of people because we always thought that knowledge work was kind of immune.
Mike Delgado: Yeah.
Tom Davenport: That’s what we were encouraged to adopt when we moved upstream from farm and factory work, and then from service oriented jobs. The idea was get a lot of education and you’ll be fine. That’s not necessarily the case anymore.
Mike Delgado: I was going through your book, you talked about some of those roles. You mentioned lawyers, also architects. That was an area I thought was fascinating. When I think of architecture, I think of somebody who needs to be very, very good with design. So there’s an artistic side, as well as the mathematical side. Are you seeing other roles like architecture fall into that realm, that could also be made redundant, or some of those tasks be augmented?
Tom Davenport: Yeah. Again, with architecture, the impact was mostly at the entry level because a lot of those people don’t do creative design work. What they do is drafting, and drafting has been radically transformed by computer aided design software, and so on, so you need a lot fewer drafts people.
Yeah, I think there are a lot of different examples. Everybody from marketers. A lot of marketers these days are highly automated. Search engine optimization, A/B testing, deciding what digital ad goes where. All of that is pretty much automated now. A lot of journalism is increasingly automated. If you’re a small company, or a small sports team at a small college, don’t expect that your story is going to be written by a human being. Chances are good that it’s going to be written by machine.
A lot of back office financial tasks are increasingly being automated with this robotic process automation software. It’s not the smartest cognitive technology, but it’s quite effective at those back office digital tasks.
If your work is digital, if it’s fairly structured and repeatable, repeated over time, if you can do it remotely, I’d say chances are pretty good that it might be automated, which doesn’t mean you can’t find another higher up job, but it’s a little bit less structured. It should give you cause for concern.
Mike Delgado: I know that whenever I read articles around finances like stock articles maybe in Reuters, sometimes at the end of the article, it won’t be a person’s name, it’ll be that it was an automated article. Now there may have been a human editor to kind of just review it, but that type of content can be scalable, like you said. It can develop a lot more content, and maybe the people that are in charge of that content are more working as editors, so that they can optimize their time and produce a lot more.
Tom Davenport: Yeah, I don’t think there’s been a big loss of journalism jobs since these technologies came out. But on the other hand it’s not a growth sector so I think you need to switch to the less structured aspects of journalism, like investigative journalism, or writing about interesting human beings, or something like that. Not the fairly road stuff like sports results. Often stories with data are better written by machines in many cases because they have access to more data, and they do a good job of that, so business stories and sports tend to be the most data intensive.
Mike Delgado: Now Tom, over the weekend, I was chatting with some friends and we were talking about your book, and artificial intelligence, and the future of work, and one of my friends was saying that she thinks teaching is safe. I was thinking, well, I’ve heard of cases where artificial intelligence, or online education can be optimized for a particular learner, and so there could be the case made that an AI could work better with us, based on our strength, and our weaknesses. I was curious about your thoughts on the educator role, and do you think that some of those tasks can be augmented?
Tom Davenport: As a professor, myself, I think a lot about that. My son is a teacher, and we have discussions about it. I think that basic function of educators involving transmitting knowledge, I think a lot of that can be better done by machine because as you say, we have these adaptive learning technologies that will carefully monitor how well is the learner learning certain things, and repeating it, or going back for simpler stuff, or moving ahead for more complex stuff if they’re getting it well. Obviously, there are some educational tasks that can’t easily be automated.
I think communicating how people should behave, and how they should get along with other human beings and so on is probably going to be a function of teachers for a long time. Maybe discipline. It’s not the most exciting part but it’s hard to imagine you have a machine minding the class. It seems unlikely that kids would behave very well. I think it’ll be a mix, and it may mean somewhat fewer human teachers in the future, but not, I think, a radical change.
Mike Delgado: I was encouraged when in your book you were talking about some of these tasks like storytelling, being a narrator, to helping to tell business stories. Having data is helpful, and having those insights is helpful, but having somebody to be able to relay that data in a narrative is something that, at least right now, machines can’t do. That’s unique to humans. Can you talk a little bit about that, humans as storytellers, as narrators?
Tom Davenport: Yeah. I’ve always believed that telling stories with data is an important function of analytical people, data scientists, and so on. It’s conceivable that machines could venture into that territory a little bit because as we were talking about, they do write journalism stories, and so on. I think the key with any human is to focus in on what the receiver, the listener would be most interested in, and personalize it, identify something that you know they care about so that they get interested. With analytical people, the tendency is always to say, “First I did this logistic regression, and that didn’t work out terribly well, so then I tried another form of analysis, but we had a heteroscedasticity problem.” That’s not a good type of story. You must learn who are the characters, give only as much quantitative detail as necessary to get the message across.
Mike Delgado: I think where humans right now have an advantage is that we can read body language, read body cues, as where sharing that story with leadership, we can get a sense of, “Is this interesting to this person, or should I be revising my storytelling strategy?”
Tom Davenport: Yeah. Some of us are better at that than others, but in general, we do have that potential.
Mike Delgado: In your book, you go into great depth talking about these five key steps for working with smart machines. Some of those sections include stepping up, stepping aside, stepping in, and stepping narrowly, and stepping forward. You go into great length explaining those. Are there one or two of those steps that you think are most important for us to be thinking about right now in a workplace?
Tom Davenport: Yeah. I think that kind of the core augmentation role is stepping in where you say, I’m going to learn how these machines work, and I’m effectively going to be their colleague. What I was thinking about underwriters before applies to the stepping in role. I’m going to pick up the ball when they drop the ball. I’m going to maybe even understand how it works well enough to improve it a little bit over time. That, I think, is an important role that particularly quantitative people can master and can pretty much ensure their future employment.
The other extreme is probably stepping aside where you say, “I don’t really want to compete with a machine, or work with it, I want to do something that machines can’t do very well.” I think those are kind of the polar-opposites of those five roles, and a choice increasingly that people must make. Do I want to work with computers closely, or not?
Mike Delgado: I look at my kids when I think about the education system today. My daughter is 11, my son is eight, and right now there is no data science programs in their schooling, at least at this point. There are other organizations like Kaplan, that do specialize helping kids learn coding and helping kids with robotics. Do you foresee our educational system beginning to bring in data science to help educate our children for these future roles?
Tom Davenport: As you suggest, there’s certainly a lot of talk about having everybody learn to code, and there’s been for a long time this emphasis on STEM, Science and Technology, and Engineering and Math. The challenge is that those tend to be the things that computers are going to do well. It’s certainly useful to know how computers think, and what kind of logic they use, and so on, and you do learn that from coding, but I think it’s important to realize that there are a number of other skills that we need to emphasize an education like narrative as you were discussing before that don’t fall into those buckets that we think will guarantee a job.
That’s one of the reasons why you hear more and more people say, “Well, liberal arts seem to have a lot of those attributes of creativity, and communication, and so on,” so maybe we’ll see an upsurge in that. Mostly, I think it’s just important to, as I said, to decide early on, “Do I want to spend most of my life working with machines?” If so, I better learn how they work. Or let’s go completely the other direction and write poetry, or whatever. Making enough money to live off of can be a problem with some of those non-computational domains. But I think they won’t be replaced as quickly.
Mike Delgado: As you consider the future, how would you advise a C-Suite to prepare for smart machines for their organization?
Tom Davenport: Well, it’s interesting, I just did a survey with Deloitte and we asked managers who were aware of what was happening in their organization with cognitive technology. Somewhat to my surprise, about 60 percent said they were already starting to train their people in the skills they needed to work effectively with machines.
I do think that for many workers, there’s going to have to be some changes in roles and skills. As I say, some jobs will go away. Not a lot, but I think there will be a lot of change in the content of the work, and the sooner senior managers can start to say, “Well here’s some of the skills we think are going to be necessary for the future. Here’s some ways you might acquire some of those things.” I think it’ll be a much smoother transition for organizations and the individuals within them.
Mike Delgado: I believe I saw research that might have been Deloitte, or maybe Gartner. It said that by 2020 there’s an expectation that even managers will have some background with data science and understand it. Does that seem to correspond with what you’re saying?
Tom Davenport: Yeah. The unfortunate thing is you don’t see too many organizations training managers about that. We have a lot of programs and universities too now to create new analytics people and data scientists, and so on. We don’t have that much in the way of preparing managers. There’s some companies, Sysco Systems has done it, they’ve programmed to create, to educate managers in what makes for good data science, but you don’t hear very much about it. I think we need a lot more.
Mike Delgado: Just one last question for those people who are still in school about to graduate college. As they’re getting ready for this job market that is just changing drastically, what would be your advice to them as they’re entering the workplace?
Tom Davenport: Well, I think as we were saying, you need to learn a fair amount about how computers are related to whatever field you’re interested in. If you want to be an architect, know that a lot of the basic stuff is going to be done by machines, so you need to learn a lot about computer aided design, and so on. If you’re going into journalism, be aware that a lot of journalism is going to be done by machine, and learn what machines can and can’t do, so you can kind of adjust your skills accordingly. Unfortunately, universities don’t do a great job of this yet, so you’re going to have to do a lot of investigation on your own.
In the book, we talk about a guy who was a lawyer, kind of fell off the partner track in a law firm, was doing document reviews on a contract basis, and he said, “I can see this is going to be automated,” so he started to go to seminars. He read a book called “E-discovery for dummies,” which I thought he was kidding, but it does exist. He went to vendor seminars, and he made himself an e-discovery expert, and now he has a job, a good job in that space with a vendor. Unfortunately, I think it’s going to be on all of us to prepare ourselves for this new future.
Mike Delgado: Awesome. For those who are watching live, or are watching the pre-recorded version of this, I want to highly recommend Tom’s new book, “Only Humans Need Apply.” You can learn more about it by going to Tomdavenport.com, or of course, by going straight to Amazon. Highly, highly recommend it. It’s a great read. It paints a more positive, optimistic future for us than some of the doom and gloom news we’ve heard about robots taking over all our jobs. Tom, thank you so much for your time today. Where can everyone learn more about you and your work?
Tom Davenport: Well, some of my critical colleagues say I never had a thought I didn’t publish, so there’s a lot out there I’ve written on my Linkedin site, or Tomdavenport.com has a lot of it, so it’s hard to avoid, I’m afraid.
Mike Delgado: I’ll make sure in the bottom of this YouTube video they’ll be a link going over to Tomdavenport.com, as well as if you’re watching this later, on Facebook, we’ll have a link in the comments going over to Tom’s website. Also, if you’d like to learn more about upcoming data talks with different data scientists and thought leaders, you can always go to Experian.com/datatalk. Thank you again for your time today. Thank you for your books. Looking forward to learning more in your further research and writing.
Tom Davenport: Thank you, Mike. It was fun.
Mike Delgado: Thanks, Tom. Take care.
Tom Davenport is the President’s Distinguished Professor of Information Technology and Management at Babson College, the co-founder of the International Institute for Analytics, a Fellow of the MIT Initiative on the Digital Economy, and a Senior Advisor to Deloitte Analytics. He teaches analytics and big data in executive programs at Babson, Harvard Business School, MIT Sloan School, and Boston University. He pioneered the concept of “competing on analytics” with his best-selling 2006 Harvard Business Review article (and his 2007 book by the same name).
Professor Davenport has written or edited eighteen books and over 100 articles for Harvard Business Review, Sloan Management Review, the Financial Times, and many other publications. He writes regularly for the online sites of The Wall Street Journal, Fortune, and Harvard Business Review. Tom has been named one of the top three business/technology analysts in the world, one of the 100 most influential people in the IT industry, and one of the world’s top fifty business school professors by Fortune magazine.
Tom earned a Ph.D. from Harvard University in social science and has taught at the Harvard Business School, the University of Chicago, Dartmouth’s Tuck School of Business, Boston University, and the University of Texas at Austin.
Check out our upcoming live video big data discussions.