Analytics, the Hype and the Hope: An Interview with Data Visionary Dean Ramayya Krishnan

A conversation with the Dean of Carnegie Mellon University’s Heinz College of Information Systems and Public Policy

"Analytics." What does that word make you think of? Do you think of statistics? Technology? People poring over spreadsheets and reports? Robots and algorithms combing through endless oceans of data? Do you think of business? Do you think of science? Do you think of analytics as something that changes the world? Or is it just another buzzword?

Ramayya Krishnan is a world-renowned expert in the field of data analytics, the W.W. Cooper and Ruth F. Cooper Professor of Management Science and Information Systems at Carnegie Mellon University, and was recently named president-elect of the Institute for Operations Research and the Management Sciences (INFORMS), the world’s leading analytics and operations research society. Krishnan is the fifth CMU faculty member to hold this honor.

Krishnan has been the Dean of Heinz College since 2008, and has been a faculty member at CMU since 1988. He was instrumental to the integration of information technology into Heinz College’s curriculum, and we would not be where we are today without his efforts and his expertise.

We sat down with Dean Krishnan to discuss the past, present, and future of analytics. And to separate the hype from the hope.

^{Dean Ramayya Krishnan speaking at CMU Qatar
---}

Scottie Barsotti: Let’s start by laying some groundwork. Analytics is a term that gets thrown around a lot, and so people may not fully understand what it really means. How do you explain “analytics” to the uninitiated?

Ramayya Krishnan: It’s using data to make decisions in a proactive, predictive manner but also to better understand and explain phenomena, which itself isn’t new. What’s new is the capacity to instrument and gather large amounts of data, as well as increased computing power and algorithmic capability, which allows us to do analysis at scale.

SB: What’s happening in the field right now that interests you most?

RK: One of the interesting questions in the field right now is one of “right-sizing”: for the problem you are seeking to solve, what is the right match of data and methods that will yield solutions that have real value? Every industry is being touched by this question, but the ones responding more energetically are those industries like retail, financial services, transportation, and health care that are being fundamentally transformed due to technology.

There is a lot of opportunity for professionals coming out with training and education in technology and analytics, so our core value proposition as Heinz College is going to continue to grow in relevance. Not just for the Googles and Amazons of the world, but for organizations in every sector that are working to move toward an evidence-based culture of decision-making in order to take costs out of their system, get services and products to market more quickly, and continuously seek to improve their offerings.

SB: Some legacy corporations like UPS have done a very good job of moving in that direction, and of “right-sizing” their approach to big data and optimization, while others have not been as successful. Why is that?

RK: This is where firms differ considerably in how mature they are. Upgrading systems and processes can be expensive, but investing in the human capital and organizational culture that will make it work are much more significant challenges to using analytics. For example, you have chemical refining plants that have had data analytics and process optimization in place for the last 30 years, but it’s only in the refining process and not the firm as a whole. Whereas at Amazon, for example, evidence-based decision-making pervades the organization from the very top down. So, companies differ very much along that dimension. The vast majority of organizations are at an early stage of analytic maturity, and still need to get the basics down.

SB: What about in the public sector? How are the pressures and drivers to adopt these capabilities different?

RK: It depends on the type of agency. The security agencies have made a tremendous investment in analytics. They’re all very sophisticated in terms of being able to gather data, process it, do sensing and actuation, and perhaps even engage in tactics of cyber warfare. They’re the most sophisticated and mature in analytics because they are competing with other nation-states and non-state actors who would aim to harm us.

For the agencies that serve citizens and businesses, they are facing pressures because of technological advancements in the private sector and new standards of service that have come with that. There’s a growing expectation among citizens that technology will make things more efficient, less costly, and more user-friendly. Government departments like social services, health and human services, transportation and so forth, are starting to ramp up their data gathering and analytic capabilities in order to do more with less from a budget standpoint. And to more effectively deliver programs and resources to the people they serve. These agencies have a long way to go to reach the kind of maturity we see among the security agencies, but again the drivers for adoption and the types of problems they face are very different.

SB: What are some key societal problems that Heinz College is positioned to address in the coming years?

RK: We have several areas of strength, like cybersecurity and privacy, arts and entertainment, criminology, and health care to name a few. Our faculty and alumni are making outstanding contributions to those fields. But some of the problems that I believe cut across all of our domains of strength are problems that have to do with smart cities and the future of work. There are huge challenges in these areas, disruptions that are coming to a head in society as we speak, which makes our research centers like Metro21 and the Center for the Future of Work so vital.

^{Krishnan accepting the 2016 George D. Smith Prize from INFORMS and UPS, naming Heinz College as #1 program for analytics education.
---}

The future of work has implications for the nature of jobs being performed by people within organizations, it has implications for the skills that will be required by workers, and it has implications for education and how citizens are prepared to enter the workforce. There is a risk that very large groups of people, in this country and around the world, will be ill-prepared for this analytic technological environment. If people don’t have the skills to compete, then inequality will grow. Policy interventions are needed to weather these disruptions and create a world where people are equipped to compete rather than be left behind. Those interventions may be related to education and training, or wage insurance, or empowering individual entrepreneurship to grow in the sharing economy by, for instance, making health care portable rather than tied to employment as it often is. These are a few examples and just the tip of the iceberg of what’s possible, but the future of work is an important element of what it is we do at Heinz, because policy innovations and technology innovations must go hand in hand.

SB: You also mentioned smart cities. What excites you in this area right now, and what’s the path forward?

RK: It’s exciting because it’s analytics in action. To improve urban design and planning in ways that create greater mobility and connectivity, are economically and environmentally sustainable, and increase quality of life, you need large amounts of data which requires sensing at societal scale. Consider autonomous transportation, which of course Pittsburgh is a testbed for. If autonomous vehicles are widely adopted, it will fundamentally transform our thinking about infrastructure. Will we need so much parking? How is space going to be used? Where will people live when they have new mobility options, and how will that affect development? Analytics helps us ask the right questions in those respects.

Beyond that, we can consider the secondary impacts on other industries; there will be fewer auto accidents, for example, and that will be a public good but it will presumably disrupt the insurance industry, and the automotive aftermarket. There will also be a period of time when autonomous cars are sharing the road with human drivers, which creates policy challenges in terms of liability. It’s a system of systems. 

We are excited that we are beginning to offer coursework in these areas to our students. Professor Rick Stafford—who helped create Metro21, our smart cities initiative—is going to be teaching an innovative course to address these very questions.

SB: Is that “system of systems” idea something that drives Heinz College?

RK: It is. It's in recognizing that the problems are interconnected and have to be dealt with holistically and, if appropriate, using a multidisciplinary approach. It’s also in the way we blend methods and analytic technology, operations research, machine learning, and statistics together with social science. A computer science or engineering school will typically be more concerned with using analytics to digitize business processes and to create new devices and technologies, while a management or policy school—if they employ analytics at all—will use analytics to understand phenomena that impact industries, communities, and organizations. Here, we are concerned with both approaches and where they overlap, and as a result Heinz College is a center of excellence around issues where people, policy, and technology meet.

SB: In a nutshell, what does that mean for students? What do they learn when they come here?

RK: Fundamentally, it is about teaching students to be practitioners of “intelligent action.” We give our students the capability to not only combine technology and analytics, but to use those capabilities to solve problems and engender change. So, yes, we teach you how to use technology models and algorithms, but in the end it’s about asking the right questions, determining the appropriate tool to solve the problem, and then communicating the complexity of the issue and the analysis. Students learn this through coursework but then also through rich experiences and projects enabled via our partnerships, both external partnerships in government and industry as well as the internal partnerships we have built throughout the university. Students will derive great value from what Heinz has to offer them as well as the rest of CMU. This makes for a unique and distinctive educational experience.

SB: It’s clear what some of the upside to analytics is, what the “hope” is. What’s the “hype?” Where’s the overpromise?

RK: The overpromise is that analytics and machine learning will solve all of the world’s problems. They can’t and they won’t. They’ve become these buzzwords that everyone uses. But machine learning can’t be thrown at any problem and solve it like magic. It doesn’t always fail gracefully. It can be buggy. There’s a hubris in the attitude that analytics and machine learning are a panacea. But truly, that’s another reason why Heinz College stands out in this space. We teach our students to wield these technologies responsibly and ethically while also understanding and respecting their limitations.

SB: How important is the ethical piece? What’s being done there?

RK: It’s of primary importance and should be at the forefront of our thinking, especially considering technology’s proven ability to disrupt. We need to develop systems that are, in effect, capable of compassion and ethical reasoning. There is ongoing work at Heinz and at the rest of university to identify and understand the ways that algorithms can be intentionally or unintentionally biased. Algorithms impact decision-making in hiring processes, in health care, in criminal justice, so it’s essential to correct biases that might contribute to unjust outcomes. Not only that, but there are deep questions about psychology and how humans are affected by technology on behavioral and emotional levels. This is another area where our being at CMU is very advantageous, due to the depth of expertise in philosophy, psychology, and ethics here on campus. Those partnerships are key to addressing those issues.

SB: Analytics has fueled developments in artificial intelligence. Can you talk about that spectrum and where Heinz and CMU fit?

RK: One thing to note is that both “machine learning” and “AI” are umbrella terms that each comprise many different methods. You could be talking about systems that have beaten the world’s best human players at various games, like Deep Blue, Watson, and AlphaGo. Then there are autonomous vehicles, which collect data so an algorithm can take a physical action. And then you have systems like Alexa and Cortana, to which I make a query in natural language and get a response. Are these all machine learning or examples of AI? Yes, they are, even though they may appear on the surface to be different. What they have in common is data, which may be structured or unstructured, and then there is processing in support of cognition, perception, or sensing, followed by some actuation that occurs.

Now where we fit into all of this is that Carnegie Mellon is the place where a lot of the foundational elements that go into analytics were invented. The operations research, AI, and computer science that drive analytics, much of that was created here, and additionally CMU has one of the most innovative statistics departments anywhere. This is a boundary-less university which means we have the capacity to engage with these other areas.

Then, combine all that with a college like Heinz that focuses on major societal issues and has a special expertise on how technology is transforming society. I expect that the best work on the consequences of technological change will be done here, as well as the policy innovations that must go hand in hand with those tech innovations. This is what makes Heinz unique.

SB: On the topic of AI, and especially since we were talking about ethics. Any danger of us creating Skynet and losing control?

RK: This is an active topic, of course. There is a worry that AI is getting to a point that it will start to control us rather than vice versa. I think we are far away from that reality existing. Others would disagree with me, some feel this is something we need to actively guard against and therefore not even fund certain types of work. But for a system to be self-aware there are so many milestones that must be crossed, where it starts making value judgments in unconstrained environments. We’re not there yet, or anywhere near it currently. But the concerns are non-trivial and that debate is crucial, particularly here at CMU and Heinz.

SB: I had to ask about Skynet because rumor has it that Terminator 2 is your favorite movie. Is that true?

RK: One of them. I also like Lawrence of Arabia.

Learn More About Heinz College Programs

Learn more about Dean Krishnan