Innovator Chat: Exploring the Impact of Cognitive Computers

Dario Gil from IBM responds to questions on how cognitive computing will help make sense of our complex world.
Dario Gil, PhD, Director, Energy & Natural Resources, IBM Research

IBM_CC_innovator-chat_large.jpg


As smart as human beings are, there are many things that we can't do--or we could do better. Moving beyond the traditional computer, cognitive machines will help us bridge between our abilities and our aspirations.  They will help us overcome our limitations.

Dario Gil, PhD, Director, Energy & Natural Resources, IBM Research, responds to our questions on how cognitive computing will help make sense of our complex world.

Q: It's been said that today's computers, despite incredible advances over the past 60+ years, are just "large calculators." What makes cognitive systems different?
A: It is true that the original design point for computers was calculation. The systems were mostly designed for automation and to enhance our productivity. This approach goes all the way back to the time when computers were basically mechanical tabulators. Computing evolved and the programmable systems era was created. In this era, in which we still find ourselves, programmers write software code that is executed on a machine that has memory and logic processing, whether it's a mobile phone in our pocket or a mainframe computer.

A new era of computing -- cognitive computing -- is emerging now because of the massive amounts of big data emanating from mobile devices, the Internet, social networks and other sources. As we add more and more instruments to the world, the sources of digital information that encode our analog and physical world have exploded.

In a significant departure from today's computers, cognitive systems will learn from experience, reason over massive volumes of big data, discover new insights, and improve their own performance over time, with or without direct programming. That is a big change.

Q: Cognitive systems have been described as "machines that can help us think." How will cognitive systems affect the human-machine relationship?
A: The human-machine relationship will evolve into something akin to a partnership. The core attribute of the partnership is to improve decisions and outcomes for the user. Cognitive systems will learn, adapt, hypothesize, and recommend in real time. To be effective, they must be able to deliver consumable, user-friendly content that communicates their insights, recommendations, and supporting evidence. In addition, these systems must be capable of continuous interaction in complex, ambiguous, and uncertain situations. They must act as our partners to solve problems and define appropriate actions, engendering trust in their capabilities and assertions.

Q: What technological advancements have paved the way for a new era in computing? Which forthcoming breakthroughs are most essential to the field's success in the long term?
A: In addition to the well-known advances enabled by Moore's Law, it is worth highlighting the progress that the field of mathematics has delivered in the area of algorithms. When combined with the exponential rise in computing power, we're no longer constrained by being able to only execute a few algorithms at a time. We can execute hundreds or thousands of them in parallel.

A second trend comes into play when we consider the vast amount of digitized information we have. Think of all the text that is annotated now because of the Internet and the mobile web. Think of all the tagging that is happening from images that people are taking with their mobile phones. All of this tagging, this digitizing of information, presents an enormous wealth of information for cognitive computing systems to learn from. In other words cognitive systems will learn from the information that humans have created by digitizing and capturing our physical world.

A third trend arises from advances in user interfaces, from voice and image recognition, to ultra-high resolution displays, to touch-enabled surfaces. They are making the interaction with the user much more natural and tailored to our needs. Lowering the cognitive friction between learning systems and humans will be a source of innovation for decades to come.

Finally, and perhaps most profoundly, the international scientific efforts that are underway to map and understand the human brain will be invaluable in inspiring the evolution of cognitive computing. After all, the brain is the best cognitive system that has been created so far.

Q: One of the key factors that distinguishes humans from machines is our ability to sense, to experience our environment and react to it. Is this something that cognitive computers will be able to do in the future?
A: At their core, cognitive systems are models of, or supplements to, the complex human skills of sensing, perceiving, understanding and interacting with the world. With cognitive systems, computation will be triggered automatically from the sensory output of the humans that are interacting with the system.

Think about how we interact with another human. We don't go to each other and say, "Okay, turn on your image processing capabilities. Turn on your memory. Turn on your hearing. I want you to process this text," and so on. We just approach each other and the sensory output that we produce with the person we're in front of automatically triggers computation in our respective brains, which in turn produces the appropriate sensory output. In a similar fashion, we will need to equip these systems with the ability to see, to hear, even to experience touch. Taste is a different sort of evolution, but I imagine that can happen as well. The sensory element is going to be central to the actual cognitive computing experience.

Q: What are the largest obstacles on the path to developing computers that can sense, think, and react?
A: Cognitive systems will fundamentally be involved in qualitatively different and more complex goals, multimodal interaction and information processing than previous generations of technology. They will offer human-like interaction abilities, which will require much deeper consideration of human motivations, emotional responses and ingrained social behaviors. These issues have been superficially addressed, if at all, in the design of earlier systems. Therefore, a much stronger burden will be on their developers to understand and differentiate the roles of user and technology to create a truly unified, reciprocal and effective human-technology system. Humans will require cognitive technologies to be unobtrusive, transparent, trusted and offer clear benefits toward their own personal goals. The human's emotions, behaviors and motivations, will be the axis around which cognitive systems revolve.

Q: Let's talk about the implications of cognitive computing. How will it impact the lives of everyday consumers in the future?
A: We have created a world of massive complexity and interdependency, and we need help making better decisions. Decades of research in behavioral science have illuminated that along with extraordinary abilities, we also have cognitive biases that impact our decisions. Whether we are choosing our investment portfolio, selecting which medical treatment to pursue, deciding our college major, or picking what house to buy, we are likely to be overwhelmed with the amount of information we have to process and the number of decisions we have to make.

Let's say you are running a small business and you are launching a product, and you have to decide the price of the product. Or, let's say you're doing some kind of hiring decision. You have to make a decision between different candidates that have many different attributes. What is interesting about all these classes of problems is that they tend to be open-ended problems with a significant amount of cognitive complexity. Cognitive systems need to be designed to enhance human cognition. They need to understand how we reason and our biases. They need to help us. Not to automate a decision, that's entirely the wrong way to think about cognitive systems. It's about helping us address the inherent complexity of decision-making in the world we live in.

Q: How soon will cognitive systems become widely available to consumers? When do you predict that cognitive computing will become the norm?
A: We will see them emerge within this decade. Systems like Watson, the machine that competed against humans in Jeopardy! are proof that the era of cognitive computing is upon us. And the best is yet to come.

About the author of this Post

Dario Gil, PhD, Director, Energy & Natural Resources, IBM Research
Dario Gil, PhD, is a leading technologist and executive at IBM. As a Director at IBM Research, he has worldwide responsibilities for IBM’s energy and natural resources research activities, leading a global team of scientists across 12 laboratories that are focused on accelerating industry transformation through breakthrough science and the systematic application of the latest advances in information technology. He is the creator and Founding Director of the Smarter Energy Research Institute, an international collaborative research consortia focused on advancing the utility of the future through the use of predictive analytics, optimization and advanced computation. 
Sponsor Content

Join the Discussion

The Atlantic does not moderate these reader comments, except to the same extent comments are moderated pursuant to the Terms & Conditions generally applicable to all content on The Atlantic's sites. blog comments powered by Disqus

Applications of Cognitive Computing