In the first week of October I participated in a Cognitive Systems Colloquium hosted by IBM at its Thomas J. Watson Research Center. IBM defines cognitive systems as “a category of technologies that uses natural language processing and machine learning to enable people and machines to interact more naturally to extend and magnify human expertise and cognition. These systems will learn and interact to provide expert assistance to scientists, engineers, lawyers, and other professionals in a fraction of the time it now takes.”
The need for such systems is a result of the explosive growth of data all around us. Not only are we now able to collect huge amounts of real-time data about people, places and things, but far greater amounts can be derived from the original data through feature extraction and contextual analysis. One of the key lessons from Watson, - IBM’s question-answering system which in 2011 won the Jeopardy! Challenge against the two best human Jeopardy! players, - was that the very process of analyzing data increases the amount of data by orders of magnitude.
This is challenging our ability to store and analyze all that data. The new generation of cognitive systems will require innovation breakthroughs at every layer of our IT systems, including technology components, system architectures, software platforms, programming environment and the interfaces between machines and humans.
In the opening presentation, IBM Research director John Kelly summarized both the promise and challenges of cognitive systems. Kelly just published Smart Machines: IBM's Watson and the Era of Cognitive Computingco-written with IBM writer and strategist Steve Hamm.
Data-driven cognitive systems are quite different from the programmable systems we’ve been using for over 60 years. Just about all computers in use today are based on the Von Neumann architectural principles laid out in 1945 by mathematician John von Neumann. Any problem that can be expressed as a set of instructions can be codified in software and executed in such stored-program machines. This architecture has worked very well for many different kinds of scientific, business, government and consumer applications but is limited in its ability to deal with large amounts and varieties of unstructured information in real time.
Our brains have evolved to do so quite well over millions of years. For example,Watson required 85,000 watts of power, compared to around 20 watts for the brains of the human players. But, while our brains are incredibly efficient, they can’t keep up on their own with the huge volumes of information now coming at us from all sides as well as with the increasing complexity of so many human endeavors, - including medical diagnoses, financial advice or business strategies.
So, just like we invented industrial machines to help us enhance our strength and speed, we now need to develop this new generation of machines to help us enhance our cognitive capabilities. In fact, the architectures of such cognitive systems have more in common with the structure of the human brain than with those of classicVon Neumann machines.
The colloquium focused primarily on the new kinds of applications that will be made possible by cognitive systems. Kelly discussed four such major application areas: assistance, understanding, decisions and discovery.
We’ve been using data-driven assistance applications for over a decade. For example, search applications have been indispensable to help us deal with the explosive growth of the Web, and sophisticated analytics have enabled the automation of a variety of operational processes, including logistics and inventory management, personalized marketing offers, and fraud detection in financial transactions. We can significantly improve these applications by adding contextual and domain-specific data, which will enable the system to better interpret what we are after. Such work is underway in a wide variety of industries.
Beyond handling well-defined questions and processes, we want our cognitive systems to help us extract insights out of all that data by uncovering hard-to-find patterns and connections, as well as to help us make complex decisions by analyzing different and potentially conflicting options. This is what data science is all about. This emerging discipline is developing algorithms, tools and methodologies to help us explore and examine the data from multiple disparate sources with the goal of discovering a previously hidden insight which can provide a better diagnosis to a complex medical problem or a competitive advantage in addressing a pressing business problem. The resulting decision making applications should significantly enhance the intuition of experts.
This is the goal of the various Watson-based research initiatives already underway at IBM. For example, “In health care, Watson and Watson-like technologies are now assisting doctors at Memorial Sloan Kettering in diagnosing patients by providing a variety of possible causes for a set of symptoms. Watson can help doctors narrow down the options and pick the best treatments for their patients. The doctor still does most of the thinking. Watson is there to make sense of the data and help make the process faster and more accurate.”
I moderated a panel discussion on The Cognitive Economy - Decision Making in an Era of Uncertainty where we explored these topics with industry experts. Dr. Douglas Johnston, a surgeon at the Cleveland Clinic felt that getting ready for data science will be a very big challenge in healthcare, because the underlying data is so poorly organized and is often quite inaccurate. He said that many of the inefficiencies in our present healthcare system are a result of the poor way that data is now gathered, distributed and accessed. The whole model needs to be rethought, and he hoped that cognitive systems like Watson could help do so.
Mike Carlen, an executive from DTE Energy in the Detroit metropolitan area talked about the work they are doing with IBM Research on applications that will help them anticipate and prepare for major storms, predict electrical outages, and prepare emergency management equipment and personnel to respond in the areas that will need it most. For example, when facing very high winds, they would like to predict which trees are most likely to fall and bring down power lines based on the winds’ strength and direction. They can then get their crews ready prior to impact, and estimate whether they will also need to bring additional crews from other areas. This is an excellent example of the kind of decision making that has mostly relied on the experience of experts, whose intuition can now be supplemented with sophisticated data, models and predictions.
Kevin Reardon, who has the overall responsibility for business development at IBM, discussed how the company is leveraging information and sophisticated analysis to evaluate potential acquisition and divestiture targets. This is particularly important to help reduce the emotions and biases that generally accompany such complex business decisions. It’s not enough to just analyze the financial aspects of an acquisition. There are many additional factors, including whether the culture of the acquired company will mesh with IBM’s culture. Will the executives and employees feel comfortable being part of a large, global organization, or will there be serious culture clashes that can derail the potential financial benefits of the acquisition? This is another example where cognitive systems can be of great help in complex decision making.
The final category of cognitive system applications is discovery, that is, applications that enhance creativity and innovation and lead to new disciplines. At the MIT CIO Conference earlier this year, MIT professor Erik Brynjolfsson observed that throughout history, new tools beget scientific revolutions. Scientific revolutions are launched when new tools make possible all kinds of new measurements and observations.
Early in the 17th century, for example, Galileo made major improvements to the recently invented telescope which enabled him to make discoveries that radically changed our whole view of the universe. Over the centuries we’ve seen that new tools, measurements and discoveries precede major scientific breakthroughs in physics, chemistry, biology and other disciplines.
Our new big data tools have the potential to usher an information-based scientific revolution. And just like the telescope, the microscope, spectrometers and DNA sequencers have led to the creation of new scientific disciplines, data science is now rapidly emerging as the academic companion to big data. One of the most exciting part of cognitive systems and data science is that they can be applied to just about any domain of knowledge, given our newfound ability to gather valuable data on almost any topic, including disciplines where people and their behaviors play a major role, including the social sciences, healthcare, business, finance, cities and government.
The development of cognitive systems and the data science applications they will make possible are among the most exciting grand challenges in the decades ahead. In the end, this is the natural next step in our quest to quantify, understand and make sense of the increasingly complex world around us.
Irving, what are your thoughts regarding computational creativity, the fifth and topmost element of John Kelly's pyramid of cognitive systems?
Posted by: IBM Research | November 01, 2013 at 01:16 PM