Cognitive computing's powerful potential

When IBM’s supercomputer Watson defeated two past champions of the quiz show, Jeopardy, in February 2011, it finally answered the question: ‘Is a computer smarter than a game show contestant?’

Its victory, along with more accessibility to the cognitive computing platform, has spurred numerous CIOs to ask whether that same problem-solving capability can be applied for more business-oriented purposes.

Watson is the first large-scale deployment of a technology referred to as cognitive computing, which is designed for resolving problems based on unstructured and ambiguous data sets – traditionally the domain of human beings.

NICTA research leader in artificial intelligence, Professor Toby Walsh, says Watson is well suited to solving problems that involve sifting through large volumes of unstructured data.

As the volume of data available to humans increases, he says tools like Watson will be essential for helping make sense of it all. An early application of Watson is in helping clinicians make better decisions.

“The number of scientific papers being written is beyond the remit of any one person to read any more, and often it is unstructured text,” Walsh says. “And yet there are vast amounts of knowledge to be tapped there. You need tools like Watson that can sift through all that evidence and point us in the right direction.”

Digital thinking

One of the first Australian organisations to sign up to Watson is Victoria’s Deakin University. Chief digital officer, William Confalonieri, says the university is using Watson initially to create an intelligent digital guide for students. Deakin has received the highest ranking of any Victorian university for learning satisfaction for the past four years, and Confalonieri says the use of Watson is intended to cement this.

“Technology is not only there to solve problems,” he says. “It is changing from a platform of record to a platform of engagement. These days, technology is critical to making experiences better.

“Student satisfaction is key for us. So it’s not only about pain points or a problem, it is to make things much better if we can do so.”

The Watson technology is now embedded in DeakinSync, a digital hub designed to serve students’ information needs from any location at any time of the day.

Confalonieri says DeakinSync replaces some of the need for students to engage directly with staff through email, telephone or face-to-face, and significantly reduces the time taken to access information.

“Potentially, we will need to think in the future how we will re-arrange internal services to accommodate the evolution that this channel is going to have,” Confalonieri says. “And it is not going to come from the point of view of savings or reduction, it is more about an extension and augmentation of services and experiences.”

While Watson will take pressure off some traditional points of service delivery within Deakin, it has required Confalonieri and his team to feed in vast amounts of information regarding how the university operates and the services it makes available. It has also raised the pressure on Deakin to operate consistently across its four major campuses.

“With this kind of technology, you need to put your backyard in order in terms of information management,” he says.

“We needed to work hard in finding inconsistencies and filling gaps, because we need to feed Watson, of course. And if we feed Watson with inconsistent things, the answers or hypotheses it can generate are not going to be the truth. “To be able to use this technology properly, you need to have your information management discipline and your information in shape.”

Deakin was only weeks into its deployment of Watson at time of print, and while Confalonieri says it is too soon to quantitatively define success, it is already answering thousands of questions with a very low level of failed results. “Everything looks good so far,” he says.

He is already planning a second task for Watson to tap into a much wider pool of information regarding the university. From there, Confalonieri says Deakin is keen to do more than just replace existing services.

For example, he intends to feed Watson information about the students themselves to create a more tailored experience. This will launch in August.

“We see the potential now to do a lot of different things that before were not possible,” Confalonieri says. “One immediate thing is the ability to deal at a massive scale with a level of intimacy that was not possible before. We can use this technology to respond to questions based on the context, based on your personal information and interests, and all the information that we know that the students provide us, and the answer can be tailored to a person.”

The power of Watson is that it can retain and respond to queries relating to thousands of individual students, and do so without forgetting individual details.

“Watson is working inside a personalised environment that knows everything about the student and responds to them personally, and is going to use that information to provide answers that are tailored to that student,” Confalonieri says.

Later this year, Confalonieri will push the boundaries still further by feeding Watson information necessary for it to provide career advice. The aim is to have Watson understand a student’s career ambitions and cross-reference these across their course work and resume.

“It is taking complex information from the market, from inside the university, from your personal profile, and combining it to really give you a roadmap for your professional life,” Confalonieri says.

While the projects Watson is being used for look significantly different to many which might typically sit within an IT project portfolio, Confalonieri says what really makes working with Watson special is that it is pushing Deakin into a whole new field of computing – one for which the rules are barely defined.

“Most other technology is mature and the boundaries are clearly defined,” Confalonieri says. “Here we are pushing.”

Deakin is also seeking to contribute to this unfolding field of cognitive computing by using Watson to teach students about cognitive computing and related concepts.

“They are going to be able to develop and build applications on top of Watson technology,” Confalonieri says, adding that Deakin is preparing to offer a full course on cognitive computing from 2016.

The university has also launched a research program looking into how cognitive computing may change other disciplines.

“The reality is that we are entering into an age where we are not able to deal with the volume of information that is produced every single day in every single profession,” Confalonieri says. “We are not able to cope with all the information that is produced to provide advice or to make conclusions.

“Cognitive computers are coming just at the right moment to help us with that problem, and those entities can read everything and come to you with some conclusions. That is where we are going to see they are saving us from the situation where we are struggling with the volume of information.”

Wider applications

According to IBM Australia’s executive for Watson Cognitive Systems, Jason Leonard, the personal assistant role Watson is playing at Deakin University is being deployed in a number of other industries, including the insurance sector, and in government service delivery in Singapore.

He says the healthcare sector is showing strong interest, particularly in the field of oncology.

“What Watson is doing is reading all of that medical literature on behalf of the oncologist, reading the patient record, and then joining the two things together,” Leonard says.

In this instance, IBM is working with the world’s largest and oldest private cancer centre, the Memorial Sloan Kettering Cancer Center in New York City. The Watson Oncology service is designed to assist the broader oncology community of physicians in determining treatment options for patients by cutting the time taken for new research and evidence to influence clinical practice globally.

Leonard says another field of strong activity for Watson has been in wealth management. “The basic idea there is the world of investments evolves pretty quickly,” he says. “Then you have investors, with their own background, and so you want to be able to provide the latest tailored advice to that person as well.”

One example is ANZ Bank’s Global Wealth division, which is using Watson to power an investor engagement advisory tool for its 400 financial planners. The bank’s goal is to shorten and improve the process of providing financial advice by a matter of weeks by matching client requirements to market offers.

Australia’s intellectual rights management agency, IP Australia, is also trialling Watson across different applications to deliver better IP rights and inventor services.

Leonard says the potential applications for Watson seem almost endless. “What we are focused on is all of the unstructured information that we haven’t focused on for the last 40 years while we have built all of these wonderful databases,” he says. “We are trying to look at the things that haven’t been effectively solved.”

Alternative platforms

While IBM is not the only company investing in cognitive computing, it is the first to bring commercial applications of the technology to market at scale. The company has committed more than US$1 billion (AUD$1.27bn) to establish the Watson business unit, employing more than 2000 people.

No other major technology supplier has come close to matching that commitment, and to date the only challengers to Watson have come from a smattering of smaller firms such as Saffron Technology.

IBM will not have the market to itself for long, however. Google is also making a play into this space, and in January 2014 acquired UK-based artificial intelligence company, DeepMind, for more than US$500 million. In 2012, Google also hired noted futurist and artificial intelligence researcher, Ray Kurzweil, as a director of engineering.

Walsh says the cognitive computing field itself still has a very long way to go before it is even close to being mature, especially if the true test is to measure cognitive computing against the model upon which it is based: The human brain.

“The human brain is the most complicated system in the universe by orders of magnitude – we don’t know anything that comes close to the human brain,” he says.

“But as the systems get better they will be more than just discovery tools, they will be able to do some of the inference and reasoning and become more and more capable. To begin with, you can see a tool like Watson being used quite readily for information discovery, but in the longer-term making suggestions as to what those decisions should be.

“Technology is going to get better and better and take over more and more of the tasks that we think of as intelligent.”

That may not necessarily be a good thing – at least not for everyone. Walsh says one of the benefits of cognitive computing is that it enables machines to become better at tasks that were traditionally the domain of people.

That may not bode well for the people who previously performed those tasks.

“It will take people’s jobs away,” Walsh says. “This is really the Information Technology Revolution that follows after the Industrial Revolution. Society changed in radical ways when the Industrial Revolution came along and completely changed the nature of work and the relationship between workers and the people who owned the means of employment.

“It is not too dramatic to say we are facing similar things here and cognitive computing is the tip of the iceberg.”

3 applications of cognitive computing

Engagement

These systems fundamentally change the way humans and systems interact and significantly extend the capabilities of humans by leveraging their ability to provide expert assistance and to understand complex matters. These systems develop deep domain insights and bring this information to people in a timely, natural and usable way.

Decision making

Decisions made by cognitive systems are evidence-based and continually evolve based on new information, outcomes and actions. Currently, cognitive computing systems perform more as advisors by suggesting a set of options to human users, who ultimately make the final decisions. Confidence in a cognitive system’s ability to make decisions autonomously depends on the ability to query and have traceability to audit why a particular decision was made.

Discovery

Discovery is the epitome of cognitive capability. With ever-increasing volumes of data, there is a clear need for systems that help exploit information more effectively than humans could on their own. Discovery capabilities have already emerged in specific domains such as medical research.

Source: Your cognitive future: How next-gen computing changes the way we live and work, IBM Institute for Business Values, January 2015