How are businesses adapting to the new technology landscape and how will they keep pace with the rate of change in computing capabilities? We have seen a huge amount of activity in the Enterprise IT field, and with the huge stores of readily available data, processing power continuing to accelerate – what are the techniques that are to enable the businesses of tomorrow to become better in every facet of their operation – Artificial Intelligence.


AI is no longer deemed purely for academics, research institutes and Hollywood films; it’s now being implemented across many business verticals to great effect. The positive, and transformative impact AI presents is an opportunity all business leaders should take note of.


At AI Business we’re focused on uncovering the latest news and case studies on what AI means for business, and we were delighted to get the opportunity to meet with the pioneers in the field at IBM Watson.


Watson has been at the forefront of the cognitive revolution, and is a platform that uses natural language processing and machine learning to uncover actionable insights from vast quantities of unstructured data.


With over 80% of todays data being unstructured, it’s becoming ever more vital to have the capabilities to extract information in a form usable to the every day employee. Watson is in use across a range of verticals and so we were very interested to understand more about how it all started, where Watson can be put to use, and the business impact of implementing the technology. We spoke with Paul Chong, Director of Watson group at IBM to find out.


Paul Chong IBM

 Paul Chong, EMEA Director, IBM Watson Group


Paul, what’s the story of  IBM Watson so far and what makes it unique?


IBM has always had the idea of ‘Grand Challenges’. We’ve always strived to inspire our scientists to reach major milestones in the advancement of technology. For example, in 1969 we provided the information technology systems to support NASA putting a man on the moon. Another milestone was in 1997 when IBM’s Deep Blue beat Gary Kasparov, the reigning world chess champion. More recently we set ourselves the challenge to win one of the longest running general knowledge TV quiz shows, Jeopardy; which we achieved in December, 2011.


Work on Jeopardy actually began a little earlier, in 2006, when the idea of Big Data was emerging. Jeopardy gave us the opportunity to really focus people’s attention on what the latest advances in technology might be able to achieve. The Jeopardy win reflected the idea of taking a 12 word question and applying it to a vast amount of information stored in natural language (ordinary text), to find the answer. This concept goes far beyond ordinary ‘search’ because you have to really understand the question. We had to work extremely hard to get the right level of efficacy to ensure it could respond quickly enough to compete with the previous all-round quiz champions. It took us nearly 5 years from inception through to winning the show in 2011.

The first application for Watson was in Healthcare. Being able to run through 200 million pages and respond to a specific question within 3 seconds becomes very powerful when you consider that healthcare data is doubling every 4 years. The challenge many healthcare consultants face is the amount of background research needed to have accurate and current information – and then being able to distil it so it’s relevant to each patient.


We started to ’train’ the system very specifically in Oncology. We focused on helping improve the treatment plans for patients, specifically in terms of lung, rectal, and breast cancer. When you look into these forms, you soon realise there are thousands of derivatives of breast cancer alone, so it’s very difficult even for experts to understand them all. A key differentiator for Watson is that it provides multiple responses based on its level of confidence. It can rank the confidence level it has to a given question – in Jeopardy, if it didn’t feel it had over 50% confidence it wouldn’t provide the response – in Oncology a similar principle is used to provide recommendations. Watson has given clinicians a real opportunity to significantly improve outcomes.


It’s important to note that this isn’t about developing a general artificial intelligence engine that makes decisions itself; we have directed Watson to specific fields of human expertise to improve outcomes and help decision support.



Beyond Healthcare, which other industries is IBM Watson working on at the moment?


IBM invested $1bn in a new Watson business unit – to effectively operate as a nimble start-up but still able to draw upon the larger IBM organisation. We started with Healthcare and then extended into the public sector and financial services, where you can already see positive disruption. Our focus on developing a robust eco-system of business partners has enabled us to scale much more quickly and deliver a much broader scope of solutions than we could have done alone.


We’ve taken what we’re learnt over the past 3-4 years and packaged it as a suite of cloud-based APIs and services. These can be used by our partners who have the specific expertise across a vast range of fields. Many are now powering their own applications and solutions with Watson and making the technology available to a wider group of people. The expansion is happening very quickly; in part because of our ability to scale via the cloud and also because of our partner eco-system are creating such fantastic opportunities.



How are partners and customers getting involved with Watson– what’s the typical process?


There are several different ways to engage with us. At its simplest, there’s the opportunity to use any one of a set of 28 Watson APIs, through our Bluemix cloud development platform, which you can access for free for 30 days. We’ve made them easy to use by providing normal programming interfaces – such as Java – making it simple for entrepreneurs or developers in established companies to access them and extend their existing applications.


We also have something called a ‘Developer Agreement’ which is a more formal arrangement designed for business partners.  It’s a great way to create a closer working relationship with IBM and gives them support and access to some of our subject matter experts. Finally there’s the direct route, where we work with clients who are looking for transformational opportunities, typically needing  IBM consulting services in support of a project.



How has adoption of AI technologies evolved over the years, especially with the harnessing of Big Data?


Everybody understands the challenge of Big Data – dealing with the vast amounts of information being generated. Most project managers will tell you that, when you introduce new technologies, you’ll only get so far before you start to encounter friction from users. The issue then becomes about how do we improve the interactions between humans and the technology itself.


One solution we’ve seen demonstrated by some of our clients is that user adoption is often driven when they start to see new value being unlocked by Big Data analytics,  For example Swiss Re have deployed Watson to help make better decisions in underwriting, which has had a profound impact on their business in terms of making accurate decisions.


Disruptors entering  new markets and industries can also trigger adoption. Banking is likely to be very different in 3 to 5 years; just think about the challenger banks that are out there now. It’s quite easy to imagine a shift in how these businesses operate both technologically and operationally. Similarly if you look at the legal profession, training Watson in this field means employees can do simple clerical tasks in a fraction of the time, leaving them able to concentrate on higher value work.



We’ve talked about Watson being more a collaborative tool working alongside humans; do you believe it can eventually replace humans in actual job roles/functions?


The ultimate goal [of IBM Watson] is to improve decision-making and to improve user engagement. Take GPs for example, who struggle to spend 10 minutes with a patient. If you were able to improve the outcome or time it takes to complete background tasks, then you would be helping to improve their interaction with their patients and perhaps even further personalise the information that is given to them.


As I mentioned, Watson can help reduce the amount of time spent on routine tasks which frees more time for higher skilled work, so it’s augmenting rather than replacing humans in the workplace.



We’ve discussed how the current focus with Watson is on narrow AI, longer term do you see the potential being with broader more general AI; and will we ever actually see it realized?


The way we’re training systems today is focused on specific industries and roles. It is interesting to consider when, or if ever, a general AI engine may occur. Some academics have suggested it could be 2049, but I’d have to say it’s very difficult to predict when we might reach the idea of a Singularity. I feel it’s still a long way off from where we are today.



Looking ahead we’ve discussed IoT being a key trend where AI will enable future development. Where is IBM focusing R&D resource longer term?


A very powerful new capability for Watson is around visualizing the huge amounts of images and video data available. So now, not only can we process text but we can also exploit imagery to help make better decisions. One of the key acquisitions that we’ve recently made in this area was Merge Healthcare, which has vast amounts of information which we’re using to help radiologists identify tumours.


Dialogue is another capability that we’re focusing on; the idea being based around the concept of virtual assistants to support the enterprise. In fact, many companies that are already working with Watson are developing virtual assistants of some kind. To extend this capability further we’ve also made some strategic acquisitions in the area of Natural Language processing.


In terms of new sectors, education is very important. IBM Watson can help through personalized learning, whether it’s through  dedicated homework or a better classroom experience or more generally around a comprehensive improvement in the curriculum.


Then looking further afield, Machine Learning is an area which will exponentially help improve the learning capability of systems as we tackle more complex problems. We’re also working in the field of Quantum Computing – which is about taking inspiration from the brain to configure new forms of hardware so that we can process data much more quickly. The principle being if you can put more data through the system then you can improve the system itself.


It’s widely accepted that the world is experiencing an exponential growth in data, and that most of this is unstructured. We can argue about the specific numbers but at the end of the day the key question everyone is asking is: how can we turn this data into valuable information?  At IBM we’ve believed for some time that Data would become the world’s next natural resource and have therefore been directing our research efforts into developing appropriate technologies – for today, what we’re now calling the Era of Cognitive Computing.



In a space, which is set to be so hotly contested among those developing new ways and techniques for utilising AI, the Watson team is clearly focused on the road ahead. Right from the start Watson has been deployed commercially to help Enterprises process their information more efficiently and IBM’s R&D focus on both growing a partner eco-system and also with advancing the hardware capabilities really sets them apart in the field.


We’re sure to see some exciting developments and announcements as we head through 2016.