IBM Watson, in the simplest terms, is a question-answering computer system.
It is able to answer a question posed in natural language.
While it was initially developed to play a TV quiz show, IBM has since reassessed and developed the system.
Today Watson’s systems are being used to develop AI applications, drive efficiencies in clinical trials and power security analytics.
Even if you follow AI closely, it still can be very hard to keep track of everything Watson can do.
In the course of this article, we will look at the history of IBM Watson, What it is? What it can do?
Ignore the hype and see how business can tap into advanced machine learning and AI.
We will also look at how IBM Watson and its applications in real-world business cases and its wider impact on AI.
The Development of IBM Watson
Developed by a research team led by David Ferrucci, IBM Watson is named after IBM’s first CEO, Thomas J. Watson.
International Business Machines Corporation, or IBM for short, is an I.T. focused multinational based in Armonk, New York.
Following a merger of three older companies, in 1911 founded as the as Computing-Tabulating-Recording Company.
The company was renamed IBM in 1924.
Today, operating in over 170 countries the company produces software, hardware and middleware.
It also provides consulting services, particularly in nanotechnology and mainframe computing, as well as internet hosting.
Since the company’s inception, it has driven technology forward.
They have developed floppy disks, UPC barcodes, hard disk drives and magnetic stripe cards amongst other things.
During the 1960s and 1970s, the IBM mainframe became the dominant computing platform.
IBM’s desire to remain relevant has seen it constantly refocusing its operations onto higher-value markets.
What is IBM Watson? From TV Quiz Show to AI Pioneer
IBM Watson is a result of the company’s desire to remain relevant and at the top of its field.
The system was initially developed to answer the questions posed on popular US quiz show Jeopardy.
In 2011 Watson competed on the show against champions Ken Jennings and Brad Rutter. Watson triumphed, winning $1 million dollars.
Since 2013 IBM has been developing applications of Watson for use in the real world.
The first launched application saw the system taking charge of utilization management decisions in lung cancer treatment at New York’s Memorial Sloan Kettering Cancer Centre.
At the time, the then business chief of IBM Watson, Manoj Saxena, claimed 90% of nurses were following Watson’s guidance.
MIT and IBM Watson’s AI Lab
Increasingly people are turning to AI to further human knowledge and capabilities.
Against this backdrop, in 2017 MIT and IBM entered into a partnership, known as MIT-IBM Watson AI Lab.
This is the first partnership of its kind, an industry-academic collaboration.
Its goal is to chart the future of AI.
John Kelly III, the senior vice president of IBM, observed, “today’s AI systems, as remarkable as they are, will require new innovations to tackle increasingly difficult real-world problems to improve our work and lives.”
It is this application that the MIT IBW Watson AI Lab will focus on.
The Focus of the MIT – IBM Watson AI Lab
Together the work at the IBM Watson lab will focus on a number of set areas.
The first of these is in developing AI Algorithms with the intention of expanding machine learning and reasoning capabilities.
This allows AI systems to tackle complex problems.
They will also be able to deal with the robust process of continuous learning.
Researchers will also look at the physics of AI, exploring hardware, architecture and devices that may support future AI models.
Here researchers will also explore the intersection of quantum computing and machine learning, helping to characterise and improve quantum devices.
This information will help to optimize and enhance machine learning algorithms and AI applications.
The next area of focus will look at ways of applying AI in various industries, such as AI in healthcare and cybersecurity.
Finally, MIT-IBM Watson, AI Lab researchers will explore how AI can bring economic and societal benefits to various enterprises, people and nations.
As well as looking at means of commercializing AI researchers and workers at the MIT-IBM Watson lab will publish their work.
This will largely be done in the form of open-source material, which is intended to foster adherence to applying AI in an ethical manner.
The MIT IBM Watson AI Lab Quickly Attracted Attention
Within a year of launching the project had received 186 proposals from 23 departments and centers.
From these proposals, forty-eight were chosen as the first projects the MIT-IBM Watson AI Lab scientists would work on.
Many of these projects were chosen for their real-world applicability.
For example, one project focused on optimizing treatment strategies in intensive care units.
Other projects were more abstract, such as one training AI to reason the moral and ethical implications of a project.
The team are also focusing on building a new generation of neural symbolic hybrid systems.
When fully realised these will embrace the statistical learning ability of modern neural networks.
This information can then be combined with symbolic reasoning techniques.
This will enable us to solve difficult questions such as tasks involving visual question answering.
This capability has so far eluded approaches purely using neural networks.
Developers and scientists are also working on tools capable of detecting “adversarial example” attacks on AI systems.
These tools will also be able to defend AI systems from these attacks.
Adversarial example attacks make it hard to detect alterations to an input.
This can cause an AI system to fail.
Luca Daniel of MIT and IBM Watson’s Pin-yu Chen are focusing their efforts on designing tools that provide precise, certifiable guarantees, ensuring the robustness of the neural network.
This will safeguard the system from attacks and catastrophic failures.
MORE – IBM to Invest $2bn in AI Hub in New York
Learning how to Leverage Learning
MIT and IBM Watson scientists are also focusing their research on a system that reduces our need to rely on large sets of hand-labeled data.
To this end, the team, led by Abhishek Kumar of IBM and Greg Wornell, Antonio Torralba and William Freeman of MIT are focusing on a different approach.
Their deep learning system will be able to leverage learning from one domain to another, allowing it to learn efficiently in the absence of a large labeled data training set.
The focus on real-world applications seeks to counteract what has historically been one of AI’s biggest challenges.
The focus on transferable knowledge as well as access to IBM’s customer base gives researchers a ready-made test group.
This enables the team to better understand commercial concerns when it comes to delivering useful AI.
IBM Watson’s Role in Financial Services
IBM Watson is working alongside financial services firm USAA to help ex-service personnel transfer into civilian life.
USAA is dedicated to helping those who are or have served their country.
The U.S. Bureau of Labor Statistics estimates around 155,000 members of the military transfer into civilian life each year.
With the help of IBM Watson’s Engagement Advisor USAA aims to provide an easy to access service, helping people with this transition process.
IBM Watson’s natural language technology means that it has analyzed and understood over 3,00 documents on topics focused on military transitions.
This knowledge powers a system that is able to automatically answer the most basic concerns.
Automation allows more people to access the information that they need quickly.
As John Gordon, vice president of IBM Watson Group observes, “Traditional methods of providing advice and information aren’t scaling”.
This approach solves this issue, providing a 24/7 service to as many people as possible.
MORE – 10 Applications of Machine Learning in Finance
IBM and the Nationwide Building Society
IBM Watson has partnered other banking and financial service providers to streamline and enhance the customer experience.
In particular, IBM Watson Assistant has been used to power Arti, the Nationwide Building Society’s virtual agent.
Arti is intended to work alongside Nationwide employees, enhancing customer experience and optimizing access to services.
Arti’s first task was to engage with first-time home buyers, asking preliminary questions.
This information helped human employees advise customers on mortgage options.
While Arti operates in a similar way to a chatbot, it is far more powerful.
It is described by IBM Watson as an enterprise assistant.
Arti knows when to search from a knowledge base and when to ask further questions of the customer.
It also knows when to involve a bank employee if assistance is needed.
Assisting Customers of the Royal Bank of Scotland
Similarly, the Royal Bank of Scotland has also used IBM Watson to create its own digital assistant.
Known as Cora, the RBS digital assistant is capable of dealing with over 5,000 customer enquiries each day.
Cora was initially trained on a dataset of over 1,000 responses to over 200 standard customer enquiries.
By constantly analyzing customer data, RBS and IBM Watson have been able to develop Cora’s algorithms.
This has allowed them to teach the system to identify how customers ask questions and deliver the most appropriate response.
Constantly learning Cora improves every time it interacts with a customer.
Like Nationwide Building Society’s Arti, Cora also knowns when a query or conversation is beyond its knowledge base.
At this point, Cora alerts a human operative, involving them in the interaction and helping to satisfy a customer’s query.
Transforming the way Education is Provided
Both Southampton University and Imperial College London are using IBM Watson to train and educate their students.
Southampton University’s Innovative Curriculum
Southampton University is using IBM Watson to power interdisciplinary PhD research projects.
These projects are not just confined to computer science.
Students are also using Watson in marine and maritime research as well as computational chemistry and medical research.
As part of the university’s Curriculum Innovation programme students engage with IBM Watson via a purpose-built Cognitive Computing module.
Dr Russel Bentley, Assistant Pro-Vice-Chancellor, described the partnership as “a natural step as we seek to transform the experience and resources available to our students and staff.”
Southampton University sees IBM Watson as an asset, allowing and encouraging participation in its Curriculum Innovation programme.
This scheme encourages students from a wide range of subjects and interests to engage with AI.
By utilizing deep learning and AI tools students develop new skills and make new discoveries.
MORE – Computer Vision Applications in 10 Industries
Student Access to IBM Watson and the Developer Cloud
As part of their collaboration, IBM Watson will provide students and lecturers with access to IBM Watson and the IBM Watson Developer Cloud.
Advice and guidance on new content and application implementation will be on hand from IBM developers and researchers.
Technical mentors will attend conferences and advise students as they build prototypes of cognitive business applications.
Finally, the company, along with other notable thought leaders, will provide guest lectures, informing and inspiring students.
Rashik Parmar IBM Watson’s Partnership Executive in UK and Ireland said, “We are entering into the era of cognitive computing which has the potential to change the way people interact with technology.”
IBM Watson and Imperial College London
Similarly, IBM’s partnership with Imperial College London will provide both computing students and researchers with access to IBM Watson.
This lets students learn about IBM Watson, the algorithms and systems that power it and similar cognitive systems.
Students are also able to work with IBM Watson.
Director of Postgraduate Studies in the Department of Computing, Dr Alessandra Russo said, “There are many advantages to this kind of collaboration. Firstly, it provides students with access to the latest state-of-the-art technology. They also then have to become proficient enough to work in teams to apply that technology to address global challenges.”
The importance of cognitive computing is increasing each and every day.
By providing access, students are able to prepare themselves for developing and working with either IBM Watson or similar systems.
While students at Imperial College focus on particular data areas, such as genomics or cybersecurity, their skills are transferable.
This access equips the next generation of developers with the skills and understanding required to develop AI and smart solutions.
IBM Watson’s Applications in the Healthcare Sector
The Mayo Clinic is exploring the potential of IBM Watson in matching patients with the most appropriate clinical trial.
The hope is that IBM Watson will be able to quickly and efficiently complete what can be a complex task.
The Mayo Clinic carries out numerous human studies every year, often over 8.000 at any given time.
However many of these trials are not successful or have to be aborted because not enough patients enroll on them.
This is not just a problem faced by the Mayo Clinic, other healthcare providers and researchers face similar problems.
To counter this the Mayo Clinic are partnering to expand IBM Watson’s knowledge corpus.
The expanded corpus includes all the Mayo Clinic’s trials as well as those contained in public databases like ClinicalTrials.gov.
Following the information input, IBM Watson is trained to analyze patient records, matching the information with clinical trial criteria.
This allows the system to identify appropriate candidates for clinical trials.
MORE – Top 10 Ways Artificial Intelligence is Impacting Healthcare
MORE – IBM and Health Startup Take on Lung Disease Using AI and Big Data
IBM Watson for Clinical Trial Matching
Early use of this system in 2016 saw IBM Watson identify patients for systemic therapy for breast cancer trial.
In the 11 months following the start of the trial, the Mayo Clinic reported an 80% increase in enrolment.
Dr Tufia Haddad observed that using IBM Watson “enabled all patients to be screened for all available clinical trial opportunities”.
The speed and accuracy of IBM Watson’s clinical trial matching system also allowed physicians to develop effective patient treatment plans.
These plans are designed to reflect the complete range of available options.
This means that patient care is fully optimized and supported.
Following the success of this trial the Mayo Clinic and IBM Watson are developing similar systems for other cancer types.
They are also working on systems to optimize cancer care in areas away from medical therapy such as radiation.
In Thailand, the Bumrungrad International Hospital is working with IBM Watson to improve cancer care at their Bangkok medical center.
IBM Watson’s applications are also being used for case evaluation in referral offices in 16 separate countries.
Bumrungrad is using IBM Watson’s systems to help their doctors access medical evidence.
The systems also allow them to access research and clinical expertise provided by their partners at the Memorial Sloan Kettering.
IBM and Arthritis Research
IBM Watson is working with Arthritis Research UK to develop a virtual assistant.
Each year the charity’s website deals with thousands of questions from concerned individuals.
Their queries range from wanting to know the symptoms of arthritis to treatment options and how it impacts on lifestyle.
The charity hopes that the IBM Watson powered chatbot will help to speed up response times to these queries.
Using the IBM Watson Conversation API this chatbot is able to provide information tailored to the individual query.
More complex queries and interactions can then be referred to a trained staff member.
Liam O’Toole, CEO at Arthritis Research UK, explained that the charity wants to “ensure that everyone has access to information and support, whenever and wherever they need it.”
The Moorfield Eye Hospital
Moorfields Eye Hospital has also partnered with IBM Watson to develop a chatbot, known as Oriel Assistant.
Like the arthritis UK bot, it is capable of answering questions from staff, patients and members of the general public.
Its initial purpose is to deal with enquiries concerning the hospital’s proposed relocation to a new facility in Kings Cross.
Oriel Assistant is able to explain the plan to interested users while also recording feedback on the proposals.
Eventually, Oriel Assistant will be used to deal with a larger range of enquiries such as questions relating to diagnoses.
The director of digital innovation, Peter Thomas, explained that the ideal purpose of the chatbot would be to deal with a patients “general questions about their condition.”
This was launched in the summer of 2019, Oriel Assistant was trained on over 500 stock questions.
These were gathered by surveying staff and members of the public.
Like other IBM Watson applications, Oriel will continue to improve the more it interacts with people.
As many of the potential users will be suffering from a form of visual impairment the chatbot was designed to be customizable.
Users can alter the color contrasts and font sizes, making it accessible to all.
MORE – Artificial Intelligence in Medicine – Top 10 Applications
Enhancing Cyber Security Applications
As the world becomes increasingly digital, cybersecurity is growing in importance.
IBM Watson is at the forefront of cybersecurity research and developments.
SparkCognition describes itself as the world’s first “cognitive security analytics” company.
Based in Austin, Texas the startup was founded by Amir Husain, the former ClearCube CTO.
Much of their work, such as their Cognitive Security Insights app (CSI), is carried out alongside IBM Watson.
Watson has been trained on a security-focused corpus by SparkCognition.
This detailed corpus includes product manuals, configuration best practices, threat research papers.
The training was done with the end aim of teaching the CSI app to think like a cybersecurity expert.
Many companies have seen the potential and usefulness of SparkCognitions work.
One of their earliest adopters was ExamSoft.
Here CSI is used to reduce the costs associated with false positives.
Usefully, CSI also incorporates automated learning systems, retaining critical knowledge and allowing for intelligent detection to occur.
Talking With Watson
Another of SparkCognition’s developments, MindSpark is the first machine capable of having an intelligent conversation with Watson.
Like Watson, MindSpark reads vast amounts of data, identifying patterns.
Since launching SparkCognition have focused their efforts in two distinct areas, the physical and the virtual, or safety and security.
In the first instance, MindSpark is able to help customers navigate their way through the industrial internet.
This is the internet’s automated infrastructure responsible for delivering utilities.
MindSpark is also used in predictive maintenance, predicting machine failures and prevent catastrophic shutdowns.
In the virtual realm, SparkCognition works alongside cloud operators to help detect security threats originating in the Internet of Things.
Both applications see MindSpark using instrumentation to gather primary data straight from the source.
While industrial internet sensors, which detect vibration levels, temperature etc, are commonplace, virtual sensors are less widely utilized.
These lie at the end of many connections, generating data that is then written in log files.
MindSpark is possibly the first application to gather and sort this information, turning it into a useful data mass.
Once analyzed the data is read for anomalies, such as excessive requests from one IP.
Any identified anomalies are then transferred into a graphical summary for easy understanding.
MindSpark is also able to construct models that characterize the possible cause of these anomalies.
It can then choose the most likely model to have caused the anomaly.
More Powerful Versions Will Allow for Increasing Automation
While MindSpark can run by itself, a more powerful version runs in conjunction with IBM Watson.
As IBM Watson is a repository of structured and unstructured information this increases the power of the system.
It also allows it to answer its own questions.
In this way, IBM Watson and MindSpark can have a conversation.
Husain sees this as just the beginning.
He believes that technology will soon be able to take over cognitive tasks from humans.
This evolution will replicate how machines replaced manual labor during the industrial revolution.
The advantage of using machines over humans, especially in complex, time-consuming tasks is that they won’t get tired and make mistakes or lose focus.
Many companies such as Boeing are already partnering with SparkCognition, keen to make use of their developments and applications.
MORE – Campbell Teams up With IBM’s Watson for AI Advertising Campaign
MORE – IBM Unveils IBM AI OpenScale to Combat AI Sprawl
Improving Customer Service
Since 2016 GlaxoSmithKline, the world’s sixth-largest pharmaceutical company, have been working with IBM Watson.
The aim of this partnership is to help GSK forge stronger connections with its customers.
The partnership has focused on developing a system allowing customers to ask questions, by text or voice, to online adverts.
Jason Andree, senior brand manager at GSK’s Cough and Cold division, said: “Watson provides a very personalized experience”.
Launched to coincide with the onset of the cold and flu season this question and answer system is a first for GSK.
It is designed to link and promote products developed under GlaxoSmithKleine’s Theraflu cold and flu medication range.
A similar system, Flonase has also been developed for launch in the allergy season.
Known as Watson Ads, these will initially run on the Weather Channel’s digital platforms such as their mobile app.
This means that they will reach a wide range of people, in numerous, accessible scenarios.
By making Watson Ads, widely accessible IBM and GlaxoSmithKline hope to reach as wide a range of consumers as possible.
An Interactive way of Recommending Products
This application of interactive advertising is an exciting innovation.
Users can ask a GlaxoSmithKline advertisement on how to treat their cough, or which medication is best for a blocked nose.
The Watson powered system will then recommend an appropriate product or course of action.
This application takes full advantage of Watson’s natural language capabilities as well as machine learning and AI tools.
These adverts aren’t intended to replace human interactions such as visiting your physician.
They are, however, an interactive, quick responding way of providing people with the information that they need.
The adverts also provide users with links to the GSK product website or coupons, which can be used to purchase products.
For users, this is intended to be a quick way to get information, such as the right product to treat their symptoms.
GSK see the adverts as a way of building a relationship between consumers and their products.
They hope that this will translate into long-lasting customer loyalty and an even bigger share of the market.
Recording and Using Customer Information
Watson Ads will also provide companies with useful information such as consumer behavior patterns and expectations.
This information will be gathered as people interact with the system and can be accessed easily by GSK.
Previously this information was generated through direct contact with consumers, such as interacting via a consumer hotline.
This has historically been a difficult way to gather and collate information.
GlaxoSmithKline hopes that IBM Watson Ads will provide a larger information base, from a larger consumer sample.
The information will also be presented in a more useful and accessible way.
This can then be used to improve and develop products, curate brand lines and plan advertising campaigns.
IBM Watson’s Evolution in the Retail Industry
IBM has also been behind the development of a mobile service that lets in-store shoppers ask questions products.
Developed with shopping chain Macy’s, the mobile app is known as Macy’s On-Call.
As well as answering in-store product questions the app is also able to respond to queries about facilities and services.
Natural language processing systems and machine learning tools allow the app to answer questions posed using everyday words and phrases.
To broaden its usability and appeal the app is available in both English and Spanish.
The app is also customized to the individual store by accessing the phone’s GPS information.
Alternatively, the user is required to enter their ZIP code.
Like other Watson developments, machine learning allows the app to evolve and improve over time.
Following a soft launch in 10 of Macy’s stores.
If successful the app will eventually be made available throughout the chain.
Macy’s vice president of digital media, Serena Potter said that “Our philosophy is to test small, launch quick, learn fast and then scale.”
She believes that this approach will allow Macy’s and IBM Watson the chance to scale and refine the system.
Potter revealed that Macy’s hopes to “implement additional cognitive services in the future.”
MORE – 10 Powerful Applications of Artificial Intelligence in Retail
Constant Innovation has Kept IBM at the Forefront of Technology
Since their formation in 1911 IBM has strived to stay at the forefront of innovation.
Their IBM Watson supercomputer is allowing the company to do just that.
Today IBM is at the heart of innovations in a number of fields including financial services, retail, education and cybersecurity.
As ever, IBM is also looking to drive technology on.
Constantly looking for new ways to evolve and adapt their applications and systems.
As AI becomes more important, powering many everyday tools and systems, IBM is intent on retaining its position at the heart of innovation.
Learn more about IBM Watson
Can an ETF Perform Better with IBM’s Watson AI?
IBM’S AI Wins Debate with Human