Just what the doctor ordered—artificial intelligence and medical diagnosis compared

Since announcing our global partnership with Celaton over two years ago, our relationship has grown from strength to strength, and we are continuing to successfully apply inSTREAM™ to automation and transform business processes for our clients.

In his last blog for Capgemini, CEO of Celaton Andrew Anderson shared insights on how machine learning can help manage unpredictable data volume and deliver significant benefits in finance and accounting. In this blog, Andrew discusses the misconceptions organizations often have about what problems can be solved with artificial intelligence.

Andrew Anderson, CEO, Celaton

When I talk to organizations about artificial intelligence (AI), they often assume it can solve all of their problems—a bit like going to the doctor when you’re sick and asking for medicine to make you better. But with so much “noise” being generated about AI it’s understandable that organizations have such misconceptions and are unclear about what problems can be solved and with which solutions.

So to take our doctor analogy further, we could draw a parallel between the way you would go about diagnosing and curing a medical ailment and the different forms and complexities of automation and AI.

In medicine	In technology
If you have a headache, sore throat or if you cut your finger, you can self-administer a tablet or sticking plaster from your first aid kit.	If you want to automate some of the tasks on your computer, you can create macros, templates and rules using the tools available on your computer.
If you’re suffering from an ailment or sickness you can’t self-administer medicine for, you can go to your family doctor or physician, who should be able to prescribe specific medicine providing your ailment is not too serious.	If you want to automate simple processes such as manipulating applications or handling structured data, the tools on your computer won’t be up to the job and you’ll need a robotic process automation (RPA) solution from an outsourcing provider.
If your ailment is beyond the capability of your family doctor or physician, he or she will refer you to a consultant doctor that specializes in your particular complaint.	If your challenge is more complex or unstructured, then you’ll need AI and machine learning specifically designed to solve the particular problems within your organization.

When you’re starting to look for AI solutions for your business, it is first important to identify the business problem or pain-point you would like to solve. You may also wish to ask yourself:

• How are you going to use the AI solution to improve your processes, innovate and grow?
• How will you integrate the solution with your workforce, processes and IT architecture?
• What data sets will you use for a proof of concept or machine learning?
• What staff training can you initiate to enhance and adapt skill sets?
• What are the risks to your business and customers?

The reality is that AI is not a single all-purpose capability, but a collection of specialist AIs, each designed to carry out an individual task really well. At Celaton, we refer to them as “skills,” and they’re designed to deal with specific tasks such as enquiries, claims, invoices and complaints. There are a dozen different AIs involved in each of these tasks alone.

Below is a guide to some of the most common AI “skills” and their application:

• Machine learning—most AIs encompass machine learning to some degree, whether it is a taught data set model or a self-learning model. Machine learning makes technology truly artificially intelligent.
• Natural language processing (NLP)—Natural Language is particularly relevant to technologies that need to process and make sense of incoming human generated language, such as social media, emails and live chat.
• Cognitive analytics—AI is applied to extract meaning and insights from data, present it to the user and make future predictions based on trends.
• Image recognition—largely used by machine-based visual tasks through creating meta-tags and recognizing images. This is a mixture of offline and live data.

Celaton’s inSTREAM™ platform utilizes all of the above skills to process the plethora of content and information that flows into organizations every day from customers, suppliers and employees, across channels including email, social media, fax, post paper and other electronic data streams. inSTREAM™ learns from historical data that organizations have already processed (often called training data), but has the unique ability to continuously learn through the natural consequence of processing and monitoring the actions and decisions of people who are involved in the process.

To learn about how Celaton’s inSTREAM™ contributes to Capgemini’s Automation Drive to deliver new ways of working, drive innovation and increase business value, contact: lee.beardmore@capgemini.com

Click here to learn more about how Capgemini’s Virtual Delivery Center brings together the governance, technology, methodology, and skills required to deliver intelligently automated business operations to our clients through a virtual workforce of robots and platform delivery.

About the author

As Chief Technology Officer for Capgemini’s Business Services, Lee Beardmore’s role is to, help Capgemini’s clients face complex challenges through innovation and by making sound technology decisions for real business gain. A computer scientist by education, a technologist at heart, and with many years of cross-industry experience, Lee has a wealth of experience to draw upon.