BEYOND THE BUZZ

Today’s trending topics. All in one place.




July 2017

Automated, Smart, Intelligent

Welcome to July’s BTB where we share examples of how the world is being shaped by robots, artificial intelligence, and automation. For some, automation is the simple replacement of repetitive tasks, but for others, it can change entire business models.


Subscribe


Of Robots and Men

Lanny Cohen, Group CTO, Capgemini

The calculus is simple: robotization is set to transform every aspect of society. It’s only a matter of timing and what pieces will emerge first in the race to cultivate the ‘digital everything’ revolution into practical, valuable solutions. 

Robotization is set to transform every aspect of society. Lanny Cohen, Chief Technology Officer at Capgemini, reflects on this incredible revolution.

12 MILLION: ROBOTS IN SERVICE AROUND THE WORLD IN 2016

Age of the intelligent robot

“Robots, you ask? They’re here, and they’re getting better all the time!” Lanny Cohen, Chief Technology Officer at Capgemini, is excited that robots are becoming increasingly able to handle complex tasks. “They can do almost anything: manufacture products, collect and analyze complex data in real time, perform services for individuals… The advantage of robotization is that it dramatically boosts companies’ productivity and optimizes processing times. In high-tech industries such as the space industry, we found that robots made it possible to carry out more tests and manufacturing trials –  and all at a 100% reliability rate. Human error is a thing of the past,” says Cohen.

"RATHER THAN ASKING WHAT TECHNOLOGY IS DOING TO US, WE SHOULD ASK WHAT IT CAN DO FOR US." - LANNY COHEN

Robots everywhere!

Robotization is spreading to all sectors and departments, from manufacturing to commerce, healthcare, education and energy. “You wouldn’t believe how many projects, experiments and prototypes we develop each day with our clients,” continues Cohen. “Businesses are taking on a whole new mindset! Digitization is a game changer that forces us to rethink all our old systems and organizations. Robots are the direct result of this digital revolution.”

"ROBOTS DO NOT DESTROY JOBS. THEY TRANSFORM THEM." - LANNY COHEN

And Capgemini is playing its part, with plenty of assets to contribute. “The first is our balance between global vision, which helps us keep abreast of technological advances, and our local presence, which enables us to offer solutions tailored to our clients’ exact requirements. The second advantage is our innovation-related experience and our ability to make innovation work for our clients. Finally, the Group is adept at forming partnerships, particularly with its clients. This is essential since innovation is increasingly emerging as a result of discussion and reflection. Companies put their confidence in us and we work with them using a partnership approach,” he explains.

What role do people play? In a society of robots, what’s left for people to do? “It’s true that some jobs will change. However, robotization is more likely to replace activities within these jobs, leaving more time for value add tasks,” says Cohen. “But we can encourage a smooth transition by helping employees acquire new skills. In addition, new professions are on the horizon, such as robotic architects and engineers, testers, software designers, robotic operators, specialized consultants and cybersecurity personnel. A new framework is falling into place.”

1.3 MILLION: INDUSTRIAL ROBOTS SOLD WORLDWIDE IN 2016 ALONE

Creativity, emotional intelligence and cognitive flexibility will be the new skills needed. This encourages more of our human potential and allows robots to complete activities within our jobs that are predictable – working as a team essentially.

There are plenty of reasons to be optimistic. For one, robotics offers new possibilities beyond productivity such as enhancing equal opportunity. “Robots can be a tremendous tool for reducing inequalities by facilitating access to education. We are already working with various companies and institutions on developing robots that can learn and transmit knowledge. The opportunities in this field are endless,” concludes Cohen.

"AT CAPGEMINI, WE’RE NOT AFRAID OF TECHNOLOGY. IT’S PART OF OUR DNA AND OUR CULTURE." - LANNY COHEN

 

Share this article

The 5 senses of Artificial Intelligence

Christopher Stancombe, Head of Industrialization and Automation

Artificial intelligence is not a singular piece of technology that can be bolted onto a business process. Rather, it is a set of five senses, when combined deliver better outcomes and progressively greater efficiency to the organization.

I believe that there is a misconception that Artificial Intelligence is—or will be—a single piece of technology that should be bolted onto a business process to make it “smart” or “independent” from human intervention. My experience to date is that the answer is more complex and more interesting.

Rather than a single solution, the real “intelligence” is in how a set of technologies are combined to create a solution. It is similar to our perception of human intelligence; this isn’t built on a single element, it’s a combination of senses, experiences, and knowledge.

Defining the five senses of Artificial Intelligence

I looked at a variety of solutions that are deemed to display artificial intelligence and concluded that they had five attributes in common based on a fusion of smart processes and intelligent automation. In explaining my findings to my colleagues, I found myself likening the attributes to senses in a human being. Hence, the concept of the five senses of AI was born.

1. Interaction (talk/listen) – This is the ability to listen, read, talk, write and respond to users of the AI solution. The aim here is for technology to ensure that the interaction feels intuitive and the customer is happy. Examples in this space include chatbots and voicebots.

2. Monitor (watch) – Here technology is used to watch and record key business data. It is used to create knowledge. This would include CCTV and IoT sensors.

3. Knowledge (remember) – This is about being able to store and find information effectively using components like databases and search engines. This is probably the worst developed area within corporations, but examples include Wikipedia and my hard drive.

4. Analyze (think) – This is the ability to detect patterns and recognize trends.  It applies algorithms to knowledge to determine appropriate action or predict future consequences.

5. Service (act) – This area uses technology to do things. We are used to the concept of Robots working on an assembly line and now they are moving into the office. Examples include resetting a password and placing a customer order. 

I find this framework useful to ensure that the solutions we design and create for our clients are complete and meet the test of being “artificially intelligent.” It also helps us to group technologies and to evaluate them against each other. I plan to share the criteria, results, and some exciting developments within Automation Drive in later posts.

 

Keep calm and think of it like a platform

There’s some uncertainty or even fear in the media about the concept of Artificial Intelligence. I think this ties back to the misconception that it is a single all-encompassing solution.

However, if we look at it more as the result of combining a range of senses that replicate attributes of human intelligence, I believe that it is where we will see real value without the fear. Multiple technologies with different attributes working together to help the organisation become more responsive, relevant and intuitive. That’s what consumers, employees, and shareholders are really clamouring for.

Share this article

The Must-Have Accelerators to Drive Automation Efforts

Xavier Chelladurai, Vice President – Group Competitiveness - Automation

It’s widely accepted that automation can mean a 40% to 60% effort reduction in application maintenance and even more in business services and testing. But despite all the positive reviews and media predictions is it living up to the hype?

It’s widely accepted that automation can bring about a 40% to 60% effort reduction in application and infrastructure maintenance – and 60% to 80% in business services and testing.

But despite all the positive analyst reviews and media predictions I see every day, the reality just isn’t living up to the hype.

I’ve found that managers and stakeholders behind the systems are active in coercing their teams to accept and react to the reality of rapid change that automation efforts bring. But in the majority of cases, the process of automation takes more time for many different reasons. These range from validating the business case, technical security clearance, and deploying automation scripts and tools in the production environment.

So how do you get around this and guarantee the success of your automation?

How do you remove or reduce the negative impact these bottlenecks, increase the speed of your automation efforts, and start to reap the benefits of huge numbers in effort reduction?

I’m convinced that the answer here lies in correctly implementing these three key accelerators:

1. Automation DevOps model

2. Automation design thinking teams

3. Automation skill development

Automation DevOps model – Increase control and efficiency

In automation, we identify manual service delivery areas and automate them. This means that automation is a software development project consisting of requirements, design, development, testing, deployment and maintenance.

Your automation projects require appropriate coding guidelines, test design and continuous deployments. With adequate version control, code management and reusability, you can achieve huge gains in efficiency. So adopting the right DevOps platform is essential for faster, cheaper and better service delivery.

Application Maintenance (AM), Cloud and Infrastructure Services (CIS), Testing and Business Services (BSv) teams are supposed to receive automation service from AD teams with specialized DevOps skills – but I’ve found that this is not actually happening in practice. Teams try to mobilize their skills and automate, but this results in pockets of automation. This not only affects your overall automation speed, but it also fails to motivate your teams to deliver faster automation results and success.

Putting the “Dev” in Development – Your DevOps model needs one clear vision

Several RPA and ITPA tool vendors cater to business process and IT support teams without proper onboarding of the DevOps development approach. Business process and IT support teams are automating simple functions with enormous effort without understanding the end game plan – Development. So the key to a successful Devops model lies in aligning your vendors and teams to create one development vision.

Automation design thinking teams – Your foundation for developing agile automation scripts

Agile development has matured over the past ten years. However, the rapid communication between users and developers has resulted in decreased focus on design, as well as lower code quality and maintenance. While architects are a big part of agile development teams, it’s clear that little time is spent on design.

Design thinking = Design quality from the bottom up

To put innovative ideas into place in the cognitive cloud, researchers use a design thinking approach for quick, quality products. Design thinking demands a quick and thorough understanding of customer requirements with the help of a tangible prototype and a few conversations with the customer before implementation. To guarantee quality, agile solutions from the bottom up, your automation implementation team must use a design thinking approach to develop automation scripts utilizing agile methods on your DevOps platform.

Automation skill development – Giving the right people the right skills

I’ve found that when it comes to candidates for automation, 25% to 30% are simple tasks that can be automated using scripts like Perl, Python or R. For example, application maintenance, report generation, server restarts, password resets, and shutting down long running processes, can be automated by using scripts.

However, reprocessing of failed orders is a complex task. There may be hundreds of checks to perform within the transaction. You need to compare with the reference data retrieved from a database before correcting the transaction. This requires an orchestrator tool that has a developer studio with a drag and drop facility and a powerful scheduler.

You can choose from popular orchestrators such as ServiceNOW, BMC Attrium and HPOO. But in Business Services, the loan approval process, account closure and insurance claim procedures are complex in nature. So they’re automated using RPA tools such as BluePrism, UIPath, Automation Anywhere, etc. Artificial-Intelligence-based automation is performed to understand structured and unstructured data and convert them into actionable intelligence you can use.

Creating an automation training plan that works

I know that as automation gains momentum, the demand for the above skills will only increase. In fact, the next two years will result in 25% of team members developing and maintaining automation scripts and systems. So you should strongly consider creating a plan for specialized training sessions with hands-on modules to meet this demand head-on.

DevOps, design thinking and skills development – Your automation keys to effort reduction

We know automation adoption is increasing at a rapid speed in almost every industry segment as the next wave of modernization and optimization expands. The volume of work in automation is enormous and every delay impacts your customers in terms of cost and quality.

In order to accelerate your automation implementation, it’s crucial to focus on developing the required skills and utilize design thinking to create tangible designs. You can then implement these skills and designs within your DevOps platform.

I know this seems like a very complex undertaking, but these three key accelerators form the pillars of an exciting journey towards a new world where machine intelligence augments human work, as machines can now react faster and process more inputs than humans. And the benefits that your business and customers reap from this will only multiply.

 

Share this article

With a Little Help from my Friends

Denis Sproten, Senior Solution Architect

Healthcare in the industrialized west is facing shortages and price escalation. AI could be the cure, or at least provide the busy clinician with inexpensive, reliable assistance with everything from capturing symptoms, to scheduling, to diagnosis.

The US and UK healthcare systems need help, lots of help. The NHS England spends roughly £180 million every 15 hours, costing the taxpayer every minute. And if that is not enough one in ten nursing posts are vacant, with one in three nurses over 50 retiring soon. Around 11% of nursing posts in England were vacant as of December 2016 (it was only 6% in 2013). 

If I would be a patient, I would be calling my AI for help. Even with simple capabilities AI could save so much time and help the nurses, the NHS.

Some of the things which even simple chatbots could do:

1. Capturing symptoms

2. Organizing a GP visit (date/time)

3. Instructions/leaflet on medication, side effects (linked to medical file, pointing out dangers), when is the best time, with food etc

4. Recommended actions once diagnosis is complete / medication

5. IoT – capture hospital bed sensors, retrieved by a chat bot.

6. First aid instructions

But chatbots should not be confused with AI, especially in public health. Also, AI is still too inexperienced to provide expert advice – and to identify a situation which requires “expert advice” is difficult. The dangers of an inexperienced AI to interpret human information wrongly could be devastating/deadly.

To point 1, considering that capturing symptoms take up 80% of a conversation with your doctor/GP, this would save a lot of time.

Points 4&6 could potentially be fatal, but with regards to first aid, if there is no-one else, it is probably better than doing nothing.

Will 2017 be the year when AI goes mainstream?

AI, no. Chat bots yes, they have gone mainstream already. A real AI will probably be another 15 to 20 years. One which specializes in just 1 area of medicine probably much sooner than that. But sharing of data (between hospitals and countries) will be essential to AI learning from experience (and mistakes).

What does the ubiquity of chatbots mean for the public healthcare industry?

No regulation could potentially pose dangers to the public. If everyone publishes their "patched together" chat bot as an expert that is very risky.

How will it affect public healthcare if "administered"?

A lot of the reasons why we currently have such a high death rate in public health services, is because mistakes just get brushed under the carpet and doctors don’t communicate their mistakes to other doctors. Even if the data is anonymized it would be essential for AI programs to work and learn from. The same way as it has worked for pilots in the aviation industry. A book from which healthcare could learn from: 

Black Box Thinking: Why Most People Never Learn from Their Mistakes - But Some Do

Chat bots will come and help the healthcare services, but more lives will be saved with AI. It has been 50 years since the Beatles published "With a little help from my friends" in June 1967, and imagine how much help we are going to get from our little AI friends over the next 50 years.

Share this article

Robotic Process Automation Gears Up for Greater Innovation

RPA is a highly integrated means to apply automation to existing applications. Until recently, this has been relegated to low complexity, high volume, routine work. But a new revolution in RPA is coming that will change many parts of our daily lives.

 

What is RPA?

Robotic Process Automation, or RPA, describes the application of technology that “allows employees in a company to configure computer software or a ‘robot’ to capture and interpret existing applications for processing a transaction, manipulating data, triggering responses, and communicating with other digital systems,” according to the Institute for Robotic Process Automation and Artificial Intelligence (IRPAAI).

RPA is typically applied to use cases involving low complexity, high volume, and routine work. An article from Harvard Business Review titled, “The 3 Ways Work Can Be Automated” calls these “swivel chair” tasks, where data “needs to be transferred from one software system to another.”

From retail to banking, from IT to HR, and job functions that may span the organization, RPA offers an opportunity for firms to take certain tasks—payroll fulfillment, onboarding, benefits administration, invoice processing, file management, returns processing, call center processes, and data migration—and eliminate activities that are today performed by human workers.

RPA advancement has revolutionized IT, HR, and manufacturing industries. So advanced is our acceptance of automation in our daily lives—particularly with apps such as Alexa and SIRI offering personal assistance through our smartphones—that the next wave of RPA adoption is set to infiltrate some of our most skilled workforces, completing data, image recognition and diagnoses reports while ultimately changing the face of healthcare and legal practice. This paper focuses specifically on this next wave of RPA adoption.

Benefits: Why RPA?

Performing tasks swiftly and accurately forms the backbone of RPA business benefits. It also frees up employees to take on higher-value work and increases productivity across the board.

Outside of these advantages, RPA’s appeal lies in its simplicity to implement. As Gartner analyst Frances Karamouzis stated in an interview with ZDNet, it "just sits on top of the legacy system" without the need for special integrations. "They're also easy to use and have a relatively low cost. For all those reasons, [RPA] has by far the highest adoption of automation tools that we've seen."

These benefits, however, have also sparked fear among employees that the robots may indeed have come to take their jobs—though in many cases, the purpose of RPA is to re-focus workers on the kind of activities that cannot yet be automated.

According to CapGemini’s webinar report Exploring the Wave of Robotic Automation in Canada, the chief benefits for employees are:

·       The ability to focus on “value adding” initiatives in their daily tasks

·       Outsourcing monotonous, mundane, and repetitive tasks

·       Streamlining workflow processes while reducing human error

Key Challenges:  What’s holding back the enterprise?

According to IRPAAI, the key challenges of RPA include:

- Defining business cases with supporting Returns On Investment (ROI)

- Preventing scope creep and added complexity

- Automating inherently poor process elements

- Managing exceptions when they occur

 

Workplace Transitions: The role of change management in the adoption of RPA Tools

Fear of the unknown, and the threat of RPA technology replacing human workers, largely dominated the conversation on the emerging tech trend. Yet, as highlighted by an article in Forbes entitled, “This is How Mark Cuban Thinks Humans Could Trump The Rise of Automation,” the introduction of an RPA workforce will see an increasing demand in sourcing workers with “soft skills” and engineering backgrounds.

For this reason, change management is vital in securing a harmonious workforce while software is developed. At the 2016 Automation Innovation Conference, attendees marked the acceptance of human employees as the greatest indicator of success or failure at implementing RPA in a company. According to Casale, developing a “people plan” strategy to introduce RPA technology was a key focus in every workshop, panel session, and individual meeting held during the event.

An organization’s IT department will be instrumental in leading change management to create a digitized workforce. Although RPA may be a business-made decision—as benefits and low-cost profits have been clearly outlined—it is the algorithms and the ability to analyze and interpret data that will determine whether a role can be automated. This testing of software and the establishment of RPA excellence centers, shows a growing need to first harness the technology, and then implement it.

Conclusion:

RPA has possible applications in a number of industries; the potential to cut out monotonous tasks across a wide variety of roles, functions, and departments; and a relatively unobtrusive technical implementation with a shallow learning curve for those in the organization who rely on it.

As different organizations roll out RPA, a configuration that works for one may not work for another, from enterprise-wide setups where every function is incorporated, to smaller deployments with more limited scope. While automation of back-office processes, particularly in finance and HR, continue to be adopted, the IRPAAI predict that late 2017 will welcome a second wave of adopters who will signal a “takeover” of RPA.

Looking forward, layering cognitive automation into existing RPA structures may carry the potential for even more advanced use cases that get even closer to doing the work that humans alone perform today. But in the coming year, the major potential for accelerated adoption of RPA remains something to watch.

To get an in-depth insight into the way RPA will change the way business and customers interact, click here.

Share this article

The Era of Intelligent Automation

Mihir Punjabi, Principal Architect (Solutions)

New technologies now make Intelligent Automation a reality. Machine learning has moved from the realm of defense and financial industries to the mainstream. Even biometrics play a role in unlocking opportunities for easier, secure, automation. 

 

Automation and Artificial Intelligence have been around for years, but recently, they have been gaining a lot of traction. So what is new and why are automation and AI being mentioned so frequently??

The word “automation” alone spans many disciplines and industries. It could be a simple script to run few commands, or an advanced algorithm to drive a car, or a complex program to read my mind and automatically serve me a cup of coffee!

The degree of automation implemented thus far has been limited by the resources available. For instance, automation tools like Selenium are used to test the user interface (UI) of products, but tools such as Applitools have taken UI test automation to the next level by using machine intelligence. This has enabled automation of UI testing across browsers (Chrome, IE, etc.) and operating systems (Android, iOS, etc.), a once-cumbersome, manual process. These tools require screenshots or design plans to outline the UI frame appearance, and Image Cognitive Analysis may then be used to detect and highlight any subsequent anomalies.

Machine intelligence has opened up various avenues and opportunities for streamlining processes. Again, though machine learning has been around for some time, resources like compute power, deep learning algorithms, and intelligent platforms to develop advanced algorithms have not. Earlier, these resources were selectively available to the defense industry, as well as certain research projects and financial organizations. Now that they have been made more available to the software world, they have become a hot development topic. For instance, biometric authentication—that is, fingerprints and facial or voice identification—is now available in handheld devices like iOS and Android, and is even built-in to laptops and desktops with Windows 10.

While automation does save humans time, effort, and cost and prevents many errors, adding intelligence and learning capabilities to software results in a smart product that could auto-evolve: a robot watches another robot fall from the edge of a table and break. From this experience, it may learn that one should stop at an edge to avoid damage.

A good example of an intelligent ecosystem is Amazon Go. Amazon Go provides hassle free checkout while shopping. In fact, it claims that its store requires no checkout—just a walk in and walk out shopping experience! One only needs to scan the mobile app while entering the shop, pick up the desired items, and then leave. The subsequent amount is directly charged to the individual’s Amazon account. Going forward, I believe such systems could include something like biometric identification (users identified by facial detection, voice, etc.) for a similar type of frictionless experience. Additionally, it could learn user preferences and make product recommendations or even keep items ready for pick up, post-confirmation.

Another example is of Microsoft Windows 10. Through the Windows Hello framework and Companion Device Framework, a user may use biometrics or a registered companion device (such as a cell phone) to unlock the laptop or other hand-held device instead of entering a password or PIN: this latter option is also available.

This has become possible by making the ecosystem smart and connected. In the Amazon Go example, the mobile app communicates with the sensors in the store for the selected items and then generates the bill. The store ecosystem relies on the mobile ecosystem to identify and bill the user. In the Windows example, the mobile element has already authenticated the user and the laptop can communicate with the mobile and rely on the mobile’s user authentication mechanism to unlock the laptop.

The software has already become an intrinsic part of anything and everything these days; the next step is to make the software intelligent enough to create a smart world. A few examples:

Ø  Automobiles could be unlocked using an individual’s biometrics and can even auto-set certain preferences like temperature and music, etc. An individual could have specific gestures as their passcode for the second level of authentication. Perhaps the car could detect your mood—happy or sad based on your facial gestures or posture—and play music accordingly to soothe you!

Ø  A house could be unlocked using biometrics—the door just opens upon seeing me! Even the car’s authentication mechanism might be used here: the car communicates with the home’s front door, verifies itself (and the owner), and the door opens itself automatically. The home could then auto-set my preferences, such as the room temperature, music, etc.

Ø  Another example is wallet payments. Currently, this is an add-on to an existing credit card or online payment systems but in many ways, it can be seen as a replacement of the physical credit card. In fact, many companies are working on facial recognition for wallet payments.

Ø  Imagine, once the ecosystems are connected, it would be possible that your car could pay directly for the gas at the gas station without any intervention!

Ø  A very simple example in the corporate world would be to auto-arrange meetings based on a participant’s calendar availability and auto-generating the minutes of a meeting, along with action items, to be automatically added to the individual attendee’s calendars. Transcripts of the meeting could also be made available for future reference.

Ø  Another useful scenario could be when the user is driving a car and has to talk on his cell phone. The ecosystem will allow only one action—of course, with the option to override in case of exceptions. And, the frequency of exceptions would have an impact on the insurance premium!

Machine learning and Artificial Intelligence are being explored to solve issues that were previously unsolvable in the fields of healthcare, heavy industries (mining, oil and gas, marine, etc.), bridge building, cars, buildings, and so on. The possibilities are limitless!

The future is all about smart devices that learn, evolve, connect, and collaborate with each other to achieve an intelligent world. To smart connectivity, automation and beyond!

Share this article
Share this edition

Meet the Authors

  • Lanny Cohen
    Lanny Cohen
    Digital Transformation
  • Christopher Stancombe
    Christopher Stancombe
    Business Process Outsourcing, Finance & Accounting Operations
  • Denis Sproten
    Denis Sproten
    Business Intelligence solutions
  • Xavier Chelladurai
    Xavier Chelladurai
    Automation, Artificial Intelligence, Machine Learning, Predictive Analytics
  • Mihir Punjabi
    Mihir Punjabi
    Cloud, Embedded Systems, IOT, Machine Learning