Turning Machine Learning Into Intelligence That Matters

Jean Loh

Part 2 of a 2-part series. Read Part 1.

Great news for employees: Travis Tompkins, senior manager at Deloitte Consulting LLP, predicts that companies will soon be removing 20%–30% of all mundane activities from their employees’ to-do lists. That means more time to do real work and think.

“Automating paper shuffling, workflows, and simple decision making will give us extra time and resources. Now companies must decide what they are going to do with that additional capacity,” he said on the May 10, 2017 live episode of Coffee Break with Game Changers Radio, presented by SAP and produced and moderated by SAP’s Bonnie D. Graham.

He was joined on a thought leadership panel by Dr. Paul Pallath, chief data scientist and senior director of the Advanced Analytics organization at SAP, and Kevin McConnell, director of Machine Learning Solutions Go-to-Market Strategy at SAP.

Travis’s bold prediction may become reality by 2027, as we’re already seeing business leaders climb on the machine learning bandwagon. They are harnessing text and speech recognition, visual image processing, chatbots, and digital assistants – and relieving employees of repetitive and boring tasks so they can act more strategically.

Click to listen to the full episode.

Reframing the most critical questions

The more companies transform with innovative technology, the more their appetite will grow to digitize every aspect of the enterprise. “Everybody wants to apply the predictive capabilities of machine learning in their daily operations,” Paul stated. “When I talk to customers, I always advise that the business problem must address the top line or bottom line. Otherwise, machine learning will not change how customers interact with their systems. Executives must first ask the right questions to create a significant difference.”

Will machine learning answer the most fundamental concern in today’s boardroom: “What can we do differently?”

“Machine learning is not about letting machines run the world; it’s having machines help facilitate how humans run the world,” observed Kevin. “The best place to implement machine learning is where there are 5 million different possibilities. People can typically look at not more than 10 or 20 scenarios. But we still need to set the parameters, evaluate results, and apply those findings, even though machines are going to do most of the thinking.”

Travis offered a use case. “The advent of machine learning is fine-tuning finance activities – from how performance is measured to how insights are analyzed.” Then finance organizations can better predict, plan, and forecast by evaluating thousands of variables and understanding the root cause of profit growth or loss. But more important, this level of insight enables the function to deliver a critical lifeline for the entire business’s survival.

More use cases: data security and more

The panelists agreed that the top use cases for machine learning include the security of personal data and intellectual property. Across networks, computers, and anything in between, “there are thousands and millions of threats a day,” Kevin noted. “Machines must start asking for the right rules and how to make them smarter.” The highest payback for investing in machine learning is found improving data security, personal security, fraud, and credit risk.

Paul added, “If businesses categorize various use cases, then sales and marketing operations, risk management, finance, and human resources are the predominant functions that could find success.”

How to build machine learning algorithms to address a specific need? “Companies must start at the top level of the organization and ingrain the technology in the strategy,” Kevin advised. “A common mistake is to start putting machine learning in lots of different places. The expense of resource and data replication does not bring a true advantage. Instead, executives should focus on key performance indicators or strategies.”

But Travis cautioned, “When businesses first get started with machine learning, they must be willing to fail – learning in that failure process how to build better models during every iteration.”

Paul added that the organizational culture should adapt as well. “In large deployments of machine learning in business processes across all functions, it is very important to have a data-driven organization. For the huge segment of the workforce that has yet be impacted, the first thing that will come to mind is, ‘Do I trust the systems that propose actions for me to look into, versus my having to evaluate all the options?’ The culture may need to be molded into being data-driven, rather than having something forced on them.”

Will machine learning become the new norm?

According to the panelists, machine learning will become more than just another tech buzzword that disappears over time.

Travis predicted that machine learning will pave the way to a more pervasive form of artificial intelligence: natural language processing. “We’re seeing it now with Google Home, Amazon’s Alexa, and other devices we’re talking and interacting with very naturally. I envision that, five years from now, we’ll view this innovation area as we look at GPS today. It will become very natural.”

Paul went further. “I predict there will be a machine-driven, completely automated world that we’ll be a part of. People will be interacting with this digital wall seamlessly without understanding that we are still existing somehow. Maybe the machine understands my responses to your questions and would be taking this call for me,” he mused.

Kevin’s perspective: “You’re going to engage with things a lot differently than you do now, and it will start with machine learning. I may come home on a Friday night and there will be a pizza waiting at my door. The machine will already know that I’m going to order it at 7 o’clock anyway. All those things are going to happen, and we’ll joke about the days when we had to do things.”

Learn more

Listen to the SAP Radio show “Machine Learning Trends – Part 2: Harnessing AI” on demand.

Automation is THE priority for global business organizations that want to drive costs down. Join Randy Garrison of SAP and Weston Jones of Ernst & Young, LLP on July 17 at 1 p.m. EDT to understand what robotic process automation can offer you today and in the future. Register now!


Jean Loh

About Jean Loh

Jean Loh is the director, Global Audience Marketing at SAP. She is an experienced marketing and communication professional, currently responsible for developing thought leadership content that is unbiased and audience-led while addressing market challenges to illuminate and solve the unmet needs of CFOs, CIOs, and the wider global finance and IT audience.