Future of Work News

RPA Needs Cognitive Learning to Truly Succeed

By

As businesses continue to look for new opportunities to operate more efficiently, AI and RPA use cases will continue to evolve.  Automation isn’t new, but AI is bringing it into a new era and offering new ways of improving operational efficiency.  But reducing the burden on employees to repeatedly perform the same manual tasks businesses, can free their teams to focus on critical components of their roles that can’t be replicated by automation.

It’s a market that is growing rapidly.  According to Gartner, it was the fastest-growing enterprise software segment, with a 60% growth rate in 2018.  Driven largely by digital transformation projects, automation solutions continue to evolve and mature, and will move into new models that far exceed the basic traditional RPA projects.  Gartner predicts a 30% increase in the use of RPA for front-office functions, like sales and customer experience.

Indico CEO Tom Wilde says, “RPA will continue to grow rapidly as it gains new users in small and mid-size enterprises.”

Indico provides Intelligent Process Automation (IPA) solutions to help organizations increase profits by helping them automate manual, labor-intensive, document-based workflows.

While there is tremendous opportunity for growth, there are challenges the market needs to overcome, including the low rate of AI project completion and RPA project expansion across enterprises.

“It requires organizations to re-engineer business processes, and that takes time, but RPA also has some built-in technological limitations,” says Wilde.  “While it’s great with repetitive, deterministic businesses processes involving structured data, it does not work when it is required to make judgments about information or learn and improve with experience.”

In order for RPA to truly reach its potential, businesses need to include AI/ML capabilities that add cognitive reasoning to the solutions that can handle unstructured data, which makes up the majority of enterprise content.  That’s valuable information that is critical to many processed, but isn’t being leveraged in automation projects. 

They also need to be designed from the outset with defined, specific business objectives.  Otherwise, projects become merely experimental, difficult to assess, and have little chance of success.

Another problem is the lack of explainability – there is an increasing need to understand the logic that goes into AI decision-making and to have an audit trail so errors can be traced back to their sources. 

“In our experience, about 80% of AI errors can be tied back to bad training data,” adds Wilde.  “The problem is finding it.”

He says the key is to tie every algorithmic outcome to training data to make it easy to trace the decision process and isolate the error point.  That’s important not only for quality assurance, but for regulatory compliance as well.

“The ability to understand how an algorithm is making decisions can surface potential problems,” he says.

Scalability is another issue.  While RPM is intended to increase operational speed and capacity, hard-coded rules make it so only pre-existing situations can be automated.  Because the data used to write the automation is based on existing experiences, current RPA solutions fail when new, undefined conditions arise.  Those may be very minor deviations from existing conditions, but enough to circumvent the automation.

In order to maximize its benefits, RPA has to evolve from hard-coding to include deep learning capabilities that not only have the ability to grow autonomously, but can integrate unstructured data into the cognitive decision-making process.

“By adding the cognitive ability of deep learning to process automation, users have the ability to automate a much larger percentage of business process decisions and minimize the manual intervention required in those processes,” says Wilde.

Indico has started seeing applications of deep learning in several use cases, including customer support analytics, contract analytics, regulatory compliance, insurance claims processes, and others. 

The latest trends in applying AI and Machine Learning technologies to business processes and experiences will again be the crux of the education at Future of Work Expo 2021.  Taking place at the brand new Miami Beach Convention Center, February 9-12, 2021 as part of the #TechSuperShow. Future of Work will again sit beside its collocated events – ITEXPO, SD-WAN Expo, MSP Expo, IoT Evolution, The Blockchain Event, and more – to deliver a compete learning and networking opportunity for business leaders who need to know what new technologies will drive their companies into the future.




Edited by Erik Linask

Future of Work Contributor

Related Articles

Workforce Logiq Intros IQ Location Optimizer for Improved Talent Sourcing

By: Stefania Viscusi    6/26/2020

As cities begin to reopen and offices start their return to work plans, there is now a need for talent acquisition solutions to assist with this new w…

READ MORE

Unified Office Awarded 2020 Teleworking Solutions Excellence Award

By: Ken Briodagh    6/19/2020

Total Connect NowSM Helps Customers Across a Variety of Vertical Markets Survive and Thrive in the COVID-19 Pandemic

READ MORE

Aruba Defines the Future of Work in a COVID-19 World

By: Erik Linask    6/4/2020

Aruba, along with many of it's tech partners, is launching a series of solutions to enable businesses to more effectively define their "new normal," a…

READ MORE

Quanergy Teams with Milexia for AI-driven LiDAR Solutions in EU Markets

By: Erik Linask    6/1/2020

Quanergy has partnered with Milexia to deliver AI-based LiDAR solutions to European organizations.

READ MORE

Study Shows Robots Tied to Income Inequality in Certain U.S. Regions

By: Laura Stotler    5/28/2020

A new study has found that the introduction of robots in the U.S. workforce has increased income inequality in certain geographic areas. The study als…

READ MORE