We, humans, love the human form. We anthropomorphize everything from Hello Kitty to elaborate machines.
For example, the modern robot began a century ago as a fictional humanoid in 1920 by Czech playwright Karel Čapek. The public face of robots remains a human form — exemplified by child-size forms with oversize eyes rolling around the floor during early 21st-century trade shows.
We like our robots cute and humanoid, but robots are machines that must be programmed by humans. However, the word connotes something alluring, which is why it’s part of robotic process automation (RPA), a buzz-acronym resonating with analysts and vendors. But what about with chief digital officers?
Nuts and bolts
Webopedia defines RPA as “a way to automate repetitive, time-consuming tasks using software with machine learning (ML) and artificial intelligence (AI) features.”
“Businesses use RPA in multiple aspects of operations to improve efficiency and reduce instances of human error,” says Webopedia. “Some common examples of RPA include performing transactions, making queries, running calculations, and processing information requests.”
Let’s unpack this. The concept of reducing human errors is attractive from both an operational and a bottom-line standpoint. It’s the ultimate goal of robotics, but business processes require human intelligence and flexibility. Machines are fast, but contextual decision-making is a stumbling block.
Automation reduces human error in simple tasks, and the more advanced the automation, the faster those tasks can be accomplished. Without, that speed is limited only by other processes in the chain. The lack of a physical component delineates a chatbot, for example, from an assembly-line robot. The algebra of an industrial robot morphs into calculus when tasks are performed at electric speed — the potential use of bots is limited only by human imagination.
Add in Moore’s Law, and it’s easy to see why bots are increasingly common in everything from customer-facing tasks to large-scale back-of-house deployments. Bots excel at computing processes involving large-scale repetitive tasks.
But what’s the business case for RPA?
Obstacles remain, says Craig Le Clair, vice president, principal analyst serving enterprise architecture professionals at Forrester.
“Deployment obstacles [includes] application development, resilient automation, digital worker analytics, and cloud advancement,” wrote Le Clair in a research note. “To fully realize value, RPA must...evolve as a general application development and management platform to support a wide variety of digital assistants, domain-specific robots, RPA digital workers, and employee robots.”
The Forrester vice president provides a granular example: financial advisors using “domain-specific bots in the background to advise on complex tasks like retirement rollovers and setting up new accounts. These RPA bots add conversational AI, act as data and application intermediaries, and leverage the advisor’s data and permission structures.”
Le Clair lists two innovations to expand platforms to this potential: “dynamic creation of robotic scripts based on recorded human activity and machine learning, and low-code principles, such as robot assembly with functional blocks or fragments.”
Much like their human counterparts, RPA bots need support and maintenance. “Infrastructure-related issues, software reliability, application UIs, and data changes cause bots to break more than they should,” says Le Clair. “Firms struggle with resiliency — keeping bots running after an incident. This makes for high maintenance costs, with bots often down for a day or more. Selecting a task that’s too complex for a bot is often the root cause.”
Bots run amok
It’s clear that while RPA holds tremendous potential, firms seeking to deploy this level of technology must pay attention to due diligence and tech architecture. And while organized use of bots progresses, the word “bot” holds negative connotations for many.
Author Richard Kadrey, who created the Sandman Slim series, finds that bots enable organized theft of his IP. “Myself and many others writers have discovered phantom books popping up on Amazon: bots run around the Web scraping any free stories or articles we’ve posted,” says Kadrey. “The bots then compile them into ‘instant books’ and sell them for a few dollars, keeping all the money for themselves.”
“I try to keep up with them, but there are too many,” says Kadrey. “Trying to take down all the phantom books of my work could easily become a full-time job.”
On a more positive note, bots don’t just scrape content and collate it for sale — they produce content as well. “AP also uses automation technology from Automated Insights to produce recaps of all MLB-affiliated minor league baseball games as well as nearly 4,500 stories covering U.S. corporate earnings each quarter,” said Associated Press in a 2018 press release.
“AP’s sports report first began using automation technology approximately six years ago and now provides most of its sports agate to subscribers through automation, primarily from Sportradar,” said the statement. “AP rolled out automated corporate earnings reports in 2014.”
“Finance and sports are the usual targets of robot reporting,” wrote Francie Diep in Popular Science magazine in 2014. “Both are a bit robotic by nature. The most basic reports involve plugging numbers from a database into one of a few standard narratives.”
Orchestrating the automation
Forrester’s Le Clair says that RPA should “become a hub for automation orchestration. End-to-end automation requires that diverse platforms and emerging technology components work together.
For example, chatbots gathering a human’s intent will trigger a bot transaction in a core system or enlist machine learning for a decision,” he says. “True RPA has a strong architecture to schedule and integrate external automation services.”
Image credit: iStockphoto/yaruta