Down-to-earth offshoots of artificial intelligence are increasingly accessible for digital transformation work. These projects tapped into machine learning – with existing talent.
Robotic Process Automation (RPA) vs. AI, explained
What is the difference between RPA and AI? How do RPA and AI work together? How does machine learning fit in? What are some RPA and AI use cases and best practices? Let’s break it down in plain terms
The expanding universe of artificial intelligence includes many terms and technologies. That naturally leads to overlap and confusion. AI and machine learning are mentioned together so often that some people – non-technical folks especially – might think they’re one and the same. They’re related but not actually interchangeable terms: Machine learning is a subset, or a specific discipline, of AI.
That’s a relatively straightforward example. Start adding other terms and technologies into the mix – deep learning is yet another subset of machine learning, for instance – and the opportunities abound for further misconceptions.
Deciphering the differences between terms and technologies takes a twist with robotic process automation (RPA) and AI. We’re going to break it down so you can explain it - even to non-technical audiences.
[ See our related post, How to explain Robotic Process Automation (RPA) in plain English. ]
What is the difference between AI and RPA?
In fact, there’s some disagreement on the basic relationship between the two. While just about everyone agrees that machine learning is an AI discipline, that same consensus doesn’t exist with RPA and AI. As our research report “Executive’s guide to real-world AI,” produced by Harvard Business Review Analytics, notes: “Some question whether RPA qualifies as AI.” (The report goes on to point out that by the AI definition it uses, RPA does meet the criteria.)
One reason for this is lack of consensus is that RPA technologies and use cases to this point haven’t been all that “intelligent.” RPA can do a great job of handling repetitive, rules-based tasks that would previously have required human effort, but it doesn’t learn as it goes like, say, a deep neural network. If something changes in the automated task – a field in a web form moves, for example – the RPA bot typically won’t be able to figure that out on its own.
Still, there’s definitely a relationship between RPA and AI, even if you’re in the camp that thinks RPA does not actually qualify as AI. And that relationship is growing.
“AI technologies that augment and mimic human judgment and behavior complement RPA technologies that replicate rules-based human actions,” says Chris Huff, chief strategy officer at Kofax. “The two technologies work hand in glove, just like traditional ‘white-collar’ knowledge-based workers and ‘blue-collar’ service-based workers collaborate as the engine to drive productivity for an organization.”
[ Want lessons learned from CIOs applying AI? Get the new HBR Analytic Services report, An Executive’s Guide to Real-World AI. ]
How RPA and AI work together
In a sense, it doesn’t matter whether you or I think of RPA as a specific branch of AI or not. What matters is that we’re aware of how the two technologies are increasingly going to work together – hand in glove, as Huff puts it.
Dave Costenaro, head of AI R&D at Jane.ai, notes that RPA has already made some significant advancements in recent years: “RPA had a major leap forward with the advancement in past years of cloud-based APIs and common data formats, which enabled all kinds of services to talk to each other in automated workflows.”
Now, as it gets deployed in concert with AI technologies, RPA is set for another boost in capabilities.
“The presently booming AI technologies – namely, deep neural networks – are adding brand-new tools to the RPA toolbox primarily in vision and language tasks,” Costenaro says. “Now, the RPA workflows can be enabled by these capabilities at decision nodes where they could not before. This allows documents and images to be ‘viewed’ holistically by an algorithm, and interpreted for downstream logic and routing.”
So as RPA gets paired with AI disciplines such as natural language processing or computer vision, the possibilities for effective automation grow considerably.
“Converging AI with RPA enables businesses to automate more complex, end-to-end processes than ever before, and integrate predictive modeling and insights into these processes to help humans work smarter and faster,” says Kashif Mahbub, VP of product marketing at Automation Anywhere.
What is intelligent automation?
Mahbub and many others refer to this convergence as “intelligent automation,” a term you’re likely to hear more often going forward. This is essentially the “digital worker” – the software bot or machine that functions like a human employee – coming to life.
“As AI algorithms become more sophisticated, software bots can transition from automating specific processes to functioning as fully cognitive business assistants capable of automatically tackling all kinds of repetitive tasks in real time and ultimately freeing humans of mundane, repetitive work,” Mahbub says. “Just as the agricultural revolution saw humans transition from farming seven days a week, this Fourth Industrial Revolution could do away with the five-day workweek and allow us to spend our time at work on what humans do best – thinking creatively.”
If that sounds kind of, well, big – that’s because it is, and it’s why so much attention is being paid to these topics worldwide.
RPA and AI use cases
But to ground it in a more present-day reality – one that IT leaders know well – the increasing convergence of RPA and AI means that actual implementations and use cases will actually be able to catch up to some of the overhyped sales pitches some vendors pushed in the earlier days of RPA.
“RPA automates tasks. RPA does not automate full end-to-end processes,” Huff says. “Unfortunately, some customers were sold RPA under the guise of automating end-to-end processes and are beginning to experience buyers’ remorse [as a result].”
Huff sees two particular technologies that are arising out of this catch-up effort to mitigate that buyers’ remorse, both examples of the integration of RPA with more cognitive capabilities:
- Cognitive capture: “Cognitive capture focuses on ingesting data via omnichannel – i.e., web forms, paper documents, emails – and then using native AI/cognitive algorithms to transform unstructured data into a structured format so RPA can begin to automate the work tasks,” Huff explains.
- Process orchestration: “Process orchestration adds rigor and discipline to automating the workflow. And since work tasks automated by RPA are typically part of a workflow, this just makes good sense,” Huff says. “But process orchestration also helps RPA by handling all the exceptions, performing traditional dynamic case management, and most importantly, managing the collaboration and work hand-off between RPA digital workers and physical employees.”
Huff offers some advice for evaluating different options that offer a convergence or collaboration of RPA and AI: “As you look to invest in these technologies, be mindful that some companies provide more pre-built integration out of the box than others.”
[ Can AI solve that problem? Read also: How to identify an AI opportunity: 5 questions to ask. ]
RPA implementation best practices
Huff also recommends a couple of other best practices for getting the most out of an integration of RPA and AI.
1. Focus on outcomes
As with most major technology trends, simply following them for the sake of it is not likely to yield strong results. You need clear goals or outcomes.
“Outcomes are typically best achieved through effective governance that properly identifies where the technologies should be deployed and then continuously monitors the outcomes against early hypotheses or business case metrics to determine ultimate value,” Huff says.
2. Treat the converged RPA and AI technologies like digital workers
“Another trick to effectively deploying the technologies is to do so in a holistic manner that views the technologies as ‘digital workers’ that simply have different skill sets, much like humans,” Huff says. “When viewed as a ‘digital workforce,’ organizational functional units begin to treat the technology as another persona in the workspace that empowers them to do more in less time, which should translate into more capacity for the organization and improved work/life balance for employees since we’re giving the individual the one thing they likely value the most – time!”
RPA vs. automation: Plenty of room for humans
Indeed, this is “big” stuff – the pairing of RPA and AI may bring significant moves forward toward some of the loftier predictions about automation and the future of work. But to stay grounded, it’s also worth reminding ourselves that the paired technologies also simply represent the growth of current capabilities of RPA on its own. So don’t count us humans out just yet.
“The net effect will be to expand the reach of RPA for higher-productivity workflows,” says Costenaro from Jane.ai. “When this wave of expansion matures, the technology will run up against a new boundary of synthesis, creativity, and strategic thinking. These are areas with no clear research path toward automating, so humans will continue to perform these tasks, boosted and aided by RPA workflows upstream.”
[ How can automation free up more staff time for innovation? Get the free eBook: Managing IT with Automation. ]