Many people often asked about the difference between Robotic Process Automation (RPA) and Artificial Intelligence (AI). Some even confused the two to be the same.
To make matters worse, many vendors are now brandying about terms like Intelligent Automation (IA) or Intelligence Process Automation (IPA).
For the uninitiated, all these jargon can be very confusing, and perhaps daunting.
To help you out, we have put together this blog post to highlight the key differences between RPA and AI, particularly in the context of process automation.
Let’s get going.
IEEE Standard 2755
First, some definitions.
The IEEE Standards Association (IEEE SA), led by a diverse panel of industry participants, published the IEEE Guide for Terms and Concepts in Intelligent Process Automation in Jun 2017. The purpose of this standard is to promote clarity and consistency in the use of terminologies in this still nascent industry.
According to IEEE SA, RPA refers to the use of a “preconfigured software instance that uses business rules and predefined activity choreography to complete the autonomous execution of a combination of processes, activities, transactions, and tasks in one or more unrelated software systems to deliver a result or service with human exception management.”
And AI is “the combination of cognitive automation, machine learning (ML), reasoning, hypothesis generation and analysis, natural language processing and intentional algorithm mutation producing insights and analytics at or above human capability.”
Sounds a mouthful?
For simplicity, you can think of RPA as a software robot that mimics human actions, whereas AI is concerned with the simulation of human intelligence by machines.
Before we go into the differences between the two technologies, it is important to realise that RPA and AI are nothing but different ends of a continuum known as IA.
Doing versus Thinking
On the most fundamental level, RPA is associated with “doing” whereas AI and ML is concerned with “thinking” and “learning” respectively.
Or brawn versus brains, if you like.
Let’s use invoice processing as an example.
Your suppliers send you the electronic invoices by email, you download the invoices into a folder, extract the relevant information from the invoices, and finally create the bills in your accounting software.
In this scenario, RPA is suitable for automating the grunt work of retrieving emails (for simplicity, retrieval is based on the email’s subject), downloading the attachments (i.e. invoices) into a defined folder, and create the bills in the accounting software (mainly through copy and paste actions).
On the other hand, AI is required to intelligently “read” the invoices, and extract the pertinent information such as invoice number, supplier name, invoice due date, product description, amounts due, and many more.
Why is this so?
This is because the invoices are essentially unstructured or at best, semi-structured data. For example, different suppliers have different invoice templates and formats. There are also varying number of line items across the different invoices.
Since every activity in RPA needs to be explicitly programmed or scripted, it is practically impossible to teach the bot exactly where to extract the relevant information for each invoiced received. Hence the need for AI to intelligent decipher the invoice just as a human would.
To be sure, it is possible to handle invoice processing through RPA alone. In this case, we will deploy what is commonly known as attended automation.
Attended automation, or Robotic Desktop Automation (RDA), is like a virtual assistant that works hand-in-hand with your human employees.
Going back to our example, after the invoices have been downloaded, they will be passed through an Optical Character Recognition (OCR) software which will attempt to extract the required information. A human operator will then validate these information, before handing over the work back to the RPA bot to create the invoices in the system.
The key advantage, therefore, of using a RPA and AI solution is that you can achieve straight through processing (with minimal human intervention). The downsides are increased costs and project complexities.
Process-centric versus Data-centric
Another key difference between RPA and AI lies in their focus.
RPA is highly process-driven — it is all about automating repetitive, rule-based processes that typically require interaction with multiple, disparate IT systems. For RPA implementations, process discovery workshops are usually a prerequisite in order to map out the existing “as is” process, and to document them in the Process Definition Document (PDD).
AI, on the other hand, is all about good quality data.
For our example of invoice processing, we will concern ourselves with finding sufficient sample invoices to train our ML algorithms, ensuring our samples are of good quality (particularly if the invoices are scanned), making sure the invoices are representative of the data set, among others.
Thereafter, the task is to select an appropriate ML algorithm, and then train the algorithm sufficiently so that it is able to recognize other new invoices faster and more accurately than a human could.
Digital Stairways to Intelligent Automation
At the end of the day, RPA and AI are but valuable toolkits which you can use to aid your organisation’s digital transformation.
The choice of implementing either RPA or AI (or both) really depends on your specific use case, and ensuring “fit for purpose” is the key.
For the case of RPA, many organisations have cited reasons such as wanting to capture the “low hanging fruits”, quick implementation and time-to-market (usually in a matter of weeks or months), low costs and complexities, and others.
And many are making the smart bet of using RPA as the first step in the digital stairways to intelligent automation.
Hope this article has provided you with greater clarity on what RPA and AI is, and we look forward to welcoming you on your own intelligent automation journey.
What is your experience with RPA and AI/ML? Do share them in the comments below.