Q. 108  What is the difference between Artificial intelligence (AI) and Robotic Process Automation (RPA). Elaborate with suitable examples. Can the highest level of automation be achieved without the two working in tandem?

 

 

Note for website visitors - Two questions are asked every week on this platform. One on Tuesday and the other on Friday.

• All questions so far can be seen here - https://www.benchmarksixsigma.com/forum/lean-six-sigma-business-excellence-questions/
• Please visit the forum home page at https://www.benchmarksixsigma.com/forum/ to respond to the latest question open till the next Tuesday/ Friday evening as per Indian Standard Time.  
• The best answer is always shown at the top among responses and the author finds honorable mention in our Business Excellence dictionary at https://www.benchmarksixsigma.com/forum/business-excellence-dictionary-glossary/ along with the related term.

 

RPA can be used where we have a define set of rules for processing that is the information is in digital format and structured data.

AI is used when the information is in an unstructured data where you can feed in series of decision making recorded by human to make future decision by the machine. Increasing the number of practice cases will make AI for a better decision.

Examples for RPA can be automatic payment chasers for pastdues of credit cards, Rule based transaction routing to various queues, Sorting of document and processing them on system which are rule based.

AI can be for unstructured data. e.g invoice information format will be vary from one customer to another however the type of data will be the same. We can feed in various types of invoices to be learnt by the machine to obtain common information ( Seller name, Buyer name, goods description and unit price), sorting of scanned  customer document/Swift messages based on content of the data to various Queues if not to miscellaneous Queues.

Highest level of automation can be achieved without the two working in tandem will be based on the process whether it can be done through rule-based where  AI will not be required and if the process has combination of both  structured and unstructured data using AI and RPA will be a signification shift.

RPA: Robotic process automation is intended to automate repetitive, low-level tasks by mimicking human actions and behavior. It may be suitable for mundane tasks that require little or no human interventions such as entering purchase invoices in an ERP system, setting up a new customer account and others. RPA is a preconfigured software that works on a predefined program to complete autonomous execution of a combination of processes, activities and tasks.

 

AI: AI on the other hand is used for automating tasks in a much efficient manner. It is capable of delivering better results as it has the capability of rewriting themselves in response to their environment. It can make decisions and predictions either based on rules (as RPA) or based on numeric parameters achieved via machine learning.

 

A key difference between AI & RPA is that the latter is used for running rule-based processes and requires structured inputs and AI is leveraged for unstructured inputs. Another key difference is that RPA software is trained exactly for the task, for example SAP’s Process Automation software.  

 

Key Differences:

Training Method

RPA: RPA is rule-based. It works on a set of instructions, or a set of rules and performing tasks accordingly. It uses a set of statements or steps to define a repetitive activity which it does by using algorithms to automate it. It is the systematic and organised inputs that RPA depends on to deliver desired results. As it is rule-based, it has no inbuilt intelligence and is prone to errors given its limited scope of intelligence. For instance, in a bill management system, if it can automatically record printed bills or automatically communicate customer issues, then RPA may be playing a key role.

AI: AI is much more than just rules. It is data-hungry and requires a large amount of data to train. The data could be anything from customer information to images that helps machine understand the underlying concept. AI systems have the capability of learning over and over again using mathematical and statistical methods. Apart from ML, NLP is also used to deliver AI capabilities, which allows a machine to acquire an understanding of the human language.

Process-Driven Vs Data-Driven

RPA: It is highly process driven, meaning, it is all about automating repetitive and rule-based processes which typically requires interaction with multiple, disparate IT systems. Every activity needs to explicitly programmed, exposing it to the risk of delivering effective performance in many cases. Moreover, RPA cannot work on unstructured information.

AI: AI, on the other hand, is all about good quality data. It would be unfair to say that AI is all about ‘thinking’ as opposed to RPA which is all about ‘doing’ a task. AI comes in handy when there is loads of unstructured data to deal with, and can easily manage variability in data to get better with time, based on its own experiences. For example, image recognition, text recognition or search are some of the most mature AI applications seen in businesses today.

Learning and Thinking Capabilities

RPA:  Robots or machineries activated with RPA will do exactly what you ask them do and in the same way all over again, every time.

AI: AI on the other hand is expected to perform a judgement based processing. This means that AI trained programs will act based on their learning from past data and trends. It can manage and understand patterns and trend over time. The bottom line is AI is an excellent self-learner and is good in capturing information such as vision recognition, sound recognition, search, data analysis and others. In short AI is where machines are trained to think like humans and possess the ability of rationalizing and take actions accordingly.

The Human Involvement

RPA: RPA is practically a software that reduces human efforts and compliments their work. As RPA mimics the steps followed by humans, it is often programmed to relieve human workforce from mundane activities, so they could focus on other important activities to accelerate business growth. Having said that, it often requires human intervention to keep the robotic processes constantly updated.

AI: AI on the other hand has the capability of eliminating human effort to a significant extent. AI comes to a rescue when RPA fails and may not demand constant human intervention after the initial process of setting it up.

Conclusion

If we are looking to opt for either of the two processes, it is always wise to first analyse the nature of your process—decreasing turnaround time, saving cost, accelerating process, among others—and then decide on opting RPA or AI or a combination of both to achieve an extremely powerful result.

 

Examples of difference between RPA and AI:

 

A supplier sends 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

Artificial Intelligence is a technology so powerful that it competes with human capabilities and intellect. It is an intelligent machine that mimics human mental capabilities like cognition, reasoning, problem-solving etc. to learn, think and take decisions to maximize success of the defined action.

Let’s understand this better with an example

A good example of Artificial Intelligence is commercial promotions and friends suggestion on Facebook, where through analytics of profile data usage and recent browsing history , AI cleverly and selectively promotes products and suggests friends. Other good example is use of Artificial intelligence is in the health industry to diagnose critical illness with utmost accuracy. Recent research shows that results acquired post-clinical test in medical field with the help of AI is accurate and error proof as compared to human driven tests.

Machine learning (ML) and Natural learning processing (NLP) are enablers of self-development capabilities in AI.

 

Robotic Process Automation or (RPA) is a rule based software/robot that mimics repetitive human tasks which are repetitive and time consuming exactly the same way, based on “if-then” commands. This automation does away with human dependency, with no changes in existing system integration. It works on sequence of commands, for software/robots under some defined set of business rules.

Let’s understand this better with an example

A good example of RPA is Customer Experience Management (CEM) tool used in by many Telecom companies, which has reduced the human intervention for network complaint handling and trouble shooting. Logics are built to capture network experience for each customer connected to network sites and accordingly the code based system autosuggests solutions. CEM has enabled frontend to effectively attend to customer queries/complaints and provide on time resolutions. Another example is IBM Watson which auto troubleshoots for any queries/complaints based on input received from frontend .It has completely removed the manual task of navigation between different applications to derive any solution/conclusion at customer touchpoints.

 

Let’s note some key differences between AI and RPA

 

Differences	AI	RPA
1.       Self-learning Vs Command Oriented	§  AI is ‘self-learning’ §  AI is capable enough to self-learn and evolve through experience	§  RPA are ‘Command oriented’ §  RPA works on set of instructions to automate a rule-based task
2.       Self Sufficient Vs Support System	§  AI is capable enough to perform human task with perfection and completely remove human intervention.	§  RPA is a software/robot used to reduces human task and help us concentrate on more creative work
3.       Decision Making Vs Execution	§  Works on concept of systems that “Think”, “Learn” & “improvise task §  Technology used in case of ambiguity or decision making requirement, §  Used usually when process/task scenario are unstructured (converting voice message to text) §  Deals with large amounts of data, as it can analyze & manage that unevenness.	§  Works on concept of “Do”. § Software/robots will do exactly what it is designed for, in same way every time. §  Work on rule-based processes where compliance and accuracy are critical
4.       Agile Vs Monotonous	§  AI algorithm can adapt to a new environment, learn from the outcomes of decisions and improve itself over time.	§  RPA works constantly on the same logic for ages unless commends are changed.
5.      Simplicity	§  AI is bit complicated. §  Needs changes in existing system integration.	§  RPA is the relatively simpler technologies. §  Works with existing system integration.
6.      Technical Barrier	§  AI requires programming skill sets	§  RPA has low technical barrier, as programming skills are not necessarily required
7.      Analytical capability	§  AI can process large data sets and analyze to make decision making easier	§  RPA lacks this ability

 

Conclusion

RPA and AI are two separate technologies, but complement each other very well. By combining the potential of AI and RPA a completely autonomous process can be achieved. RPA along with  AI and machine learning capabilities can handle high-volume, repeatable tasks with effective solution and deliver precise outcomes. Synergy of RPA and AI can help in natural transition from automation to intelligence. As AI adds value to RPA, in order to achieve the highest level of automation, these two technologies needs to work hand to hand.

In general, we can brief Robot Process Automation is a subset ( one of the unit) of Artificial Intelligence. RPA is widely used in many manufacturing & non manufacturing firms in India for few years. It is the replica of human actions for completion of particular task or activity, techinally some therbligs for motion economy (like pick & place, load & unload, assemble or disassemble, transport load & unload).In general, many organizations adopt RPA only if it is maximized ROI with minimal payback. RPA reduce human intervention at areas where repetitive tasks & high human fatigue with physical strain take place in manufacturing areas.

 

On other hand, Artificial intelligence is a broad picture that enables & demonstates machines to maximize the chance of achieving organization goals by enable simulation of human cognitive skills towards independent decision makings based on historical data without human intervention. In recent days, AI is mostly depoying at SHE ( Safety, Health & Environment) related tasks to monitor & track where decisions are made by considering external business risks mostly.

Simply, RPA is related to human hands& its motions & postures derived by if then rules, where as AI is related to human minds& emotions to exhibit independent & effective decision making, corrective& preventive action planning, effective root cause identifiaction, key stoke analysis of real time data ( behaviour of 6M's over PQCDSM at any interval of production schedule).

 

In recent days, to maximize OLE (overall labour effectiveness)& operational efficiency, RPA is most likely choice over AI based on the internal & external business risks at present scenarios.

 

We can acheive highest level of automation by combined RPA& AI where RPA focus on external improvements like manpower reduction & maximized machine utilization, optimized material flow & AI can be enable to reduce or eliminate process variation and identifying root cause(s) & sustenane of solutions. Both together will give highest impact to business value to organization if we select and deploy at the right place & at right time.

For examples, in service industries, if you want to reduce multiple approvals, we can adopt RPA for electronic signature with human touch. in other case for the same objective,  if It combines with AI, it can sense,  verify, finalize, approve without human intervention.

Definitions:

Intelligence: Per English Dictionary, it is the ability to learn, understand and think in a logical way about things. In general human beings shows that intelligence predominantly and distinctly, when compared with the rest of the species in the Earth.

 

Then What is Artificial Intelligence? A very simplistic definition : If the same intelligence is being shown by a machine, or a computer, then it is Artificial Intelligence (AI). Therefore, AI essentially simulates the human intelligence.

 

What is Robotic Process Automation(RPA)?
Simple Definition : As a software robot, it apes or mimics the human actions.

RPA is all about the habit of doing automation of your day-to-day business processes by a s/w robot as automatic tasks.  

 

Let us compare and Contrast RPA and AI.

 

In my opinion, i feel the points described here , are some of the key differences.

 

1. RPA needs structured inputs and/or well defined rule-based human processes whereas AI can handle even unstructured inputs/data(like email, audio, video files,...),semi-structured inputs/data (XML,Java Script Object Notation(JSON,....)
2. RPA cannot take decisions on its owns and aligns with what is programmed whereas AI has the potential intelligence to take decisions, depending on the needs.
3. RPA is 'dumb' in that it does what it is supposed/programmed to do, while AI is 'Self-learning' and can adapt itself as per the situation.
4. RPA does not possess cognitive skills (refer #3 for the reason) whereas AI has that Cognitive skills.
5. RPA has some limitations in its design as shown by aforementioned points whereas AI does not possess any specific limitations.
6. RPA 'does' things whereas AI focuses on 'thinking'
7. RPA is relatively less complex when compared with AI since AI has many branches/fields (like Machine Learning, Pattern Recognition, so on...)
8. RPA might be relatively less steeper to learn when compared with AI, primarily because of #7.

 

Benefits of RPA:
1. Good in doing routine/repetitive tasks.
2. Eliminates manual errors(human induced) and unnecessary rework/delays and therefore increases the productivity.
3. Ensures 'Lower Turnaround time'.
4. Cost Efficient
5. Provides Better ROI
6. Helps the Stakeholders in focusing solely on strategical aspects alone (thats the very purpose of RPA . Is it not ? - to alleviate the workers from doing routine tasks).

 

Benefits of AI:
1. Maturity of the AI software can make the decision making easy to stakeholders/mgmt.
2. Helps in providing accurate and precise information to the organisation
3. Identifies patterns/trends based on past data(historical).
4. Provides data on things which otherwise is difficult to get by human beings.  

Let us see some examples for RPA And AI and portray their Capabilities.

 

RPA examples:

1. Imagine few processes such as cheque handling process and cash transaction process in a bank. You want to ensure that all manual operations/transactions/movement

involved in this to be automated by RPA. This will be a good candidate for RPA.

2. Another specific example: Imagine a Money counting machine. In old days, the cashier used to count the money and used to give us the amount that we had specified (say using a withdrawal slip). The same task is done by the money counting machine and given to us. This is a typical RPA stuff.   

 

AI Examples:

1. Alexa , SIRI , Ok Google (Google Assistant) are popular AI products from Amazon. Apple, Google respectively.  
2. Imagine you have some photos of yours in your mobile. You want to show your kids all the photos that you are in. Now you choose one photo of yours and when you  click the details of that photo, it will show the rest of your photos in your mobile.  This is nothing but a facial recognition of your photo which is an AI aspect.
3. IBM's Watson is an example of Cognitive Computing of AI. It is used across industries and very useful for health industry.
4. Take the game of chess. Remember 'Deep Blue'. The famous chess computer had a pitched-battle with the then World Champion , Garry Kasparov. The AI machine might say

take account of algorithms like depth-first or breadth first as required. Professional players practice a lot with such AI chess systems to gain more insights into the nuances of the game and thereby in the process, fine-tuning their specific needs.
5. Shop sweeping bots
6. Drones can be controlled by AIs , which can collect critical information in war affected zones.

 

Now we have seen benefits , examples, charactersitics of both RPA and AI.

 

Let us see what can happen if both can be put into use together.

 

Before doing that let us do a mimicking of these two aspects , with some imagnition.

 

Since RPA is dumb,i will equate it to 'animals'. Since AI is having the ability to Self-learn, i will equate it to 'Human Beings'(only for this article :-) ).
Now imagine Animals vs Human Beings. Can you tell who is powerful ?  Think Elephant, Lion, Tiger, Cow, Dog..as animals. Can you handle them 1-1 without any weapons(if

you face any of these 1-1, when if any one of them tries to attack you). Just for your support, imagine there is an elephant trap nearby and you are too close to that. You know if you run near that , the Elephant (if it is planning to attack you) can fall into that trap. You have 'thinking' ability has your biggest weapon here.Now imagine you have a physical weapon and you are holding a wood with a fire in it(Tranqulisers are not for you unless you are a forest official!!) to drive away the animals.  What do you understand from this ? Both Animals and Human beings have their powers. You therefore , definitely want to harness both their powers. How many times,you would have seen 'Tarzan' movie, where Tarzan and his animals work in tandem and succeed with great and right results efficiently !! Now imagine replacing that 'Animals' and 'Human Beings' with the actual powers - RPA and AI. Don't you want to get this power. Well RPA and AI does this , when combined together.

Let us some examples where RPA and AI systems can work in tandem (the answer could be theoretically almost in all spheres - but though depending on an Organisation's

needs)

 

1. AI comes into picture where RPA falls short. Imagine cheque clearing, cash dispensing processes. Any routine tasks which are repetitive in nature, can be done using RPA, doing Optical Character recognition is part of AI. This is just a very basic example of combined usage of RPA and AI.
2. Similarly this can be the case on invoicing systems or any other systems where the need is there.
3. On systems where decision making is needed for the data obtained, post the business automation.

 

Now to the question of achieving the highest automation without the two working in tandem, is that possible?
The answer may not be 'yes'.  We have seen above, both the imaginary and the practical examples. But there may be some scenarios where RPA alone might be suffice which

may not need AI presence. It could be because the amount of investment(in terms of energy, effort) made for having that AI aspect may not be worth considering the information that you get it for that specified entity for which you do RPA and AI. If ROI is not enough for a given activity, the organisation may not think it worth.
In that case, the organisation could be comforted with RPA alone. For the record, to bring about a change in the productivity,quick turn around time,some teams/organisations use the approach of implementing RPA first and see the ROI drastically changed. Then they go with AI combined with RPA. But nevertheless i feel, bringing both the aspects to the party will produce the maximum result.

 

Conclusion:  

RPA and AI are two seperate aspects which however mostly can coexist in today's growing business needs. Both are at different points of the automation phenomenon.  RPA has its own characteristics and AI has its own flavour. AI is very deep in its branches . Both RPA and AI have a history . RPA had its initial origin which was called Screen Scrapping and AI has been there for quite some time. Both have evolved. While RPA helps in doing effective automation on repetitive tasks, AI steps in where RPA is lacking - cognitive learning,self-adaptiveness, taking independent decisions (ability to deduce information and act upon) and so on,... Caution has to be put in place, to ensure that right data is captured by AI for processing information, for safety and security reasons. The language of use for RPA and AI can be decided based on the organisational needs or the team comfortness. It can be Java, LISP, C/C++, Python or any other AI or RPA based languages. Most of the organisations see rapid change in their Productivity, turn over time and have a great ROI and excellent customer satisfaction , by using RPA and AI. Imagine the superior performance of the Money counting machine and the way how you get response from Alexa/Google Assistant/Siri. Are we not happy with these things ?  How about the Unmanned cars driven on the road ? Welcome to the world of RPA and AI!!

 

 

 

 

Artificial Intelligence is the system that uses a set of images as a training data to help machines understand the concept. Once the machine develops enough level of cognition it begins making guess about the objects which weren’t even in the data initially fed to the system.

 

Robotic process Automation uses a set of instructions to automate a rule-based task. It uses a set of statements (steps in some cases) to define a repetitive activity and uses an algorithm to automate it. It demands systematic and organized inputs else it fails to bring the desired results. Due to no inbuilt intelligence, it has a very limited scope and is prone to making errors if inconsistent inputs are presented to the system.

Edited November 13, 20187 yr by Puneet Vohra
THought deeply on the same topic

Hi, 

as per my understanding both serving different purposes.  RPA is ‘doing - works basis on set of instructions’ and AI is  ‘thinking / self learning’.

 

Examples:

Chat BoT  &  e-mail BoT are part of  AI - works automation of rule based tasks - replies basis the scripts developed

 

where as RPA is minimize the human repetitive activities like task demanding extract, compute and compare etc.

 

Thanks,

Srini

 

Following is the answer to the RPA question which I have posted in the link,

 

Robotic Process Automation (RPA) is a preconfigured & predefined software tool used for autonomous execution of repeated & manual processes, transactions, activities and tasks with human exception management. Manual processes are inefficient, prone to errors and lead to client & employee dissatisfaction. RPA allows organizations to automate their workflows and helps in overcoming these issues. Organizations are opting for RPA because,

1.       Most non tech organizations still use legacy systems

2.       Large number of work force still complete work manually using multiple systems

3.       Advantage over outsourcing

Benefits:

1.       Reduction in manual effort

2.       Improved delivery, quality & compliance

3.       Improved client & employee satisfaction

RPA is apt for processes which are well defined, standardized, centralized and repeatable. RPA has gained popularity with organizations across different industries for automating wide variety of tasks.

Examples:

1.       Onboarding & profile creation of client/ customer

2.       Profile edit/ updation

3.       Claim processing

4.       Reconciliation & report generation

5.       Compliance reporting

6.       Data cleansing

7.       Compliant processing

8.       Order update & processing

9.       Sending notifications

10.   System access & setup

 

Artificial intelligence (AI) emphasizes the creation of intelligent machines that work and react like humans. It has become an essential part of the technology industry. Research associated with artificial intelligence is highly technical and specialized. Some of the activities machines with artificial intelligence are designed for include,

1.       Speech recognition

2.       Knowledge

3.       Reasoning

4.       Learning

5.       Planning

6.       Problem solving

AI has two core parts Knowledge Engineering and Machine Learning. Sufficient information is a must for machines to act and react like humans. There should be full access to objects, categories, properties and relations between all of them to implement knowledge engineering. Developing common sense, reasoning and problem-solving power in machines is a herculean task. Machine learning enables an algorithm to evolve. The algorithm is feed huge amount of data so that it can adjust itself and continually improve accordingly. AI also has wide variety of application across different industries.

Examples:

1.       Auto pilot in flights

2.       Self-driving cars

3.       Voice recognition in devices

4.       Navigation in maps

5.       Smart homes

6.       Social networking

 

RPA and AI are the different ends of a continuum known as IA. RPA is a software tool that mimics human actions, whereas AI deals with the simulation of human intelligence by machines.

 

Robotic Process Automation	Artificial intelligence
Doing	Thinking
Process centric	Data centric
Rules based	Rules & logic based

 

Both RPA and AI as a valuable tools can be used by the organizations for technology transformation. The choice of implementing either RPA or AI (or both) depends on the specific use in organization, and ensuring “fit for purpose” is the key.

RPA is rule based, follows a coded logic and can do repetitive tasks. Whereas AI is a step beyond RPA; and can mimic human behaviors such as cognition and reasoning and take complex decisions. 

 

The chosen best answer is for Ehtesham because of differences shown in lucid tabular format alongwith examples for both. The answer also has just the right detail that was asked for without any extra dimensions.

 

R Rajesh's answer cites additional examples.