• Predictive Analytics
  • May 16, 2017

The Tools of the Future Today – What is Robotic Process Automation, Artificial Intelligence and Machine Learning?

Interest in taking advantage of robotic process automation as well as a variety of intelligent systems, such as artificial intelligence or machine learning, has exploded over the past few years. The technologies in question will have a dramatic impact on the future of knowledge work. However, how many of us know what the terms actually mean?

Robotic Process Automation

Robotic process automation – or RPA in short – involves the production of automation with the help of software. Robotic process automation differs from artificial intelligence in the sense that software robots must always be provided with instructions because they themselves are not intelligent — at least, not yet.

Robotic process automation is ideal for, for example, entering purchase invoices in an ERP system or the establishment of a new customer account in a number of systems simultaneously, and the routines related to the hiring of a new employee and the start of employment relationships. The common denominator in all of these cases is the creation of various user IDs and the importation of the same data into more than one system, meaning that automation reduces mistakes and may generate a lot of savings in terms of work that has formerly been carried out manually.

The creation of automation requires a human to switch on the robot assistant when first taking it into use and provide it with instructions. The rest is taken care of by the robot, autonomously. This ensures that all of the necessary user IDs and access rights are generated at the same time without errors and that the quality remains good as the number of mistakes declines. 

Artificial Intelligence

Artificial intelligence (AI), on the other hand, is an umbrella term for a machine's ability to imitate a human's way of sensing things, make deductions and communicate. An example of this is machine vision, which allows for real-time traffic counting from the feed of a traffic camera, the anticipation of congestion, and alerts concerning potential emergencies. Machine vision and image recognition enable the automation of control room work and preventive intervention in situations.

IBM's Watson cognitive system is capable of human-like inference and communication, meaning that it has cognitive abilities. Watson has already proved its abilities in a variety of fields, from quizzes to healthcare projects. In Finland, Watson has been piloted in screening prematurely born babies for a sepsis risk. The software uses the data of previous patients who have suffered from sepsis and compares them to the real-time data accumulating in a digital patient data system, looking for early signs of the condition's development.

Another advanced manifestation of artificial intelligence is Google's DeepMind, which learns from its experiences and from feedback. When an artificial intelligence gets to play chess for the first time, you don't need to teach it the rules. Rather, it learns the game by playing it and can even beat a human at it.

An artificial intelligence is nowadays fairly adept at understanding natural language, due to which it can identify certain terms used in a real-time phone conversation, for example, and use them to make inferences as to what the conversation is about. This is not about frequencies alone. Instead, the artificial intelligence can interpret the terms and their context, thereby forming a deep understanding of the situations, things, and attitudes the words relate to. Applications based on textual analysis and the understanding of natural language are already in use in customer service solutions, insurance companies, and public sector organizations which process vast amounts of digital material, such as emails, online service feedback, chats, and documents.

Machine learning

AI solutions often make use of the methods of machine learning. A machine can, for instance, be taught to identify phenomena with the help of mathematical and statistical methods. In this case, "teaching" means loading numerous images, numeric values, or text that represent the phenomenon to be learned into an algorithm. As a result of this teaching, the algorithm is gradually able to become increasingly better at identifying a particular phenomenon.

In the future, robotic process automation and forms of machine intelligence will become increasingly amalgamated. Different technologies can be used side by side to perform the same task because of their ability to complement one another. 

This kind of joint operation is nicely illustrated by a customer service situation in which the customer first communicates with an AI customer service representative through a chat function or by talking. During the process, the customer is given recommendations and suggestions on products with the help of machine learning algorithms. Finally, a software robot handles the customer's order automatically and sends an order confirmation which accounts for the customer's profile and the service situation in a natural way.

Jyri-Pekka Makkonen
Head of Sales