Today’s world of automation technology has been taken over by computers, from IBM’s Watson to self-driving cars, robotics and now artificial intelligence. How do you strike the right balance between just enough automation and too much?
We’re thrilled to share a series of questions and answers from a recent interview with Billtrust’s Research and Development Scientist, Farhad Khalafi. Farhad is responsible for incorporating the state-of-the-art technology in machine learning and robotics into Billtrust’s next generation software in Credit and accounts receivable (AR) automation, and has over 20 years of experience in software development. He also holds a Ph.D. in Quantum Physics from the University of Cambridge, UK.
Farhad sat down to discuss and answer questions about robotics, AI, and machine-learning, and how they can be used in the future of AR automation.
QUESTION: There are so many buzz words in the industry now, such as artificial intelligence (AI), robotics process automation (RPA) and machine-learning. Could you explain how these differ from each other?
FARHAD KHALAFI: I’ll explain each of these terms as I understand them and how they apply to software.
Artificial intelligence (AI) is the process of simulating the human learning process in computer programs. Instead of applying the same routine and producing a predictable result each time, a computer can be programmed to identify nuances and variations, take different actions, and produce more accurate results. It’s similar to how humans make decisions. We handle each situation differently each time depending on our past experiences and information.
A software robot is a program that automates a process by acting on a set of well-defined input and creating a predictable set of output. It is perfect for automating mundane and tedious tasks that could, in principle, be performed by human operators. It doesn’t involve learning, but it saves time by automating manual tasks. When used in accounts receivable automation, software robots automate the AR life cycle by processing invoices, payments and remittances as they move through the process workflow.
Machine-learning (ML) refers to the ability of software programs to adapt and learn from input data, the environment or user feedback. Machine-learning can be considered as a subset of Artificial Intelligence. The learning program may try to mimic the human brain (e.g. neural networks) or use a variety of other statistical learning algorithms (regression, Bayesian networks, Markov chains, Support Vector Models, etc.). A major function of machine learning is to recognize and classify the input that it receives in order to apply the appropriate processing. This function is divided into two categories: supervised and unsupervised.
Check back next week for another interesting explanation from the Billtrust data scientist.
Do you have a question for Farhad? Ask him on Twitter @FarhadK