Professor Ilan Oshri teaches strategic management and technological innovations in the Graduate School of Management. He has published 18 books on outsourcing and offshoring, and is currently exploring the world of intelligent automation.
Intelligent automation: How far have we come?
“Intelligent automation” (IA) is an umbrella term covering three distinct technologies: robotic process automation (RPA), machine learning and artificial intelligence (AI).
Robotic process automation (RPA)
RPA is a recent technology based on software robots (“bots”) that are configured to carry out business processes previously done by people. The software is programmed to carry out operational procedures and thus acts as a virtual worker.
RPA service providers have been promoting the benefits of this technology since its initial implementation, claiming that RPA reduces costs while increasing service productivity, accuracy and speed. However, in a recent study of RPA implementation, based on media coverage of nearly 160 reports and cases, Ilan and his research partner Georgia Moore found that we lack clarity about the real benefits of RPA and firms fail to report on the return on investment. In another study, Ilan found that around 30% of data in most organisations using RPA is currently not well-processed. Why? Because the incoming data is often not accurate, not relevant or not complete. For example, banking systems acquire data from different sources, which then needs to be collated by humans. The study’s findings suggest that the expected cost savings have not materialised for the RPA cases reported in the media.
In implementing RPA, Ilan says, companies do not significantly reduce headcount as humans are still needed to handle ‘exceptions’ (i.e. those incidents that bots don’t handle well). We can therefore see RPA as a local solution rather than a platform to improve the entire value chain.
In addition to questionable efficiencies, there are psychological challenges for human workers, who often feel threatened by bots – not least because the bots are invisible, leading to a lack of trust in outcomes.
“But ultimately, humans and bots will work side by side”, Ilan says, “and, in many cases bots will be able to analyse data and answer questions, often faster and better than humans. What bots won’t be able to do is define the questions and problems that need to be solved, iterate deeply on the responses, and prioritise solutions.”
Machine learning describes a process whereby algorithms and statistical models are used to perform a specific task without using explicit instructions, relying on patterns and inference instead. Algorithms learn by running through massive amounts of data, in the process becoming smarter in predicting inconsistencies or abnormalities.
Machine learning technology is more advanced than RPA, and is revolutionising decision-making processes in companies around the world. However, a problem arises in organisations that are heavily regulated, such as the banking sector. Although data can be more efficiently processed by algorithms than people, regulators do not as yet accept algorithmic analysis/conclusions. Again, the desirable outcome is for machine learning to support human activity, providing input and validation.
Ilan is currently working with Julia Kotlarsky from the Department of Information Systems and Operations Management to study the impact of machine learning on business outcomes. This endeavour is supported by a grant from the Ca’ Foscari University in Venice, Italy, and the results will be published in 2019 during the Global Sourcing Workshop.
Artificial intelligence (AI)
Most people’s understanding of AI refers to the type of adaptable intellect found in humans, a flexible form of intelligence capable of learning how to carry out vastly different tasks, or to reason about a wide variety of topics based on its accumulated experience. This is the sort of AI we can only see in movies, the likes of HAL in 2001 or Skynet in The Terminator, but which doesn't exist in reality as yet - and AI experts are fiercely divided over how soon it will come.
So where are we at with IA?
Ilan’s recent research belies the hype and fear-mongering prevalent in the media today. His analysis of 100 big organisations round the world, undertaken with the support of KPMG, has found that by and large, these companies are not ready to fully integrate IA into their business processes. The reasons:
- They are not prepared to invest sufficiently in the technologies
- They are unsure of the potential benefits
- Internally, even senior management is not very well-informed, relying on suppliers to fill the knowledge gaps
- Suppliers themselves are learning as they go
In short, progress is slow, investment negligible and expectations unrealistic. IA is not going to take over the world any time soon, and the optimistic outlook is that by the time the technologies are fully implemented and understood, humanity will be well-prepared.