By Simon Shah, Chief Marketing Officer, Redwood Software
I saw a fascinating chart the other day, produced by Michael Donough, Bloomberg Intelligence's Global Director of Economic Research & Chief Economist. It shows a line that meanders along, then upticks sharply in 2014. What does this line represent? An analysis of the number of times that companies have mentioned artificial intelligence in their quarterly earnings calls.
This puts a revealing figure on a trend that we know to be true – businesses across a range of sectors are promoting their expertise and innovation in integrating this technology into their offering. As ever with technological advancements, software providers are quick to jump on the buzzword. IBM Watson is the classic illustration, rebranding all software properties over the last 20 years as ‘cognitive’. The concept of AI is now very much on trend in enterprise tech – but how much of this is truly new technology, and how helpful is this term in the wider conversation around robotics in business?
AI has been ‘the next big thing’ for several years, though the reality is that there has not been a quantum leap in Natural Language Processing, Machine Learning or AI. These tools and algorithms have been around forever - the majority of what can be enabled and achieved through AI comes from the availability of abundant data, the power to process it, and the ability to build rules based on processes that are completely unstructured.
The use of AI will most likely evolve as a set of intelligent data processing services, used to provide insights. Let’s take a look at IoT – whether in the supply chain, for remote diagnostics on your car, or control of your heating – the ability to attach a phone line to a piece of equipment has been around for a long time. What’s new now, is the ability to distinguish between noise and signal to deliver useful information. For example, the ability to predict with a high degree of certainty that the boiler will fail in the next three weeks. Another good example is the app that tells you about possible traffic jams as you are heading out from the office. The phone is inferring a process (that you’re going to leave). The real utility is in the ability to infer and build rules based on processes that are unstructured.
In all the excitement, AI and Robotic Process Automation (RPA) are often bundled into the same sentence, without enough practical proof of their connection or even applicability. The truth is that most of the automation we see today is not yet about intelligence, but ‘automated stupidity’. This doesn’t undermine the potential of RPA – quite the opposite. It offers the opportunity to automate the dull, repetitive, rule-based work so that humans are freed up to leverage their own intelligence, eliminating the costs, risks and wasted time associated with an over-reliance on manual labour.
What is starting to be very interesting on the other hand, is Intelligent Automation (IA). Developers are wrapping their minds around this, with a view to harnessing intelligent insights into process execution that can drive intelligent improvements. With this layer of intelligence, we move away from ‘assisted robotics’ that require constant monitoring. The true value of robots lies in their ability to identify and fix anomalies as well as start to predict issues or areas for improvement, based on previous patterns. It is these increasingly efficient processes that will lead to a continuous cycle of process improvement, rather than just short-term cost savings.
We need to re-wire the conversation around robotics in a business context. It’s not about doing intelligent things – it’s about masterminding a series of ‘simple things’ in an intelligent way.Categories: Automation Robotics