The kind of AI used in process automation may not be smart enough to pass the Turing test, but it does involve increasingly more complex processing algorithms, iterative learning and sophisticated data abstraction, replacing decision-making and administrative skills that have, until now, necessarily been performed by people.
The boundaries of the technologies that are collectively referred to as AI are being challenged every day in labs and startups around the world, covering everything from machine learning, natural language processing, and image recognition, to any manner of applications that were once the exclusive remit of human intelligence.
While many have heralded the advent of a new machine age, this is, at the moment, mostly still hyperbole. It is true that machine-to-human interactions have increased in our daily lives in situations as varied as ID verification, responding to customer queries and automated investment advice (robo-advice), but this technology isn't yet at a level where it is going to replace the more detailed and analytical work that, say, a skilled financial adviser may engage in when it comes to complex portfolios.
There is, however, a quiet revolution happening with this nascent technology that many will be unaware of, and it is having a very human impact. It is taking place in the back offices of large corporations, where sophisticated software using various strands of the AI technologies or 'robotics' (as the industry has labelled it), is being used to automate a lot of the simpler and more straightforward processes that have historically been provided by humans.
At the moment, it is industries such as financial services that are embracing the technology with open arms. Certainly most of the business process outsourcing deals we have advised on in the last 18 months in this industry have included a significant element of robotics, resulting in a large component of re-engineering of the business processes involved to include for them and to streamline what remains.
For once, it seems to be the insurance industry rather than the banks leading the charge and, if you consider the sorts of business processes involved in that industry, you can see why. Robotics are perfectly suited to a situation where the back offices have to administer millions of policies, contracts and claims, and make decisions using processes with defined parameters; basically the AI 'sweet spot'. Perhaps one of the most visible of examples is Lemonade, an American insurance company whose USP is its focus on AI. Lemonade reported back in January, that it had determined and settled a claim submitted by mobile phone within 3 seconds using AI tools - considerably faster than the usual 30-45 day process that many insurance companies would normally have to go through.
Over the past decade, corporations have tended to cut the cost of repetitive business processes by offshoring them to a country with lower labour costs or by outsourcing to a specialist company which has economies of scale. They will now be considering increasing their investment in technology to cut out the labour cost altogether for certain functions. Replacing basic human tasks with technology is a centuries-old concept, but businesses are now seeing the value of machines that can be trained to understand how to make decisions and automate processes to a previously unimagined extent.
Process automation is increasingly being used to analyse policies and claims, detect fraud, and for marketing. Even in its current early guise, it has demonstrably improved efficiency and minimised the risk of human error, all the while providing a compelling business case with its cost advantages.
A lot of this technology, especially tools like chatbots which can only operate based on what they have been trained to do, is heavily reliant on the data that is given to it. We saw evidence of this last year when Microsoft's chatbot Tay was trained by users to spout offensive content almost as soon as it was launched!
The trick for customer-facing tools appears to be to figure out what specific need you are trying to fulfil, and then to make it clear to customers what the tool can and cannot do. It is vital to ensure that you know why you are using AI for a particular process, and to understand how 'intelligent' your AI tool actually is. The tool needs to be able to identify the limits of its own capability and call on human intervention when needed and there must be an absolute assurance of the privacy, accuracy and completeness of the data being processed by the tools and of the output that the tool delivers.
The continuing need for process assurance and the importance of contextualisation help explain why most current deployments of AI technology in business processes have been part of a wider structured transformation. The smart algorithms behind AI products must be integrated with legacy systems, rest on reliable datasets, and support the end-to-end processes that humans must still do in a way that is efficient and risk free. In short, you cannot just upgrade a particular function in isolation – you need to ensure that all other systems and people who interact with that function are going to work seamlessly with the upgraded AI technology, so that both the AI function and your legacy systems are running on the correct data and are communicating effectively with each other.
So a few final thoughts on AI trends in 2018:
This quiet revolution has already started and will only grow during 2018. When you strip away the industry speak of “leaning processes” and “optimisation” what is essentially happening is that AI technology is indeed replacing humans in certain areas – perhaps the hype is not so overstated after all.
If you have any questions on this article please contact us.
Artificial intelligence algorithms and big data are already being used in hospitals and other areas of healthcare and diagnostics. What exactly does this mean for patients and for those seeking to overcome the legal challenges?
1 of 6 Insights
3 of 6 Insights
Facial recognition technology (FRT) tends to dominate the headlines when it is used for law enforcement purposes, but its commercial use is one of the more widely visible forms of AI.
4 of 6 Insights
5 of 6 Insights
Countries including the US and China have traditionally led the way in developing strategies to foster and develop AI technologies, but 2018 has seen the UK and EU taking meaningful steps to position themselves at the forefront of the AI market.
6 of 6 Insights