More than 200 astronauts have visited the International Space Station since it was launched three decades ago but the latest guest, who arrived in June, is unlike any other.
(pronounced Simon) has no arms or legs, is shaped like medicine ball, and has a screen displaying a simple animated face. But inside lies sophisticated artificial intelligence (AI) that allows it to help crewmates perform routine work and offer solutions to problems.
CIMON, which stands for Crew Interactive Mobile Companion, is just the latest example of how AI is reshaping the way we interact with the world and each other. While AI has been around in various guises for decades, its ability has skyrocketed in recent years thanks to algorithmic advances, big data and increasingly powerful computation.
CIMON is powered by IBM Watson – one of the first AI systems developed, which in 2011 famously beat two former winners on the popular quizshow Jeopardy! taking first place and the $US1 million prize.
However, its underlying technology is now applied to a wide range of applications and processes, such as making treatment recommendations for lung cancer or providing recommendations to financial planners about suitable products for clients.
Financial services is one of several sectors where AI can underpin innovation and productivity growth.
Professional services firm Accenture
recently predicted that AI technology can deliver $US1.2 trillion in additional gross value to the financial services sector by 2035 by relieving knowledge workers from repetitive tasks such as generic customer queries, mortgage reviews and market research.
BPAY Group is also making moves in this area with the launch of new company Sypht, which uses machine learning and natural language processing technology to extract and interpret document data with an impressive level of precision.
AI to bolster integrated customer service
Fergal Murphy – who previously led the IBM Watson Industry Platforms unit and now heads the AI team within Accenture’s digital business in Australia and New Zealand – calls AI the current “alpha trend”.
However, the conversation has recently shifted away from standalone AI, such as chatbots, which were largely propelled by marketing hype rather than their natural language processing and learning capabilities.
The AI Index 2017 Report
, an initiative to track the industry’s progress hosted at the Stanford AI Lab, highlighted many recent AI advances but noted a lack of widely accepted measures of success may create an optimistic bias.
“Chatbot dialog falls far short of human dialog and we lack widely accepted benchmarks for progress in this area,” the authors wrote. “Similarly, while today’s AI systems have far less common-sense reasoning than that of a five-year-old child, it is unclear how to quantify this as a technical metric.”
Murphy says firms are now taking a more holistic approach to AI.
“Rather than those standalone chatbots, which is where things were going, we’re seeing a much more integrated experience where AI helps integrate all the digital aspects into that customer experience."
For example, a customer could call their bank on their smartphone and ask to open a new bank account, which the AI-backed system would understand, and then continue the process by delivering content through its authenticated banking app.
“We can create a consistent experience and maintain context right through that multichannel experience. It does require some investment, it does require some faith, but that's where we're going to see differentiation.”
Compliance in AI sights but quality data remains key
The Royal Commission into Financial Services has also changed the nature of the AI conversation and re-focused it towards compliance, conduct, trust, and transparency. Technology offers speed and scale to analyse areas that have been in the spotlight such as insurance documentation or financial advice.
“When you look at the financial planning industry, historically maybe 3 to 5 per cent of the advice would have been analysed and checked for compliance. We need to change that to 100 per cent and the only way to do that is with AI. Natural language processing cannot just read that at scale, but understand it, and I think it’s the way forward.”
AI by its nature should improve over time – but is its starting point good enough to undertake such crucial activities as compliance? Murphy says AI’s accuracy at least matches many human checking procedures, such as fault determination in the insurance sector, where AI has recorded accuracy in the mid-to-high 90-plus per cent range.
AI has already matched or exceeded human performance in other areas.
AI-powered automatic labelling of images published on the ImageNet database produce errors of less than 3 per cent (down from 28 per cent in 2010), compared to human performance of about 5 per cent.
Meanwhile, researchers in a 2017 Nature article
described how an AI system trained on a data set of 129,450 clinical images of different diseases was able to classify skin cancers at similar levels of accuracy as 21 board-certified dermatologists.
Despite those impressive results, the quality of data remains a key factor. If AI is being fed say, poorly scanned documents, the ability of algorithms to analyse that information is immediately curtailed.
"If I've got one piece of advice for the market about what they need to make AI work well, it’s to get on top of your data governance now. That's a challenge because as we move into a big data world the volume of data is increasing and the speed of data is increasing with real-time transactions. But a few dollars well spent on governance today will be paid back in terms of what you can do with AI down the track."
This article represents the views and opinions of the author and do not necessarily reflect the opinions of BPAY.
Published by BPAY Pty Ltd. BPAY payment products are offered by over 150 Financial Institutions. Contact your Financial Institution to see if it offers BPAY payment products and to get the terms and conditions. This is general advice – before using BPAY payment products please review the terms and conditions and consider whether BPAY payment products are appropriate for your personal circumstances.