11 June 2018
From the birth of AI in 1952, through the IBM supercomputer ‘Deep Blue’ beating chess grandmaster Garry Kasparov, and to today’s mobile phone personal assistants, AI has remained one of the most exciting and promising areas of science and innovation.
Industries from construction through to retail, logistics and media and entertainment are increasingly seeking to explore the benefits, and it’s high time the pensions industry woke up to the benefits available to pension scheme members and long-term savers.
With personalisation, comes engagement
A personalised approach to pension communications is vital to capture the interest of those people, especially millennials, who are often focused on the here and now, with long-term saving for retirement often at the back of their minds.
AI has shown how personalisation can engage people more effectively in entertainment. Netflix, for example, analyses the data collected from people’s streaming habits to suggest other films they may be interested in watching. The mathematics behind this is part of the actuarial exam syllabus, but is rarely used in practice by pension scheme actuaries. The company is so confident in the interpretations of its data that it is willing to commission multiple seasons of a new show rather than just a pilot episode, based on their findings. If this kind of certainty can be utilised in an industry as volatile as television, why not pensions?
Imagine if we in the pensions industry were able to take the data people input in answer to their current financial situation, future goals and aspirations, and offer them a tailor made savings plan to help them achieve that goal, completely automated and explained in just the right way.
Step up investment in AI to maximise the benefits
What’s clear from the history of AI is that there will need to be significant investment in order to fully realise the possibilities machine learning can offer the pensions and benefits industry. Maximising the benefits will require both patience, and a willingness to experiment with ways that the technology can be harnessed most effectively.
When Google introduced speech recognition in 2008 (technology which essentially went on to become Siri and Alexa-style personal assistants), it pioneered a new approach to a system that up until that point had never been cracked. With thousands of powerful computers running parallel networks and learning to spot patterns in the vast volumes of data streaming from Google’s users, the system went from what the company itself described as “fairly inaccurate” to more than 92% accurate within a decade (the app had an 8% word error rate in 2015, according to company statements).
Any innovations the pensions industry embraces in this area may need a fairly long time horizon to fully realise the benefits, but if “the juice is ultimately worth the squeeze” then time should not be a prohibitive factor in an industry that stretches out for decades. The continuous improvement of the voice recognition tools discussed above shows that the sooner we get started, the sooner we will see the results filter through to better outcomes for pension scheme members and long-term retirement savers.
Looking at occupational pension schemes, there’s no reason why pension scheme trustees can’t use big data techniques to go one step beyond personalisation. For example, they can have two very slightly different versions of their member communications, including their website, and test which one gets the best engagement – known as “A:B Testing”. Then, choose the best one; and experiment with changing something else. As well as the precise wording used, font size and even the palate of colours can make a surprising difference to engagement levels.
There are already instances where AI is being used to address the growing need for savings advice with the number of people auto enrolled now reaching 8 million. Take chatbot, Ameila, for instance. Amelia is used by Swedish Bank SEB to act as a customer service agent, able to answer some of the more simple queries that people have, with the ability to learn from their human compatriots when referring on a question that they cannot answer. More chatbots like Amelia would be a great step forward for the industry and I look forward to seeing how this concept progresses.
So why wait? In my view, it’s time to begin the journey towards fully embracing AI in the UK pensions industry – to achieve better outcomes for members and savers.