The pensions industry is perhaps not the first port of call for implementing the new technology, especially with multiple other regulatory and innovation projects going on, but some from the industry are looking at how it could be utilised for certain purposes to improve outcomes for members and increase innovation.

Professional Pensions spoke to some representatives from the industry to find out if the technology could bring enough benefits to pensions to justify its use.

He said in general, the area of pensions can sometimes be seen as a “slow adopter” to some of the brand-new technologies, especially compared to banking and general insurance, although noted this could be an opportunity for the industry to follow through on a new way of working.

Elliott said Gallagher already uses AI “on particular client needs” today, but it is “on everyone’s agenda” and is being utilised in areas like business transformation and the wider strategy.

Having the right controls and checks in place is essential, Elliott says, pointing out AI can change on a “weekly, if not daily” basis, with businesses needing to work closely with partners and clients on any wider rollouts, and to perhaps collaborate with other providers, who can be taken on the “journey” towards implementation.

“I think people are inherently cautious. You don’t want to be going out and just saying that we’ve got this in place and then find out that there’s risks that you need to navigate… whether that’s data privacy, GDPR, or regulatory, so we need to make sure that we’re tackling that in the same way.”

However, it is important for businesses to think about how they can drive things forward, with utilisations including helping with routine tasks and freeing up capacity being crucial benefits businesses would not want to miss out on.

Elliott explains: “We know there’s a capacity crunch in the industry – there’s business as usual keeping teams busy, strategic projects, GMP [equalisation], buy-ins and buyouts, dashboards… there’s so much data work that is being asked of the industry at the moment… it’s a perfect storm in terms of capacity”.

In addition, having improved data insights can help implement more “targeted” communications and personalisation for members, especially specific cohorts who don’t already engage with online portals. AI can even help with signposting and with reformatting the portals themselves, whereas previous yearly statements issued as a blanket communication are less likely to have the same impact.

Elliott also notes the changing demographics of those becoming interested in pensions, meaning providers should think about how they will approach the opportunity.

“I think some people are seeing a sort of ‘Terminator 2′ post-apocalyptic view of AI, and some are already engaged with it… where the pensions industry is good is because it is quite cautious and reasoned, so how you communicate is going to be key.

Elliott says members should always be well informed, be encouraged to make the right decision, and hopefully see improved outcomes along the line, with providers needing to manage any concerns they might have along the way.

Oversold and under-delivered?

Isio partner and head of administration Girish Menezes said AI can bring many benefits but added that it may be “often oversold and underdelivered”, adding providers need to be cautious about what is achievable and deliverable in a reasonable timescale.

He said: “AI is already helping in its ‘intelligent search and transform’ functionality embedded within our laptops, electronic document repositories and website software. You can search for specific information, translate, transform, summarise and report on it.”

Menezes said chatbots and telephone software have already embedded AI, adding this can be used to support members, reduce the human support requirements, and even identify fraud – however “specialised skills” are needed to implement such solutions, rather than a “straight from the box” approach.

He said: “Data analysis, data cleanse and reporting can be improved exponentially. But remember the joke about being great at math but not knowing which numbers to count. Leveraging AI in this area requires individuals who understand complex UK pensions, AI itself and be able to think strategically, laterally and commercially regarding valuable outcomes.”

Menezes added: “‘Robo-Advice’ is a case study in itself which reminds us how much investment AI requires, the importance of embedding it within a broader ecosphere and that AI augments, rather than replaces humans. An investor may use AI-powered tools to explore and analyse. However, one still needs a fact find and, especially when money is at risk, the reassurance of a trusted human being.”

However, the “threat” of redundancies and large-scale upheaval should not be ignored – yet Menezes pointed out previous transformational changes, such as the introduction of spreadsheets, managed to produce as many jobs in data analysis or accountancy as it made redundant the role of traditional bookkeepers.

AI investments

Also singing the praises of AI is AllianceBernstein head of multi-asset solutions David Hutchins, who said AI is “very much in the nascent stage” in pensions, as with other industries.

“Currently, the biggest link between AI and pensions is the enhanced returns it has provided to investors invested in US large cap tech stocks linked to AI. However, it is not AI that has generated this directly. In fact, looking at the four largest AI powered exchange-traded funds (ETFs) we see they have underperformed their benchmarks by on average 10% per annum! The irony of this should not be lost. Clearly as AI learns and becomes better this should improve, but the case for replacing portfolio managers completely with AI is far from compelling.”

However, AI is already an essential tool to allow data and information to be more rapidly analysed, and for ideas to be generated. Improved member engagement can also help make digital member engagement more effective in its outcomes, he said, with “faster processing, natural language capabilities and more sophisticated advice processes”, however there are still barriers to achieving this.

Member engagement, for one, is still a “perpetual problem” and a “hurdle” that’s impeded progression for a number of years – AI could therefore more rapidly adapt to members’ style and timing. “But let’s not forget humans have always adapted to their environment and it is likely that they will learn to ignore even the cleverest of AI just as much as they learned to ignore human engagement if they so want to”.

Hutchins also noted AI is “neither telepathic nor clairvoyant”, especially in customising investment strategy or retirement options. “Indeed with learning it will realise, in the same way as good advisers, how much is unknown at the personal level about an individual’s uncertain future needs and will seek to limit overly specific personalisation based on what it knows are erroneous assumptions – an issue that has dogged poorly designed and managed defined benefit and DC plans for years using human intelligence.”

Source: PP