Digital TransformationFuture of AI

Progress in AI: too slow or not yet on show?

By Ronan Donohue, Q4 Capital Advisors

Whether from grainy news images of ‘Black Monday’ in 1997, or more memorable representations depicted in Hollywood blockbusters, most people can easily visualise the frenetic warehouse-sized enclosures known as trading rooms. Though arrested by the sheer size, the noise and the peculiarity of employees holding a phone to each ear (giving way to headsets), for many the most striking aspect is the sheer numbers of people present. This may be set to change.

Since the launch of ChatGPT in November 2022, banking and finance has ranked high among industries identified as vulnerable to disruption. Yet, to the casual observer it appears not much has changed. At the same time senior management, while paying lip service, seem reluctant to move away from the watchwords of stability and continuity in their public statements.

The infrequent use of the loud hailer to champion AI should not be misread as evidence of inactivity. Banks, and for that matter other financial institutions such as insurance, are in the main not sitting on the sidelines. Look below the surface and it’s clear the shadow of AI is lurking in almost every business unit.

It is also clear that AI, machine learning and related areas (henceforth AI) has huge potential to enhance productivity and the customer experience, but with a corresponding knock-on impact on employment. This perhaps explains the less-than-enthusiastic public embrace of the technology. Automation in the sector has a direct line to employee numbers.

Wherever the searchlight drops, it is clear some business functions stand out more than others as being ripe for AI-led enhancement.

Begin with an early win

Wealth Management is one such area marching steadily and comfortably ahead alongside what AI has to offer, rather than trying to shield from it with an emphasis on the old ways. Smart managers realise that key aspects of AI, robo-investing chief among them, are already sufficiently advanced as to make resistance futile.

Arising in the vast retail investment platforms in the US (as well as newer markets such as China), cheap access to securities investment portfolios, selected and managed according to pre-determined risk criteria and objectives, has proved immensely popular. Add the steady adoption of user-friendly investment options such as Exchange Traded Funds (ETFs), which seem designed to play into the automated investment space, and the unambiguous direction of travel is as clear as a well-polished gem stone.

To ignore these developments in the somewhat similar, albeit higher-value, world of wealth management would appear, not just commercially unwise but negligent. Banks offering the wealth management service (with its attractive fees and lower capital requirements) in the main are open to embrace what technology has to offer.

Alongside client portfolio modelling, security selection and risk mitigation, the involvement of AI in client services points to a more tailored approach in general, with clients’ particular preferences and outlook catered for in real time.

Furthermore, AI algorithms can pick up risks not immediately apparent to a wealth manager, including concentration risk, credit downgrades and critical news items or company announcements. Portfolio rebalancing on certain triggers can be pre-programmed to suit the risk and preference profile of individual clients, thus ensuring timely reactions to events.

Many of these technologies can in time be repurposed to other areas of banking such as debt and equity capital markets where enormous (underutilised) data sets also exist. Here too, a more accurate and tailored client interface is likely to emerge along with real-time updates to pricing and balance sheet modelling. Data engines can also be trained to identify optimal windows of market access, incorporating movements in interest rates, investor appetite as well as geopolitical events in general.

From detect and repair to predict and prevent

Though perceived as being somewhat late to the AI table, the insurance industry is well positioned for value enhancement from AI technologies, which according to German insurer Allianz could add a potential $1.1trn annually.

Central to this is the exponential growth of connected devices creating an avalanche of data with which to more accurately and easily price risk. Add to that the ability to streamline verification processes and the cost of onboarding new clients is set to fall, as is the associated time commitment. As a result, the old model of Detect & Repair is moving towards Predict & Prevent.

Adding blockchain-based smart contracts to the mix and the claim evaluation/settlement side of the business should benefit enormously. This is turn will have a knock-impact on customer perceptions of the value proposition from non-essential insurance offerings (for example, immediate pay-out direct to a customer account on a flight delay), opening the door to significant business expansion.

Motor insurance too is ripe for a makeover, including customer options to go full cover on a trip-by-trip basis, maintaining only third-party while the vehicle is sitting idle. Other examples could include phone battery insurance, or washer versus dryer insurance.

For insurance providers, earlier notification of weather events and other impending loss allows for more timely settlement, lower overall operational costs and better capital management.

On the investment side of insurance business, we can take a lead from the developments in WealthTech above. More tailored matching of assets and liabilities could also be an advantage, especially in life assurance where longevity risk poses a problem in this regard.

Hub or spoke

Advice to organisations on the extent to which they should embrace their changing operating environments varies significantly from threading cautiously to wholescale transformative change.

Proponents of the ‘go big’ approach point to the advantage of laying foundational tracks capable of accommodating additional business units, and in the long run at far lower cost. They also stress the need to decide up-front as it will determine the quantum and quality of data captured during the process and the configuration of associated models.

Even among those favouring a graduated approach, a further debate concerns oversight. Should the business units themselves drive change, or should it come from the top? The worry is that the development of individual silos may struggle to come together in the event a group-wide vision eventually takes hold.

Though an industry standard approach has not yet emerged, there is growing evidence to suggest management favours a more centralised grip on risk and cyber security functions while ceding service enhancements to the business heads, for now at least.

Behind the safety car but still on track

AI’s vast potential in risk mitigation has no doubt turned heads at board level. Reducing risk has a direct impact on available capital, which if properly deployed shows up on the bottom line. If AI can help with this ambition, expect its adoption to become mainstream.

According to a recent study by the Bank of England and Financial Conduct Authority, most perceived benefits of AI in the financial services sector are to do with data and analytics insights, anti-money laundering and in fraud prevention. Cybersecurity is perceived as the highest risk over the next three years, though it is believed the benefits outweigh the risks.

Looking across a broader set of sectors, a recent McKinsey survey identified risk and compliance along with data governance as the two areas most likely to be managed centrally. Other functions (such as the adoption of certain AI solutions) are frequently handled within the business units or a hybrid combination of both.

The same study identified larger companies (those with annual revenues >$500 million) leaning into AI adoption and investing also in internal communications to demonstrate value creation. It makes sense to ensure employees are on board. Not only are they likely to provide extremely insightful feedback on use cases, but the sense of inclusion in the face of impending change should reduce job uncertainty and stem a fall-off in productivity.

The never-ending question of when

Actual and potential use cases for AI in finance proliferate across the sector, but a question commonly touted with increasing frequency is when we will see the sector effectively upended? More common again is a misconception that as time moves forward and seismic change is not yet in evidence, analysts and commentators have somehow gotten it wrong and what we are witnessing is merely overdone hype.

International wholesale finance is a frenzied business attended by short attention spans. This mindset can be an advantage, some would say a necessity, to effectively navigate an economic backdrop changing by the minute. This might explain some of the apparent surface-level inertia and reluctance to look too far ahead.

Tech observers would do well not to misread these signals and not to be overly influenced by the state of play on the customer-facing side when so much of the progress to date at entity level is taking place in the engine room, especially in less obvious areas such as risk management and other developed AI technologies already clipped into workstreams.

In addition, such perspectives rarely incorporate progress already made to industry parameters. The EU AI Act passed by parliament last year (approved by Council on 21 May 2024) constituted the first attempt to define AI. Others such as the UK’s AI Bill are progressing with its own definition along with other provisions.

Alongside statutes, accounting and tax issues must be factored in, as must company-specific aspects including corporate culture, as well as limiting factors to technology itself such as energy and water.

Progress is well underway across banking, finance and related ecosystems. But, as anyone who has ever waited for a bus knows, it feels longer if counting the seconds.

Author

Related Articles

Back to top button