AI won't replace investment managers, but it could improve returns

It shouldn’t be news to anyone that artificial intelligence (AI) is developing rapidly and drastically reshaping our future. Yet what the average pension-contributing individual might not have thought about is how this development will affect the management of his or her pensions and savings in the years ahead.

Here are some questions we should be asking ourselves now:

How might AI affect pensions and savings?

Asset and wealth management firms are working on ways to use AI to improve their investment decisions and to extract insights from historical data, which should increase efficiency, accuracy and compliance. Other services, like robo-advisory, are also taking off. With the global AI asset management market size expected to be worth almost $13.5 billion within the next six years, it’s clearly a booming part of the business.

The development of technology in finance and computer-aided algorithms has also contributed to rising interest in low-cost, rules-based, quantitatively-oriented passive investment strategies. Over the last five years, such strategies have grown by about $2 trillion while traditional active management strategies declined by roughly the same amount.

Traditional market capitalization-weighted indices – like mutual funds or exchange-traded funds (ETF) – have also grown over the last five years. This has been aided by minimal costs and strong performance due to the extended bull market. However, these indices are now being rivalled by other weighting systems that deploy technology to deliver better outcomes and mitigate risks. For example, as stakeholders increasingly focus on the harmful role of carbon in global warming and the importance of other ESG-related issues in mitigating portfolio and social risks, asset owners are using technology to adjust market capitalization-weighted benchmarks.

In this way, we expect the role of AI in index investing will have a knock-on effect of improving asset owner and asset manager decisions, focusing them on working with portfolio companies over longer time horizons and supporting companies that, for instance, need to transition away from poor carbon outcomes. We further expect that this will lead asset owners to engage more transparently with their constituents (for example, pensioners or retail clients) to explain why such decisions are optimal.

How does AI compare with human judgement?

AI will gradually play a greater role in critically assessing human judgements in investment decision-making frameworks, to moderate or counter short-term behavioural biases or even to make judgements without human input. This will mean a streamlined ability to predict what will happen and the best course of action for financial forecasting. This is an improvement on a currently very manual process that suffers from inherent human biases.

Could AI displace investment managers? We don’t think so. For AI to be successful in any of this, AI still needs humans. Firms will need to invest in experts to monitor the AI and ensure it functions and make adjustments when needed. This means that investment decisions will still be largely reliant on individual judgement and common patterns of biases, much like they have for the past 20 years.

What are some of the ethical considerations?

Using AI in investment management also creates new risks and challenges. A model is only as good as the data you put into it – so model opacity and data integrity matter. And questions about the ethics in justifying or bounding AI investment decisions will also need to be answered, as well as issues surrounding the regulation of AI systems, the risks of heightened pro-cyclical investment activity and the current lack of a monitoring framework, as AI plays an ever-increasing role in the investment decision-making process.

These questions matter because of the need to maintain and improve the level of trust and confidence in investment management and the broader financial system in general. This is partially achieved by explaining transparently how ethical and social responsibility factors can be included in the investment decision-making process at the portfolio, strategic and regulatory levels, so that AI can be developed with proper balance.

At the same time, trust and confidence can only be gained if frameworks are developed to control pro-cyclical investment activity, monitor AI decision-making impacts in financial markets, and enforce the role of human interaction in the investment-management process to oversee and communicate decisions (whether by human or AI).

What should we be doing about these questions?

In order to allow for these positive developments, we need to identify the key stakeholders to manage and encourage positive AI development in the investment management process. These include regulators, asset owners, asset managers, custodians, investment consultants, legal experts and representatives of retail investors. We also need to identify methods to increase awareness of the AI challenge in the investment management industry.

We need to bring stakeholders together to agree upon an appropriate framework of ethical inputs, regulation and monitoring systems for AI.

Regardless of regulation and monitoring systems, asset managers also need to take it on themselves to address some of these issues in their existing investment process.

We are ultimately optimistic that properly managed AI will support human judgement in managing pensions and savings – and maybe even improve your returns! – by reinforcing best practices and mitigating known biases.

This is part of a series authored by current and former members of the Global Future Council on Investing. Views expressed are those of the author alone.