Artificial intelligence: Innovation or threat for portfolio management?

While technology experts grapple with the increasingly real existential threats posed by artificial intelligence (AI), the portfolio management business is quietly considering its own survival plan.

OpenAI’s launch of the conversational bot ChatGPT in November last year has the investment community questioning its own future.

Goldman Sachs estimates that 300 million jobs could be lost over the next 10 years, but the bank also predicts that the combination of significant labour cost savings and new job creation raises the possibility of a productivity boom.

Cleantech investor Paul Sandhu, founder at Prometheus Corporation and formerly head of multi-asset quants at BNP Paribas in Hong Kong, told AsianInvestor: “What’s going on in AI right now is going to fundamentally change how we do most things. Our infrastructure, our systems of governance, are not set up for a mechanism with the potential of ChatGPT."

Up to now, the use of AI in fund management has mostly involved machine learning (ML) applications to improve the effectiveness of algorithms in investment processes, or using AI techniques to process big data for investment insights.

Early adopters

Dutch pension fund manager APG and Chinese insurance group Ping An, for instance, were early adopters, employing AI to help them gather and analyse data to help achieve their environmental, social and governance (ESG) investment goals.

“The industry’s needs have grown from asking for proof that AI and big data work, to asking for an action plan that can provide support to firms’ strategy, as ML methodologies rapidly become mainstream,” said Larry Cao, senior director of research at the CFA Institute.

CFA has produced a handbook aimed at equipping investors and policymakers with the tools to evaluate and incorporate AI and big data techniques into best practices.

Asset managers have had success in collecting and incorporating big data into their investment processes. In particular, since ESG-oriented strategies have become more mainstream, asset managers are looking for ways to assess ESG-related activities in their investment companies and to monitor their progress toward their ESG goals.

Global asset manager Robeco's concern is the threat posed to civil rights and liberties, for example regarding privacy and surveillance in socially sensitive situations such as the search for jobs and housing.

“Safety and accountability lag behind actual technological innovation in many cases, and our engagement with companies seeks to encourage them to catch up,” said the firm in an online posting.

The alignment problem

The downside of AI is the potential for it to manipulate markets, to falsify data and to code programs with malevolent intent in such a way that humans won’t be able to detect. The clearest indication that AI represents a threat comes from the very people at the forefront of bringing it into the world.

OpenAI’s own safety and governance people are quite candid about the risks.

In an academic paper on the alignment problem, Richard Ngo, an AI governance researcher at OpenAI and fellow computer scientists Lawrence Chan and Sören Mindermann said that if AIs are trained in ways similar to today's most capable models, “they could learn to act deceptively to receive higher reward, learn internally-represented goals which generalise beyond their training distributions, and pursue those goals using power-seeking strategies.

"The deployment of misaligned AIs might irreversibly undermine human control over the world.”

Pause for thought

Hence the call for a halt on the development of AI in an open letter signed by Elon Musk and computer scientists at Apple, Amazon, Google and raft of other tech giants.

“Powerful AI systems should be developed only once we are confident that their effects will be positive and their risks will be manageable,” said the letter.

Right now, no such assurances can be given, but not everyone sees the threat outweighing the opportunity.

“I’m actually quite excited; I can understand why people are saying we need to take a pause on it, but fundamentally that’s not possible,” said Sandhu. The biggest risk for investors is that fund managers lose their ability to innovate, he said.

"By regulating and constraining fund management in the way that we have, we have stopped a lot of errors from happening, but it really has restricted innovation immensely.

"This AI knows nothing of constraints, nothing of experience other than the data it has collected. It’s purely empirical and so it can start innovating in really interesting ways, that humans may have been able to but they have (since) lost that ability. That’s where I think things are going to get a little hairy.

"AI can do administrative tasks much better, but it’s on the innovation side where the real risks lie in terms of humans becoming obsolete."

The headline of this story has been changed.