Executive Briefing: AI changes everything. . .except in investment management

Description

The consensus view is that AI is disrupting asset management. Angelo Calvello takes the contrarian view: it will be years before such disruption occurs.

There’s no doubt that traditional investment methods and tools have failed us (and this failure has adversely affected the lives of millions of pensioners, students, and other beneficiaries). We are in dire need of new, innovative approaches to solving investment problems; AI (including both nontraditional data and the new, powerful computational methods used to extract insights from this data) holds the promise of providing such innovation.

At most, a handful of asset managers—early adopters—genuinely use AI in a consequential manner to create better investment outcomes. Yet almost daily another asset manager proclaims its use of AI in its investment process, while another foretells of its intentions to adopt AI. The former’s claim generally reduces to a repurposing the use of big data, a misappropriation of the term “machine learning,” or simply corporate window dressing. The latter’s plans will go largely unfulfilled because the actionable framework required to adopt AI in a way that generates better investment outcomes (i.e., more consistent returns per unit of risk in various market environments) disrupts the manager’s entire business.

For example, incorporating AI into the investment process requires the firm (especially its leadership) to radically alter their worldview and accept that conventional investment approaches and the supporting academic dogma on which they have built their current business and careers do not fully circumscribe the boundaries of investment possibilities and that AI could produce superhuman results that might not be explained by traditional measures and tropes. It also forces management to transform its existing business model (culture, human resources, compensation, etc.) and management style (successfully integrating AI into the business is a nontrivial task), drastically expand its existing budget knowing the ROI is uncertain, and rebrand itself in a way that demonstrates that this transformation isn’t an act of capitulation.

True adoption requires true disruption, and given that asset managers have a hegemonic market position, earn a 30% profit margin on mediocre to disappointing investment products, and aren’t seeing a wholesale dislocation of their client base, they have little incentive to self-initiate such disruption.

Angelo explains why asset managers will inevitably (but slowly and haltingly) incorporate AI into their investment processes in a meaningful manner. (But expect the hand waving and misdirection to continue unabated.) Angelo argues that this incorporation could be accelerated by the entrance of an external AI-based actor (see his article on Kai Fu Lee) or the success of AI-based investment startups.

Please note that the presentation’s scope is using AI directly to make investment decisions; it does not discuss how AI could improve investment operations, compliance, or other critical but noninvestment processes. Moreover, the presentation’s focus is institutional investment management—the large pools of assets like pension funds, endowments, and sovereign wealth funds managed for the benefit of others. The presentation will not discuss retail applications of AI like roboadvisors, as these are trivial from a technological perspective.