Rebooting the global asset management industry

Rebooting the global asset management industry 35 This imperative is further reinforced by organization-related constraints (Figure 3.2, lower panel). The key one here is the need to protect clients by only using technologies already tried and tested by time and events (54%). AI and GenAI have not yet reached that degree of maturity. Yet another factor in this cluster is legacy thinking on the part of senior management. It is based on the familiar innovator’s dilemma: while profit margins are high, pressure to change the operating model is limited (36%), thus reducing the urgency to create a governance process to identify and prioritize the use cases, ranging from quick wins to moonshots. Allied to this fact is the overemphasis on short-term profits to the detriment of long-term planning (33%). This is because the incentives for top executives are tied to short-term results (22%). So much for the moderators. Historically, they typically dictated the pace in the early phase of implementation. Anything with no track record of success has invoked fear of the unknown. However, over time, accelerators gained ground as early adopters had an edge and reshaped competitive industry dynamics. This, in turn, created the fear of missing out, as price makers turned into price takers. For now, the focus of our survey respondents is on developing AI literacy, governance and security, while duly recognizing that AI and GenAI are about strategic differentiation in value creation, not just efficiency gains. Anything with no track record of success has invoked the fear of the unknown. Insights Data quality must improve before the adoption of GenAI takes off The worldwide diffusion of GenAI has been remarkable. Two previous transformative innovations – the telephone and the mobile phone – respectively took 75 years and 16 years to attract 100 million users globally. TikTok achieved that feat in nine months. ChatGPT did it in two months, so revolutionary is its potential. Yet, at corporate level, its adoption remains evolutionary – at best . A key obstacle has been slower market acceptance. GenAI requires a major effort in updating existing data standards and data assets to provide viable input in training its complex models. The growing volume, velocity and variety of data from external sources is impressive. But assessing its veracity has proved an uphill task to the point where the final model outcomes often raise more questions than answers at this stage. Indeed, online misinformation has exploded on Web 2.0. Our biggest challenge is ‘hallucinations’ – inaccurate outputs that seem indistinguishable from true ones. Data hygiene is vital in harnessing the vast embedded value in GenAI. Without that , outputs from our algorithms driving machine learning and large language models will be only as good as the data they are trained on. Smart AI cannot work effectively with dumb inputs. The problem is compounded by the historical data demarcations within our own business between investment, sales and finance functions – to the detriment of developing an agile integrated governance process for the effective identification, adoption, and scaling of use cases, given the pervasive nature of GenAI across the value chain. The necessary expertise is evolving gradually through learning by doing and trial by experimentation. GenAI is not a turnkey plug-and-play technology, much as we would like it to be. A Swedish asset manager “Not sorting out the veracity of data before adopting GenAI is like putting the cart before the horse.” “The extent to which regulators will actually permit technologies to operate in highly autonomous ways is an open question.” Interview Quotes