Artificial Intelligence in Financial Services, 2019: Demand-side Analysis

Analyzing the use of AI in financial services reveals a fascinating combination of old and new. AI is often considered ‘new’, but Chartis’ view is that it is one of several statistical tools that have been with us for some time. Elements of AI and machine learning (ML) have been embedded in financial services for a while, and nowadays AI tools are everywhere in the sector, used mostly as ‘cogs’ in bigger, established processes and systems.

Most of the AI tools that FIs currently use fall into the ML and natural language processing (NLP) categories. ML is essentially an extension of the regression techniques most FIs are already familiar with, while NLP is used extensively in text analysis. Evolutionary programming, the another main category of AI, is considered far more ‘exotic’, and as a result is harder for vendors to market to stakeholders – which is ironic, considering its historical success in several capital markets areas including portfolio management and securitization.

Digging deeper

Like every dedicated process or system, AI must be understood properly to be used effectively.

As part of our ongoing research into AI, based on more than 20 years of observations of AI implementations, we have been considering what AI is, how and where it is used effectively in the financial sector, and the motivations and challenges that FIs and vendors face in supplying and using it. In doing so we consider the ‘five Ws’ of AI (see Figure 1 in executive summary).

As we shall see, however, there are nuances to each of these that highlight how AI adoption in the financial industry has been an evolving phenomenon. We also identify four key ‘value chains’ where AI tools are making an impact as ‘cogs’ within larger processes. In the companion market landscape report we’ll explore how vendors are exploiting four main areas of these value chains: data management, workflow and automation, analytics and packaged applications.