Decisions in AI: Complexity vs. Function

Systems have become and continue to be more complex and more interconnected. This has led to astounding advancements and uptake in the practical applications of technology – from the analysis of huge datasets to automation, processing, and the application of logic. Add to the mix the more recent advancement in autonomous machine learning and reasoning, and the proliferation of systems designed to automate many of the functions that were deemed only capable by a human is set to grow.

Paradigm Shift in System Development

Consequently, the realisation that the traditional methodologies of architecting, building, testing, and auditing of these systems are no longer appropriate has triggered a paradigm change. Model-based software development, advanced automated testing, and even, ironically, the use of AI to verify AI-based systems. As a result, we find ourselves in situations where we do not fully understand the systems which we increasingly use.

There is a risk vs. reward conversation around the complexity of the system and the extent of the function it should fulfill. Take autonomous cars. The technology has been around for years. However, the reason we do not see roads filled with driverless cars (yet) is due to the risk of system malfunction and its consequences.

Application of AI in Wealth Management

The same risk vs. reward applies to the use of Generative or other variants of AI in Wealth Management – the most likely outcome in the foreseeable future is that it will not replace the Wealth Manager but act as a co-pilot. Think more SatNavs than fully driverless cars.

The other aspect to consider is the complexity of the use case. Again, using the autonomous car example, making technology responsible for the safety of a passenger traveling across town may be too big to swallow. Similarly, using AI to completely automate the Wealth Management function is too broad a use case. Breaking this down and using AI to automate specific manual tasks and provide contextual, accurate insight based on datasets is the nearest stepping stone.

It is, however, important to highlight that the technology roadmap for the vast majority of firms will include an AI stream – any that do not will be at a disadvantage to their peers so it is in their interests to embrace rather than fight the tide. The challenge for each firm is determining how Generative AI can benefit each specific process within their business and what the cumulative value would be.