Why a chatbot will manage your money  -  Perspectives on AI in Investment Management

Florian M Spiegl
January 24, 2017

Artificial Intelligence is emerging into a new spring and hopefully extended summer, breaking with the pattern of intermittent AI Winters the discipline has seen before. Against this backdrop, we see AI entering a broad field of applications, including Financial Services, where it is one dominant factor transforming the profession, alongside changing demographics and new regulation. The impact of increasingly powerful machine intelligence on the future of Stock Brokerage and Digital Wealth Management has been a recurring topic on recent panels and discussion forums. Taking a step back, I would like to offer a brief summary of my perspectives as shared at some of these events.

Assessing the domain of Investment and Wealth Management, the most significant impact of AI can be expected on two central and interrelated dimensions of the value proposition: the investment process and the client interaction.

Artificial Intelligence will be leveraged to provide automated advice to client segments-of-one.

Looking at the investment process, AI will be leveraged to provide automated advice to client segments-of-one, following two thrusts, portfolio construction, and continuous optimization.

Portfolios will be tailored to the particular client, a significant progression of the dominant Robo Advisor model that assigns clients to one of several pre-defined buckets with a fixed asset allocation and set of financial instruments. Customers will be provided with their own, customized bucket, derived from a holistic understanding of their financial situation and goals.

In addition to that, machine learning-powered algorithms will allow a continuous evaluation and optimization of all portfolio positions, in contrast to the current practice of periodic re-balancing of positions that typically aims to merely preserve the determined asset allocation. Automation will also allow a more holistic advice on a client’s wealth planning across domains and sophisticated risk management. Recent advances in specific techniques like deep learning have triggered renewed interest in AI-driven investment management. Some of this excitement will indeed leave into dead ends, but it is now more reasonable than ever to see machines help to manage portfolios and eventually take control autonomously, with humans only providing general oversight and direction.

In effect, AI will make the personal, overarching portfolio manager a reality for anyone, not just individuals with significant amounts of assets.

AI will transform the “last mile” of interaction with users and fundamentally change how client relationships emerge and evolve.

Reflecting on the second and just as interesting dimension, AI will transform the “last mile” of interaction with users and fundamentally change how client relationships emerge and evolve. This transformation is important as the winners will ultimately be determined by a superior user experience. There are two angles to this as well: data-driven personalization and new communication channels.

Within the trusting relationship with their investment advisor, clients are prepared to voluntarily share significant amounts of structured and unstructured data, which feeds algorithms to create highly personalized customer experiences. As a result, communication style and content are customized for the specific user. This will be further propelled by the growing ambition of financial service providers to truly understand their clients, increasingly drawing on methods grounded in fields like psychology.

The AI domain of Natural Language Processing enables machines to understand language and unlocks additional channels to communicate this content. The emergence of chatbots can be expected to shift more of the client interaction from costly human-to-human conversations to efficient, always-on machine-to-human interactions, again enabling broader access to personalized wealth advice. On a personal note, it was a perplexing experience for me to realize how quickly even simple chatbots enter the realm of social connection — one that we usually reserve for humans. In other words, the quasi-natural and more frequent conversations humans are going to have with machines will create a stronger bond than most of the traditional channels we have been using to interact with service providers.

While it is likely that this shift will happen via a waypoint of a hybrid, human-plus-machine setup, it can be reasonably expected that clients with simple money management needs will be served exclusively by intelligent machines. We are seeing early developments into this direction by pioneering digital wealth management firms like 8 Securities. The ever-improving accuracy of parsing and contextual understanding of meaning will make sure that system capabilities move up the curve to handle more complex client needs and setups.

Quintessentially, AI will enable machines to communicate with clients as naturally as a human would — within the domain limits of investment services.

The building blocks for this future ambition are all available today, but barriers to a broader adoption have to be overcome.

Taking these developments together — and with the appropriate scepsis — we can imagine a future of semi-automated and at some point fully autonomous investment advice from a machine that is always there with us, continuously optimizing a personalized portfolio and communicating like a human advisor that knows and understands us well. Given the advances we have seen recently, the question appears to be rather when this will happen, not if.

The basic building blocks for this future ambition are all available today — and we will progressively see them combined to form new platforms, technically powered by connective API layers like the Fabrik. Undeniably, this development may still be further out than we think today and certainly will meet substantial barriers to a broader adoption. One to highlight in the context of this reflection is the regulatory environment and the question whether an automated investment service provides merely guidance or investment advice — with the respective regulatory implications and questions around liability.

Barriers apart, there is utility in reflecting on the ultimate destination in order to play an active part in the journey.

Above are a few considerations in the domain of automated investment advice; interesting discussions lie ahead. These selective highlights are not a comprehensive investigation of this subject  -  merely a starting point for further thinking. If you are interested in joining the conversation and building next generation investment services, do reach out.