Edouard Legrand, chief digital officer at Paris-based BNP Paribas Asset Management, discusses the potential of artificial intelligence in the wealth management industry and the best way to exploit it. 

Artificial intelligence – now a hot topic – is full of potential and can be captured by the world's asset management sector, according to BNP Paribas.

Service offerings and companies touting their use of artificial intelligence for all kinds of purposes are springing up like boom towns in the time of the gold rush, Edouard Legrand at BNP Paribas Asset Management said. 

Benefits range from automating repetitive tasks, providing more personalised, data-driven advice in specific areas such as portfolio optimisation, risk management and tax analysis. In agriculture, AI enables farmers to monitor crops more effectively, reduce errors and minimise the risk of crop failures, for instance. But will we all benefit from this new technological gold? (This news service had thoughts about the issue here.)

Legrand believes that the rush for AI is comparable to the discovery of oil or any new energy source to improve human productivity. Indeed, it is a source of operational efficiency and transformation, but its exploitation comes at the cost of significant investments and an organised approach. “As with any energy, it is necessary to control its extraction, storage, flows and conversion into a business use beneficial to its activity,” he said.

So how can an asset management firm adapt to this new dynamic? Legrand thinks several parameters must be prepared to get the most out of AI, as outlined below. 

Building a solid infrastructure
First and foremost, data is important of course because it is the raw material for developing artificial intelligence, Legrand said. And to exploit the data, an ecosystem must be gradually built up. It means, for example, setting up a data hub that combines significant storage capacities (for structured and unstructured data), but also different levels of services so that the data is as close as possible to business challenges (dataviz, API – or application programming interfaces – data science platform).

In parallel, it is also important to establish a data dictionary, governance with data objectives for many employees, quality control plans, he continued. Finally, by capitalising on this work, it is important to remain open, by integrating external data or know-how on its platform. Also, with this ecosystem, a management company may be able to work more easily with new technological elements: synthetic data, integration of large language models (LLM), Legrand said.

Controlling risks and anticipating impact
There is a second area on which Legrand said is important: the risks associated with artificial intelligence and data. He noted that in an ultra-connected world, data leaks are an ongoing concern, and cybersecurity teams must constantly work towards developing models and data in a healthy environment. Being vigilant to ethical topics should not be neglected either, such as how to guarantee the traceability of results produced by an AI model. The same goes for reliability: AI has biases like humans, so it is necessary to evaluate the models and their answers before proposing them to customers, he continued. In addition, the operating costs of the most efficient AIs are significant. Each query generates complex calculations and is energy intensive. Legrand thinks that it is therefore critical to properly assess the energy and environmental impact of the use of AI, when sustainability is put at the heart of the strategy. It is still difficult to accurately assess these consumptions, but it is imperative to optimise them, Legrand said.
 
Educate teams and transform data into business benefits
The science and technology behind AI is complex and constantly evolving. Legrand believes that it is therefore necessary to have training to get the best out of it. Finally, and most importantly, the exploitation of data must be for the benefit of the business.

BNP Paribas Asset Management said it has already created natural language (NLG) comments for 300 of its funds, a chatbot (NLU) developed in-house for its employees, designed automated multi-factor allocation (MFA) portfolio management solutions for management and digitalised client journeys. Because using AI without a specific purpose related to the business does not make sense.

By applying this organised approach, Legrand believes that asset management firms will benefit from the development of artificial intelligence and take advantage of it in their management processes and customer relations. “Some will say, as the popular saying goes, that during the gold rush it was mainly pickaxe and sieve merchants who made their fortunes,” he added.

“Sales of cloud infrastructure or graphics cards/GPUs have a bright future ahead of them, but those who have met the conditions for the effective use of AI (like all those who have not missed a technological shift before them) will also be in a position to transform their business and gain competitive advantages, asset management being no exception,” Legrand concluded.