Regulatory technology or RegTech, along with wealthtech and proptech (property technology), is on the rise. Regulators are using it more and more and venture capitalists are investing in it, even though its take-up among firms seems to have plateaued temporarily.

Firms are now using machine learning in a range of business areas and in both front and back offices. They use it most commonly to detect fraud and in anti-money laundering operations. They also use it in customer-facing applications (e.g. customer services and marketing). Some firms also use it in credit risk management, trade pricing and execution, general insurance pricing and underwriting.

Firms could not think of any new risks that machine learning created but thought that it might, in the fulness of time, amplify existing ones. They thought that machine-learning apps might not work as intended if model validation and governance frameworks failed to keep pace with technological developments.

Firms 'validate' machine-learning apps before and after deployment, most usually by means of outcome-focused monitoring and testing against benchmarks.

Firms use a variety of safeguards to manage the risks associated with machine learning, especially alert systems and so-called "human-in-the-loop mechanisms." These can spot circumstances in which a model does not work as intended (as in the case of 'model drift,' which can occur as an app is continuously updated and asked to make decisions that are outside its original set of parameters).

Firms mostly design and develop machine-learning apps in-house. They sometimes rely on software vendors for the underlying platforms and infrastructure, such as cloud computing.

Most users apply their existing model risk management rules to their machine-learning applications. They know, however, that in future they might have to reconfigure them to take account of the increasing sophistication of machine-learning techniques.

In order to provoke further discussion on the subject, the Bank of England (where the PRA resides) and the FCA are going to set up a public-private talking-shop.

The price of not innovating

On a broader note, wealth managers in the UK have told a survey by Nucoro that only 55% of their brethren will survive the next decade if they fail to innovate digitally. Nucoro has also found that only about one-fifth of wealth managers' "client-facing interaction" is done digitally at the moment. When looking at the areas of their operations where they feel that digitisation can "add the greatest value," 29% cited administration, 24% cited communication with clients and 13% cited the onboarding of clients, but only 4% cited regulatory reporting.