Most risk professionals see value in AI, but feel they lack skills to implement or maintain it - survey
- 01 March, 2019 06:30
Artificial intelligence (AI) is transforming all industries, including financial services.
A new survey from the Global Association of Risk Professionals (GARP) and SAS showed that 81 per cent of risk professionals in the financial services industry have already seen benefits from AI technologies.
The top areas where respondents are reaping those returns include improved process automation (52 per cent), credit scoring (45 per cent) and data preparation (43 per cent).
Around one-third of respondents also reported seeing benefits from model validation, calibration and selection.
The survey was conducted last December among more than 2,000 representatives from across the financial services industry, and included respondents from New Zealand and Australia.
For the survey, AI included machine learning, natural language processing, computer vision, forecasting and optimisation.
“At this point there’s little doubt that AI is here to stay, and that is no different for risk professionals and financial services firms,” says Mark Carey, co-president at GARP.
“While more than half of survey respondents described at least moderate knowledge of their firms’ current and planned use of AI, the survey suggests institutions are still very much exploring AI, with a lot of questions remaining.”
Risk and financial service professionals who haven’t yet deployed AI plan to do so soon.
According to the survey, of those not yet using any form of AI, 84 per cent plan to be using machine learning and natural language processing in the next three years.
Additionally, in the next three years, almost all respondents expect AI to improve their jobs at least somewhat.
They anticipate AI will lead to higher productivity (96 percent), faster time to gain insights from data (95 per cent) and more data insights for faster, better decisions (95 per cent).
At the same time, the respondents expressed concern about how their jobs may change - or even be eliminated.
Another main concern is the skills gap to using AI.
More than half (52 per cent) of respondents said they were at least somewhat concerned that their firms lack the necessary skills to implement and maintain AI.
Other challenges loom. A large majority of respondents said that among the biggest obstacles are data availability and quality (59 per cent), key stakeholders’ lack of understanding of AI (54 percent) and interpretability of models (47 per cent).
SAS senior data scientist Katherine Taylor recommends equipping existing risk talent with data-science skills, enabling them to better map the new technology to the problems that must be solved.
Establishing central teams for AI governance appears to aid in adoption of the technology across the company, Taylor says, adding that it is important to introduce the technology by tackling a well-defined, real-world problem.
She says one investment bank, for instance, turned to neural networks to improve approximations of real-time portfolio repricings, because traditional methods were breaking down under stress.
The neutral networks indeed ran faster and provided much better approximations.