AI, Deep Learning, and Financial Services
Financial services can benefit from the many competitive use cases enabled by artificial intelligence and machine learning, but adoption has been slow.
- By Brian J. Dooley
- March 13, 2017
As we discussed in a previous article, "Analytics in The Stock Market: Made for Each Other," the financial services industry would seem to be a particularly good fit for current advances in artificial intelligence (AI) and machine learning. After all, financial services are based on understanding risk and balancing a wide range of numeric factors as well as predicting trends.
However, uptake of machine learning and AI has been relatively slow in this industry. The chief reason is the need to maintain a conservative outlook and to accommodate a wide range of processes and systems based around legacy information technology. Despite this, we will likely see some change soon. One notable sign: start-ups are beginning to offer financial services based specifically on emerging technologies.
Financial Use Cases for AI
A survey of 424 senior executives in financial services industries released by Baker McKenzie in March 2016 found that 49 percent of respondents expected their organizations to use AI in risk assessment within the next three years. Additionally, 29 percent expected to use AI in customer understanding and anti-money laundering solutions, and 26 percent expected to use it in risk, regulation, and compliance activities. Most of these activities are peripheral to the core business, but they provide experience with the technology and support established AI use cases.
The adoption of AI applications in financial industries is partly driven by the amount of data available. The increasing availability of storage and mandates requiring longer record retention have recently led to a massive increase in stored financial data. This data can now easily be used by deep learning and other AI tools that derive patterns from enormous pools of data.
For corporate financial departments, the growth of AI and machine learning is likely to be more variable. Specific applications might depend on which skills are available within the firm or initiatives being undertaken in other areas. Marketing is often a leading driver, but enterprise risk avoidance is also likely to be important.
What activities will deep learning be applied to within this sector? We are already seeing it used in fraud detection, compliance, correlation and linking of multivariate transaction data, analyzing behavior data captured from online activities, analysis of international trade flows, analysis of corporate interconnections, portfolio management and creation of virtual advisors, mortgage risk analysis, algorithmic trading in securities and FOREX, and loan and insurance underwriting.
Although many of these activities are currently confined to islands and to niche start-ups, the growing acceptance of the possible benefits will lead to continued expansion.
Adoption Challenges
One of the problems with financial decision making by algorithm, particularly through "black box" machine learning, is the gap between transparency and trust. Users of the machine learning solution need to be able to trust its output because they have no insight into the evaluation itself.
Before trusting a new algorithm, however, the enterprise must review it thoroughly. With the numerous regulations governing financial data, it is possible for such a process to unintentionally breach ethical rules -- for example, profiling individuals according to race or gender.
Another problem could include false incoming data if a data source is poorly chosen, incorrectly integrated, or able to be manipulated by a third party. These types of factors will need to be reviewed on a case-by-case basis, but awareness of these issues is an important starting point.
The Future of Finance
Deep learning and other AI techniques provide tremendous advantages to forward-thinking companies both within the financial services industry and within the financial departments of other firms. Dealing directly with the profit and loss mechanisms, new services in this sector can yield a critical competitive edge.
Although IT advancement has been relatively slow and conservative in finance, AI is capable of changing this and creating an arms race that just might transform the way financial technology evolves.
About the Author
Brian J. Dooley is an author, analyst, and journalist with more than 30 years' experience in analyzing and writing about trends in IT. He has written six books, numerous user manuals, hundreds of reports, and more than 1,000 magazine features. You can contact the author at [email protected].