Impact of Artificial Intelligence on Banking
Like many financial institutions, Security Pacific National Bank had a problem with fraud, specifically fraudulent use of debit cards at automated teller machines and sales counters. And, like other financial institutions, Security Pacific discovered that fighting the problem by trying to recover losses was far less effective than preventing fraud in the first place.
Fraud investigations, mostly conducted after the fact, were hampered by the cumbersome process of manually reviewing reports from several sources. So, in 1987, the bank’s corporate security department authorized its new Fraud Prevention Task Force to use an “expert system,” applying artificial intelligence technology.
The intent was to use artificial intelligence in an early warning system to prevent fraud. In three months, a prototype was built on a small departmental computer using the bank’s and the industry’s knowledge of rules weighing evidence of fraud.
The prototype was successful enough that the application was moved to a larger computer and, in 1988, the on-line fraud detection system paid for itself in reduced losses in two months.
Despite such dramatic results, the use of artificial intelligence in financial services is still relatively low. But enough companies have had sufficient experience with these systems to demonstrate their value. So far, expert systems have been used more widely in manufacturing, but financial services should be fertile terrain for the growth of artificial intelligence because the industry relies heavily on mental activities and the growth of knowledge. In the financial world, the quality of knowledge constitutes the basis of competition.
Notwithstanding the apparent match between the technology and typical functions in the financial services industry, this sector accounts for a very small percentage of artificial intelligence vendors’ business. There are a number of reasons for this disparity:
* There is a mismatch between the tools, hardware and languages--many of which are highly specialized--required for artificial intelligence and the equipment and skills in the financial community.
* Most efforts to develop artificial intelligence systems reflect an engineering and scientific approach toward experimentation and prototyping, an approach at odds with the controlled system development approach usually used in the financial services industry.
* Turnkey expert systems, which often employ a generic representation of knowledge, do not have good track records in the financial services industry, where companies rely more heavily on their own knowledge of markets and systems.
* Artificial intelligence has taken a back seat as financial institutions have had to deal with a blizzard of new, more pressing technologies such as personal computers, distributed processing and transaction processing.
Despite artificial intelligence’s modest reception in the financial services industry, there is reason for optimism. A recent survey showed that nearly 33% of brokerage firms, 50% of banks and 80% of insurance companies are considering investing in artificial intelligence. Moreover, it helps that the technology is changing in terms of what vendors are offering and what is in place at companies. Developers and users are moving from mainframes and PCs to increasing reliance on business workstations and networks.
In addition, applications where artificial intelligence has a clear and direct benefit are being identified. In credit analysis and authorization, network management and product presentation, for example, artificial intelligence is particularly appropriate for financial planning, which requires knowledge of a broad variety of products.
Financial planning is a good example of the advantage that artificial intelligence has over conventional information systems. For a conventional system, the more complex the depth and breadth of its knowledge, the less ability it has to recommend actions. If such a system knows which kinds of investments are eligible for an individual retirement account, for instance, it can recommend an action.
However, if the system is asked to consider 10 alternative investment vehicles, it has more difficulty making a selection.
With artificial intelligence systems, however, both the level of knowledge and the level of ability can be expanded to offer a more complete planning tool.
One New York bank was looking for a way to provide affordable expert financial planning to people with incomes between $25,000 and $150,000. The artificial intelligence system that was developed over a five-year period produces a customized plan incorporating expertise in investment planning, debt planning, retirement savings and plan settlement, education funding, life and disability insurance planning, budget recommendations, income tax planning and saving for major financial goals such as a home purchase.
The system’s $5-million cost included system development, product development, marketing and promotion. The bank’s fee for this service starts as low as $500, while comparable plans offered elsewhere can cost as much as $10,000.
While artificial intelligence is used far less in financial services than in manufacturing, many of the benefits for the financial services industry sound like the same sort of goals being sought by other service industries and manufacturers. These include boosting productivity by putting complex knowledge in the hands of more people who can use it; managing quality, particularly when it translates into a consistent and timely approach to handling cases, solving problems by tapping professionals’ knowledge to use in training programs and gaining the competitive advantage from employing artificial intelligence to use shared knowledge.
Firms that have used artificial intelligence to reach these goals have gained a two- to five-year edge on their competition.
The ability to apply artificial intelligence widely throughout the financial services industry makes it ripe for exploitation. Total investment in artificial intelligence products and services could grow by more than 40% a year, from slightly more than $50 million in 1987 to more than $310 million in 1992--roughly twice the overall growth rate of information systems in financial services.
For banks, brokerages, insurance companies and others in the financial industry, the link between knowledge and competitive advantage is becoming increasingly strong. It is now up to vendors to show that artificial intelligence can help financial companies gain a competitive advantage through the use of technology.