If responsibly managed, the rise of artificial intelligence can greatly improve financial services.
Advances in artificial intelligence (AI) have made a big splash over the last few years. Autonomous vehicles, warehouse worker bots, and package delivery robots are all making headlines and stretching the imagination. The financial sector is not immune to these advances in AI, and Congress has begun to seriously explore the implications of this. For instance, the House Financial Services Committee has recently set up the Task Force on Artificial Intelligence and the Task Force on Financial Technology.
Policymakers will be expected to pay serious attention to the potential problems and challenges with artificial intelligence, such as algorithmic bias and privacy concerns. This is a good thing. These issues will greatly affect how well the financial system serves Main Street and should be addressed.
However, only focusing on the problems would paint an incomplete picture of AI’s role in the financial system. Discussions about how the rise of AI can benefit society are also important. Policymakers should ask about how AI can responsibly expand access to credit, promote stability, and better serve consumers. Financial services can be greatly improved by AI, and we should not lose sight of this when managing the challenges.
The recent advances in AI most applicable to financial services tend to be in machine learning. Machine learning algorithms use data to “learn,” identify patterns, and make predictions. A good example of machine learning algorithms are those used to read numbers on checks. These algorithms are fed large data sets that include images of numbers and their value to help develop a way to predict the value of inputted checks.
Check reading algorithms can help explain a couple of important points about the limitations of machine learning.
First, these algorithms are heavily reliant on data to improve their accuracy and “learn.” If given too few data points, the algorithm will have trouble identifying numbers. The algorithm will also have trouble if it uses data sets that are biased in a certain manner. For instance, if the algorithm is only trained with numbers that are printed, then it should have trouble reading numbers that are handwritten.
Second, these algorithms are dependent on the objective given to them by humans. If the algorithm is designed only to identify numbers on a check, it will just focus on that. It will not focus on trying to identify whether a check was forged. However, if the creator modified the algorithm and gave it the appropriate data, the algorithm could be designed to identify forgeries as well.
Check reading algorithms also illustrate the potential of machine learning.
Advances in machine learning, processing power, and data collection can result in faster, cheaper, and more accurate predictions in many situations. In the case of check reading algorithms, these improvements can reduce costs, decrease error rates, and speed up the process for cashing checks to the benefit of consumers.
The role AI plays in helping identify patterns and making predictions is important, but so is the role of human judgment in guiding AI and its applications. Advances in AI can play a key role in making the financial system more stable, inclusive, and effective, provided financial firms implement AI responsibly. A few interesting applications of AI that benefit consumers and the public include:
- AI can monitor credit card transactions to better detect fraud and protect consumers. A machine learning algorithm can monitor a person’s spending habits to predict when a transaction is potentially fraudulent or a credit card has been compromised. This can help reduce fraud and create savings that can be passed on to consumers through lower fees and interest rates.
- AI can monitor traders to ensure they don’t go rogue and jeopardize the stability of a financial institution. Machine learning algorithms can monitor a trader’s work emails and trading habits to help determine whether they are engaging in suspicious activity and making unauthorized trades. The compliance team can use this information to better oversee traders and prevent them from taking reckless risks that endanger the firm. This can help promote greater financial stability and protect the financial firm’s various stakeholders, including shareholders and employees.
- AI can give regulators tools to better monitor the financial sector and protect consumers. Regulators can use machine learning algorithms to identify bad actors or suspicious behavior, so they can more effectively allocate their limited resources.
- AI can help improve customer experience by speeding up the process and better tailoring services to consumers. Natural language processing can help identify the purpose of a customer’s call and direct the customer to the appropriate person at a financial firm. Improved call sorting would free up customer service representatives from directing calls, allowing them to specialize in answering specific types of questions instead.
- AI can give insurers new ways to evaluate risk, so they can more efficiently set prices. A property and casualty insurer can use a machine learning algorithm to better predict weather patterns to determine the likelihood of a natural disaster in a region. This can help more efficiently set insurance premiums, which could better signal the riskiness of a location to potential homebuyers.
These applications and others can benefit the public, but only if they are done in a responsible and consumer-friendly manner. An algorithm for guiding customer calls designed to direct those most susceptible to high pressure sales tactics towards unscrupulous sales representatives would be harmful for society. This is a big reason why human judgement and ethics will continue to be critical.
The challenges in adapting to new advances in AI are real, but they should not blind us to its promise. If responsibly managed, the rise of AI in the financial sector can greatly improve living standards, promote stability, and serve consumers. Policymakers across the political spectrum should keep an open mind and consider these benefits when making policy decisions about AI in finance.