are machine learning and artificial intelligence the future of economics?
achine learning is becoming increasingly important in finance and economics, reducing operational costs through automation and increasing productivity and therefore revenue. A recent IBM study revealed that 82% of enterprises are considering AI. Machine learning uses statistical models to draw insight and make predictions. In finance, large data sets are common and for machine learning, more data means more accurate results. The quantitative nature of the finance industry makes machine learning perfect for use in finance.
So, what is it used for? There are a growing number of uses of machine learning in the finance industry, one being in the stock market for prediction in algorithmic trading. For algorithmic trading, the programs use coded models to make very fast trading decisions, often making millions of trades per day. They are used in the stock market for analysing price patterns and forecasting future stock prices and index changes.
Another use of machine learning is “robo-advisors” which can offer up to 70% reduction in cost for certain services. These use algorithms to collaborate information and analyse it creating a financial portfolio geared to the goals and the risk tolerance of the user. The “robo” will then tailor investments across assets in different classes of risk to reach the user’s goal. It will adapt to changes in the user’s goals and real-time changes in the market to find the best fitting investments for each user.
Increasingly, artificial intelligence is being used for fraud detection and prevention. The system works around a list of risk factors, highlighting behavioural anomalies that may suggest fraudulent activity. It may block the user or it may require additional information, to determine whether fraud is being attempted. These systems can also evolve and adjust to new potential security threats. Paypal have adopted this deep learning technology resulting in only a 0.32% revenue loss due to fraud, compared with the 1.32% average that other merchants see.
Machine learning is less prevalent in economics. Without AI, non-linear mathematical models used to predict economic phenomena are too complicated to be used in a controlled way therefore simplifying assumptions are made to make them solvable (e.g. perfect information or perfect rationality – both of which are implausible).
Complex adaptive systems are systems which are formed from a network of different processes. They perform the task of recognising patterns in different economic environments. These systems have the ability to adapt to their surroundings due to their evolution in different “economic worlds” over an extended period of time. However, each different situation has different ‘local niches’ that means that one complex adaptive system can only ever be programmed for one specific situation. As well as this, the world is ever evolving and so by the time one system is programmed to work in one situation, the economic environment may now contain new variables which the system isn’t engineered to understand.
In the future the use of machine learning and artificial intelligence is expected to grow massively to dominate online security. It is expected to make much greater use of biometric data. As well as this, attempts are being made to make AI that understand emotions, for finance and economics this could be very useful especially for recognising doubt and uncertainty.
M