As e-commerce has risen in popularity, so has online fraud. Yet fighting online fraud is very challenging. Declining transactions too aggressively to prevent fraud can be a self-defeating goal. According to a 2015 study by research firm Javelin Strategy, false declines, legitimate transactions that are wrongly rejected, account for $118 bln in losses for retailers. A third of false decline cases result in lost customers, and in US alone they incur damage that is worth 13 times the value of actual fraud.
Artificial intelligence can come in handy here. By analyzing various data points, machine learning algorithms can detect fraudulent transactions that would go unnoticed by human analysts while improving the accuracy of real-time approvals and reducing false declines.
A number of companies are exploring AI-based fraud prevention. One example is Mastercard’s recently launched Decision Intelligence technology. Instead of limiting itself to predefined rules, DI gleans patterns from historical shopping and spending habits of cardholders to set a behavioral baseline against which it will compare and score each new transaction.This is a major improvement over traditional prevention technologies, which rely on a one-size-fits-all approach to evaluate all transactions. While Mastercard isn’t the first financial firm to employ artificial intelligence in fraud detection, the billions of transactions it processes every year gives it plenty of data to train and hone its algorithms.
Other companies such as Sift Science employ a more holistic approach. Sift Science collects data from more than 6,000 websites where its fraud detection solution is deployed. This enables it to track and analyze data across multiple channels and devices. The engine correlates different data points including payments and activity on websites to create better models of customer behavior and detect fraudulent transactions.
Banking chatbots
In past years, chatbots powered by natural language processing (NLG) and machine learning algorithms have become a powerful tool with which to provide a personalized and conversational experience to users in different domains.
There are several ways that AI chatbots can improve the banking industry, including helping users manage their money and savings. Plum, a chatbot accessible through Facebook Messenger, helps you save money in small increments. When registering, you connect Plum to your bank account, after which the AI engine behind it analyzes your income and spending habits and predicts how much you can afford to save. It then deposits small amounts to your Plum savings account at opportune moments, and reports to you periodically.
Another example is Cleo, a chatbot that assists you in tracking your income and expenses across multiple accounts. The chatbot lets you query your financial data in a conversational manner, as if you’re speaking to a personal accountant. The assistant can also help you by providing tips on how to manage your money and save for future plans.
Banks are also dabbling in the chatbot business to improve their self-service interfaces, an area that is generally attributed with poor quality. Bank of America plans to launch its AI chatbot Erica (a play on the bank’s name) later this year. The digital assistant, which is available through voice or message chat on the bank’s mobile app, will help you make faster and smarter decisions. Instead of navigating the app’s UI, you can command Erica to, for example, send money to a friend or pay a bill. The chatbot’s AI engine also leverages analytics to assist you in managing your personal finance. For instance, it can help you achieve a savings goal by making suggestions based on your income and spending patterns.
Algorithmic trading
If there’s one thing computers have always been good at, it’s crunching numbers. Thanks to machine learning, they can now take on the subtleties and complexities involved in tasks such as trading stocks. A handful of hedge funds are exploring the concept, and have managed to obtain results that rival the intuition of human experts.
Sentient Technologies, an AI company based in San Francisco that also runs a hedge fund, has developed an algorithm that ingests millions of data points to find trading patterns and forecast trends, which enable it to make successful stock trading decisions. Sentient runs trillions of simulated trading scenarios created from the vast amounts of public data available online. Its algorithms use those scenarios to identify and blend successful trading patterns and devise new strategies. These techniques enable the startup to squeeze 1,800 days of trading into a few minutes. Successful trading strategies, which it calls “genes,” are then tested in live trading, where they evolve autonomously as they gain experience.
Another hedge fund, Numerai, uses artificial intelligence to make trading decisions. Instead of developing the algorithms themselves, they’ve outsourced the task to thousands of anonymous data scientists, who compete to create the best algorithms and win cryptocurrency for their efforts. Numerai shares trading data with the scientists in a way that prevents them from replicating the fund’s trades while allowing them to build models for better trades.
The jury is still out on how effective AI will be in mastering the intricacies of stock trading, which are often impacted by the most unexpected and unpredictable parameters, such as Twitter rants by the US president. The practice still has many skeptics, especially with traditional traders who are dubious about the lack of transparency in AI algorithms. However, what’s evident is that algorithms can provide invaluable insights and suggestions that will help the humans running the operations make better decisions.
The future of AI in finance
Artificial intelligence as we know it today is still in its infancy, and has hurdles to overcome, including legal, ethical, economic and social challenges. However, the prospects for smarter trading, less damage and a more personalized experience are great. The future of money just got more exciting.
