Thanks for conveying your interest!

You will get notified on the email you provided.
twitter Connect with us on linkedin

AI in finance

Jure Brence, November 2021

According to EC Europa, more than half of adult Europeans used online banking in 2017. Since then, the figure has risen to over 75%, and is projected to keep increasing, a process that has without doubt been accelerated by the Covid-19 epidemic and the associated lockdowns and social distancing measures. Similar trends have been observed in retail investing, with an explosion in popularity of online brokers suchs as Webull, Degiro, Robinhood, IBKR and many others, especially among younger adults.

The field of finance has a long tradition of developing and employing methods of data analysis and relying on forecasts of the behavior of highly complex systems. As such, it should come as no surprise that banks, investment funds and insurance companies have been quick to grasp the value of modern advancements in artificial intelligence.

In this blog post, we take a look at how financial institutions apply artificial intelligence to enhance their risk management, obtain better market predictions, protect investors from fraud and help retail customers manage their personal finances.

Cybersecurity and fraud detection

As customers put more and more trust into online finance, the impact and magnitude of potential security breaches increases dramatically. In a similar vein, opportunities for malevolent actors to deceive and defraud multiply as many customers struggle to keep up with the accelerating pace of technological and societal change.

To deal with these issues banks and personal investing firms already rely heavily on AI. With mind-boggling numbers of transactions occurring every minute, it is impossible for banks to manually verify whether each transfer of money is legitimate. Instead, AI is trained to recognize behaviors, specific to fraudulent activity, and alert humans to look into it closer. One such example are F5 products, which include AI for fraud detection and prevention of security breaches, used by 48 Fortune 50 companies.

Personalized finance

The digitalization of banking is opening a new frontier for the experience banks can offer their customers. From virtual assistants for customer support to budgeting tools and subscription optimization, personalized finance is making its way to customers.

Chatbots for customer support are becoming more useful and humanlike with every advance in natural language processing. An early example of a banking virtual assistant was ENO by Capital One, while today many enterprise solutions are on the market, such as the conversational AI platforms, offered by Kasisto. On the other hand, third-party services like AskTrim employ artificial intelligence to analyze the subscriptions and spending habits of customers and identify personalized money-saving actions and strategies.

Quantitative trading

Profitable trading often requires accurate predictions of trends in financial markets, which can be understood as collective behaviors of highly complex systems. This is a difficult task, where artificial intelligence analyzing huge amounts of data can make a big difference. Modern tools take into account data from many different sources, ranging from company fundamentals and technical analysis, to AI-processed documents such as SEC filings and new articles, as well as social media sentiments.

Solutions such as Canoe and AlphaSense perform automated market research and provide clients of varying sizes with condensed information that enhances the performance of decision makers in investing. Due to the rise of high-frequency-trading, most big hedge funds and institutions, such as Citadel LLC, Tower Research and Virtu Financial, have already been trading in a fully automatized manner. Advancements in AI are taking algorithmic trading to the next level, with products such AIAutoTrade entering the market and many financial institutions developing their own solutions.

Risk assessment

A key concept when making investment decisions is risk assessment and management, which is at the centre of the insurance and lending businesses. Traditionally, banks relied only on credit scores, basic employment information and downpayment to assess whether to approve large loans, often erring on the side of caution.

Artificial intelligence can process much larger collections of information on potential borrowers and provide more nuanced estimates of financial risk to the lenders. This can help not only lenders by maximizing profit, but also customers by allowing credits for traditionally underserved, younger populations through more accurate and holistic risk assessment. An example in this application is Enova’s AI platform Colussus, which improves the risk assessment and loan pricing for NetCredit, Simplic and CashNetUSA.

Interpretable and accountable AI

Insurance and load underwriting are decisions that have large and very real effects on people’s lives. Leaving such decisions completely to AI can lead to unanticipated issues, such as the gender discrimination case of Goldman Sachs from 2019, where an algorithm assigned a man a credit limit 20 times higher than that of his wife. This problem arose due to bias in the model and the lack of human oversight in the decision making process.

Issues of this sort can be prevented by being conscious of biases in data and more importantly, by employing interpretable AI that can explain and justify its predictions. An example of such software are Birlasoft products. Many insurance firms and banks prefer to leave the final decisions to humans, while using explainable AI to provide them with valuable decision support. Others seek to fully automate the process and work to ensure its fairness.

What about crypto?

Cryptocurrencies are on the cutting edge of modern financial technology and apply AI in various ways. Most of the concepts discussed in this article apply just as well to crypto as to traditional finance, such as trading coins, dealing with fraud, improving customer experience on exchanges and assessing risk on decentralized lending platforms.

Stay tuned for a delve into crypto-specific applications of AI.