How AI Transforms Financial Services
Financial services is a sector that is typically slow to adopt new technologies. Barriers due to regulation and privacy laws like GDPR and CCPA, legacy systems, sensitive data, and aversion to risk, ensure a slow rate of change. However, increasing consumer demand for better services in a digital ecosystem forces firms to look at AI investments to move forwards. One report shows that a significant number of financial institution executives from 33 countries expect to become mass adopters of AI within the next two years. The technology will be a key business driver.
This article will discuss how AI is a transformative technology for financial firms.
Using data for fraud prevention and anti-money laundering strategies
The Covid-19 pandemic has seen accelerated growth in businesses moving to digital channels for e-commerce and customer service. For financial institutions, this raises the level of threat from cybercriminals. Consumers are very aware of how their data can be used maliciously, thanks to high profiles cases such as Cambridge Analytica. AI can help institutions protect themselves from fraud.
A machine-learning algorithm uses data to find patterns and anomalies in transactional data. In most cases, that includes purchases, personal data, online behaviors, and social media information. Advanced deep learning models will process the data in real-time, flagging any irregularities to the financial services provider’s bank. When a mistake is made, it will understand the problem and correct itself next time.
A typical example of this is in anti-money laundering strategies. AI is prone to generating false positives, which means legitimate transactions create alerts of suspicion. In a paper by Accenture, they propose a triage process to help machine learning recognize false positives and automatically suppress them next time they occur.
Despite some minor flaws in AI for fraud and money-laundering purposes, it is far more efficient than having a human sift through thousands of applications and transactions.
AI-Powered analytics for granular insights
Data can help firms make critical decisions based on the likely future result of an action. The field of predictive analytics uses AI, known as machine learning, to spot patterns in data and generate real-time actionable insight. One of the leading examples in financial services is credit scoring.
AI helps make the credit scoring process more accurate by ingesting multiple data points, such as personal information, transaction history, and documentation. Applicants receive a rate based on those with similar profiles as the firm providing credit gains insight into the future risk of providing credit.
Amazon partnered with Barclaycard to offer financing for purchases made through Amazon.de above EUR 100. Barclays and Amazon link data with AI analysis to approve credit (or not) and predict what customized services clients will want next. Automobile companies in the US report that bringing AI onboard cuts losses by 23% annually.
At Ford Motor Company, machine learning algorithms help predict customers’ risk that only have a thin-file credit history. The large tech firm SAS offers a predictive analytics application called Credit Scoring for SAS Enterprise Miner. Piraeus Bank Group uses the platform to speed up data analysis and the generation of reports. The software optimizes the development of risk prediction models and understands the future better.
The benefit of AI for credit decisions is that everything is fact-based. There is no emotion involved, and credit can either be provided or rejected depending on the supplied data. Ultimately, it makes AI a more rational credit tool than its human counterparts.
Robo-advisors for trading - faster and more accurate decisions
Trading platforms need to use a lot of data to make the right investment decisions. Machine learning algorithms will augment traditional trading strategies by automatically processing data, make recommendations, and predict how successful a trade is likely to be for the trader.
Using AI in this way allows human advisors to focus on the customer rather than spending hours sifting through data. It is less prone to human error, works 24/7, and offers greater accuracy in results.
Algorithms can review a more extensive pool of data faster than any human can. Investors are getting quicker decisions and have more confidence that a broad spectrum of information is used to create insights.
In trading, humans can be influenced by their own emotions or opinions about a specific trade. Where AI bases recommendations purely on facts, the investor receives decisions that are devoid of emotion and likely to be a safer bet. Robo-advisors combine neural networks and decision trees by exploiting non-linear relationships in historical data. Evidence-based decisions are made instead of following a gut feeling. Major enterprises are already deploying a large number of successful robo-advisors.
At Faktion, we’ve deployed AI robo-advisors for Bank van Breda to generate smart cross-selling recommendations for its account managers. Read our customer case, “Lead Scoring and Recommendation Algorithms.”
We’ve also implemented AI for Vivium. The application we built automatically analyzes complaint emails to the sales department to understand better what was happening in the insurance brokers’ world. The gained insights have enabled the firm to optimize several processes and proactively address detected problems before becoming unmanageable. By using the Faktion NLU engine, each email receives one or several labels (e.g., product, stage in the customer journey, sentiment), and relevant entities are extracted (e.g., product name, policy no.)
Using Conversational Chatbots for better service
Conversational chatbots are starting to become a must-have application for customer service teams to reduce resource-heavy tasks and improve efficiency. Consumers are now accustomed to chatbots due to the popularity of Alexa, Google Home, and Siri over the last decade. The technology is driven by AI, which takes human speech, converts it into data, and converts it into the most appropriate response through machine learning algorithms.
Conversational AI can help customers with queries in the financial services sector that are generally quite complex. For example, reading through the terms of a mortgage contract can take time, and they need to speak with someone to ensure they understand everything. There are already several examples of finance firms leading the way with chatbots.
Amy from HSBC is specifically designed for corporate banking customers to provide comprehensive answers through a convenient interface.
Chatlayer.ai, a Belgian chatbot provider, developed a chatbot for the assurance company Vivium. This project aims to answer questions from brokers that have electronically submitted an M202 claim regarding mandatory but missing fields through a chatbot.
Bank of America is a leader in mobile banking and AI implementation in the US. The Erica chatbot can send customer notifications, provide balance details, and recommend how customers can save money.
As customers continue to favor digital communications, financial services need to meet the demand with chatbot technology.
Back-Middle office automation
Whether it’s the credit department in a retail bank, the new issues department for global custodians, or the claims department of insurance companies, they all have something in common: managing vast amounts of unstructured documents and emails that still require the human eye to process and interpret the materials for further handling.
Applying natural language processing techniques makes it possible to automate the human language’s interpretation even for unstructured text sources. Whether it is for intelligent routing of emails and documents or complete STP (straight-through processing), machine learning makes the life of both employees and customers more comfortable. Reducing the time spent on repetitive manual labor will enable your employees to focus more on intricate and valuable tasks, e.g., customer interaction.
A concrete example of an STP case is the automated control of supporting documents linked to credit applications during the loan origination. Nowadays, intelligent document processing solutions, such as Metamaze, enable retail banks to automatically scan/read and interpret the documents that identify their type, verify the presence of signatures, and extract relevant fields to match with data within the same document and check against data from other sources. The latter is possible by combining the latest, state-of-the-art technologies in the field of OCR and NLP.
The second example of a low-threshold machine learning project for financial institutions is automated email analysis and routing in the insurance industry. An enormous amount of emails are exchanged between insurance brokers and their clients daily. The automatic classification of these incoming emails (e.g., product, stage in the customer journey, sentiment) enables insurance firms to reduce the time spent on routing emails to the correct person, optimize internal processes and proactively address detected problems before they become unmanageable.
What does the future of AI in financial services look like?
In this article, we have looked at some of the most significant ways AI is transforming financial services. Major banks, including JP Morgan, HSBC, Citi, and US Bank, are already heavily investing in AI solutions as they look towards the future. New technology like 5G, edge, and quantum computing will continue to disrupt the financial landscape. Simultaneously, unprecedented events like Covid-19 show how crucial it is to have automation in place when human resources become limited.
AI is still in its infancy within the financial services industry. The world in 2030 could look quite different as new technology platforms become mainstream and FinTech startups bring new ideas to the marketplace.