Banks began installing ATMs in the 1960s, and electronic card-based payment systems were introduced in the 1970s. The 2000s were inundated with 24/7 online banking, the 2010s saw the proliferation of mobile-based "banking-on-the-go". Now, it is indisputable that we are currently in the midst of the Big Data and AI era.
Even if you're not aware of how your financial institution utilizes, for instance, complex machine learning to thwart money launderers, or sifts through mountains of data to detect fraud-related anomalies, you have probably interacted with its customer service chatbot, which uses artificial intelligence. So let’s take a sneak-peak at the top AI and Big Data trends to look out for in 2022!
People deride banking hours for a reason. It seems that the banks are never open on holidays or weekends when you most need them. In the past, call centres were known for long waiting times and queues, and when operators finally attended the call, they were sometimes unable to resolve the problem, all of this was an inconvenience for the consumers. Artificial intelligence is changing that with Chatbots on call 24*7.
The use of chatbots, or automated conversational assistants, is one of the biggest advantages of AI. Unlike an employee, a chatbot is available 24/7, and customers have become increasingly comfortable using this software program to retrieve answers for mundane and simple questions as well as handle many standard banking tasks that would otherwise necessitate person-to-person interaction resulting in a longer turnaround time and a greater drain of resources.
Bank of America introduced Erica, its AI digital assistant, in 2018. With Erica, customers can perform simple banking tasks such as paying bills, getting financial advice, updating credit reports and viewing e-statements via Bank of America's mobile banking app.
Watch Erica function here - https://www.youtube.com/watch?v=0lrg83riPzo
Banking institutions can also inform clients about their potential to significantly reduce fees based on a thorough and rapid analysis of their transactional history using AI. The mobile app Empower flags duplicate services and high bills and suggests measures such as reducing subscriptions or negotiating for lower phone fees, and also suggests ways to reduce bank fees.
By using AI-based budgeting tools, customers can improve their financial discipline. Acorns, for example, allows customers to set budgets and receive alerts to help them stay on track.
Banks can employ AI-driven systems across the enterprise to combine relevant data to create a comprehensive view of a customer's total inflows and outflows so that they can offer advice on balancing daily and annual expenditures with wealth-building objectives.
Wealthfront, a digital wealth management tool, offers customers an investment plan based on their answers to a few questions. The process allows the customer to set their goals in practical terms, such as learning how much to invest to buy a home in five years, taking a year off to travel next year or retiring at 40.
When banks create and offer intelligent propositions, they need to make them available not only on their own platforms but also in other ecosystems that their customers are a part of. SBI launched YONO in India, designed to meet a wide range of retail customer banking as well as non-banking needs. With more than 100 merchants embedded in its online marketplace, customers can purchase groceries and book tickets through a single app.
Hence, banks, as well as fintech firms, are leveraging technology to provide superior customer service, the quality of the service being the main differentiator.
Advances in technology have enabled a torrent of real-time data, which can be leveraged to optimize every aspect of a financial institution, from marketing to cyber-security. Meanwhile, the incessant shift toward digital banking has posed numerous cybersecurity challenges and fraud risks. In the financial year 2021, the Reserve Bank of India reported bank frauds amounting to an astronomical 1.38 trillion dollars.
Financial institutions and banks are using technology and big data tools to combat economic fraud. With Big Data Analytics, companies can agglomerate and analyze data, uncover underlying customer patterns and behaviour typical of normalized behaviour, and identify and detect anomalies of these normal behaviours that may point to a phishing attack or security breach.
Through the development of predictive models based on historical as well as real-time data on wages, medical claims, attorney costs, demographics, weather data, previous claim records, criminal records, social media data, call centre notes, and voice recordings, companies are better equipped to detect fraudulent claims in their earliest stages. When a claim is registered, the insurer needs to scrutinize and examine the claim beyond mere claim details. It should review and evaluate the claimant's information. It should flag the claimant for further investigation if it detects anything suspicious about the claimant record.
The following example illustrates identity theft fraud, and how technology helped catch the culprit: In 2018, David Matthew Read, 35, impersonated actress Demi Moore and reported that her American Express no-limit card had been stolen. The fraudster obtained her social security number online as well as other personal information. Next, he intercepted Moore's new credit card at FedEx and went on a 25-day shopping spree while pretending to be Moore's personal assistant. Read spent over $169000 on the card online and in luxury New York stores. Surveillance cameras spotted him shopping at stores with the card, which led the FBI to identify him.
Banks have access to a large amount of customer data from sources like social media, call centre conversations, logs, and log files, and using this data can point out possible anomalous behaviour. If a credit card holder is flying by airplane and posts the status on Facebook, any credit card transaction during the flying period is abnormal and the bank may block the transaction. Social media data can also reveal information that indicates that the conditions described in a claim did not occur on the day in question. As an example, a claimant may claim that his car was totalled in a flood, though social media evidence shows that the car was elsewhere at the time of the flood.
Agents at call centres possess a greater level of insight into possible patterns of behaviour and relationships between other claimants and service providers when the first call is placed. Suppose an agent receives notice that a claimant has testified in an identical case six months ago. As a result of discovering other unusual behaviour patterns, the claim process can be stalled before it even starts.
This is just a glimpse of a number of ways data can be utilized to extract valuable insights that can alert us to fraud that is impending, which can serve as a deterrent to fraud and cut losses.
There has been an ongoing, unrelenting wave of disruptive startups challenging traditional banks and financial institutions. These digitally transformed entities are taking a huge piece of the financial services pie by leveraging data-driven technologies in an agile manner. Square Inc, which states on its website that it is a financial services company and not a bank, bought the "Buy now, pay later" Australian business "Afterpay" for $29 billion. Square is worth $113 billion, which is more than HSBC, the most valuable bank in Europe, which is worth $105 billion.
These rapidly emerging startups operate primarily through smartphone apps and websites, opposed to employing huge amounts of labour at high costs to run traditional brick and mortar banks. These companies that classify as fintech, as opposed to banks, have no branches and consolidate the entire financial ecosystem into a smartphone app or website thereby bringing huge cost savings and reaching new geographical areas without access to banking infrastructure. The savings in costs are transferred to the customers, allowing them to penetrate deeply into the customer base of traditional banks.
Adopting data-driven models requires analyzing around 2.5 quintillion bytes of data created each day at the current pace and making informed decisions based on the valuable insights derived from such vast amounts of data. This can be especially important when it comes to offering loans and investments.
Also, for the customer’s benefit, he can verify that transactions are not fraudulent, view recent transactions, make instant purchases and payments, and transfer money to family and friends with a few clicks, available 24/7.
Due to the phenomenal growth of these services, 38% of personal loans are now made by businesses classified as fintech, rather than banks.
Even in the year 2021, managing and handling virtual and traditional filing cabinets that are crammed with contracts, proofs of residences, proofs of identities, and other essential documents pertaining to background checks and customer due diligence is strangling productivity.
Throughout the history of AI, it has made significant advances in what is commonly known as computer vision, or its ability to recognise and perceive what it sees and make rational, data-driven decisions based on that. These capabilities are being applied to the management of documents with the use of optical character recognition (OCR) technology, which allows a document management system to automatically recognize, classify and process the documents without human interaction. Through OCR, paper documents can be easily scanned and converted into actionable digital information.
A document retrieval system that harnesses AI can efficiently and accurately extract information hidden in documents. A data extraction tool, for instance, is used to ingest data from forms, documents, and contracts and extract key-value pairs and entities
By reading more documents, AI becomes more adept at understanding how employees interact with documents and identifying and processing relevant information.
In an AI-enabled document management system, the bank's colossal library of documents can be more accurately categorized into different topics or hierarchies (especially helpful when topics or hierarchies are not yet known), identify the relationships between documents within a context broader than the individual documents themselves, derive inferences and hypotheses, and uncover underlying similarities among them. In the end, company documents can be more easily categorized, organised and searched when a deep dive is needed.
As opposed to merely replicating traditional document workflows in digital format, artificial intelligence has become a catalyst for irreversible changes in document management. The process has the inherent potential to improve, accelerate, and streamline every step in the workflow, from the processing and classification of the documents to their storage to the extraction of information from them. Already, it has reduced wasted time, improved collaboration, and shortened turnaround times on common workflows.
“How may I help you?” Pepper asks cheerfully as he beckons observers to choose from half a dozen options, including comprehensive and extensive tutorials on self-service channels like mobile banking and ATMs. There were mobs of tourists cooing and clapping at Pepper, a 4-foot tall humanoid robot with a high-pitched voice and a tablet screen strapped to its chest. This was the scenario when HSBC bank in collaboration with SoftBank Robotics deployed the humanoid robot Pepper in its flagship Manhattan branch and became the first bank in the US to use the robot. Buoyed by its unprecedented success, it has been rolled out in branches across the US.
A growing number of humanoid robots are enhancing customer experience across the banking industry. Because bank receptions are the primary interface between the consumer and the bank, this is an essential avenue for the implementation of digitization. Robots will undoubtedly change the face of the banking industry for the better. Think ATMs, virtual tellers, and robo advisors for wealth management, which will become a staple in the financial services industry, and now we are in the midst of an era of robots.
Robots offer unique advantages - they greatly reduce the time it takes to do repetitive and tedious tasks, they are extremely productive because they do not require breaks or rest, and in the event of a malfunction akin to a human being getting sick, they can be restored immediately. Additionally, you are not required to pay for overtime, insurance, or paid leave. Since it does not sleep, it can work 24/7. As an added feature, robots are now capable of charging themselves when they are at the end of their batteries by putting the plug in the socket automatically and ejecting it when fully charged.
At home, India's HDFC Bank unveiled the latest version of its interactive humanoid robot in 2018 at its branches in Bangalore, Kochi and Mumbai. IRA 2.0 was developed in collaboration with the robotics companies Invento Makerspaces and Senseforth Technologies in Bengaluru. An indoor humanoid robot built with GPS capabilities has a speech recognition module that can be trained to understand customer speech and answer 4000 questions using voice-based navigation. IRA 2.0 uses ultrasonic sensors to move inside the branch, as well as a face-detection algorithm to recognize customers. It has a 7.9-inch tablet display for visual interaction with customers.
City Union Bank has deployed Lakshmi, a two-foot robot who speaks English, can gesture, can interact with customers on more than 125 subjects, and can answer questions about loan interest rates, account balances, and more.
As we see in Hollywood movies, robots are taking over, not to eliminate humans, but to propel them forward, to increase the profitability of business exponentially, and to bring in more people to the party to help with more complex jobs that require their expertise.
New technology and digital tools have enabled neo-banks and digital-only banks to flourish, bringing a wave of commercial lending to new heights in the last few years.
When it comes to data and consumer insights, the financial services industry is probably the richest. Because of the endless applications for new accounts, loans, and insurance, along with comprehensive credit reports, credit scores, etc., the financial industry has access to a goldmine of data.
Without taking advantage of this resource, it can not only make banks obsolete and outmoded, but also cripple their revenues, knock off their profits, and put them at risk of extinction as more and more fintech firms saturate the commercial lending landscape, thereby taking their share of the pie.
The dynamic duo of Machine learning and AI can not only process data at rapid speeds but also do so with great accuracy and sans bias, and hence churn out more profitable loans, thereby preserving or restoring the position of traditional banks in the pecking order.
Let’s understand the two ways AI is used in making lending decisions: supervised and unsupervised
Supervised AI is found in large, data-driven organisations with their own lending rules in place based on market conditions and the bank's risk appetite. The AIs are fed these rules and then assigned the onerous task of weeding through the enormous amounts of loan applications based on these predetermined rules.
Alternatively, unsupervised AI systems are ideal for smaller banks and lending institutions with flexible sorting criteria and no rigid rules. Data scientists need to feed massive amounts of data into an unsupervised AI, then let the algorithm work on its own. A piece of AI then begins analyzing vast amounts of unstructured, disparate, dispersed data to uncover structures and patterns across variables, and gather insights on them.
AI is assisted in this onerous task by machine learning. It runs a variety of if-then scenarios using algorithms and statistical modules based on patterns identified by AI. A machine learning algorithm absorbs the input from the vast sea of data, runs it through its algorithms, and produces predictions or decisions as output.
Using artificial intelligence, one lender found historically women need to earn 30% more income than men to qualify for loans of the same size. In the past, lending practices have been distorted by bias against protected characteristics such as race, gender, and sexual orientation. When it comes to determining who gets credit and how credit is issued, these biases are evident in institutions' decisions. In most cases, human bias undermined the underwriting process, resulting in application rejections and higher/lower interest rates applied to certain loans, causing customer dissatisfaction and substantial losses to the exchequer.
The algorithms of AI and machine learning eliminate bias altogether by crunching data and churning it through its algorithms, uncovering patterns and based on those insights making decisions without any emotional involvement. The objectivity of Machine Learning compensates for its lack of warmth. In this way, lenders can create more profitable loans and reduce error rates.
Blockchain technology could revolutionize the financial ecosystem. Blockchain is basically a decentralised, distributed ledger that can store facts like who owns a particular piece of land or a bond. Representing the decentralised distributed ledger, blockchain enables distrusting, autonomous parties to carry out financial transactions without any intervention of intermediaries. As the technology is based on cryptographic algorithms, this technology can be used to keep an immutable record of ownership, and thereby ensure the security of sensitive data.
Cross-border transactions from payments to letters of credit generated a humongous $224 billion in payment revenue in 2019. Facilitating cross-border transactions is highly profitable for banks. In spite of the high profitability of cross-border transactions, banks are under no pressure to reduce their considerable transaction fees.
Cryptocurrencies like Bitcoin and Ethereum are built on public blockchains that anyone can use to send and receive money. The public blockchain technology renounces the necessity of third parties intervening in transactions and provides a free passage to fast, cheap and borderless payments.
The settlement and clearing of orders on the exchange involve multiple intermediaries and failure points. The process of transferring ownership is complicated because each party maintains its own version of the truth. By enabling the creation of a decentralised database of unique digital assets, blockchain technology can revolutionize financial markets. With a distributed ledger, it's possible to transfer the ownership of assets off-chain via cryptographic tokens. While Ethereum and Bitcoin have achieved this with only digital assets, new blockchain companies are trying to tokenize real-world assets, such as stocks and commercial real estate.
By reducing transaction times and facilitating settlement and adjustment, the technology improves efficiency. With the advantages of immutability, node distribution, and asymmetric encryption, blockchains could offer an edge in mobile banking for faster authentication, money transfers, and payments, as well as interbank payments and credit history checks. There are many applications of this technology, and it is certain that it will redefine mobile banking in the future.
The assessment and evaluation of a range of probable outcomes is an indispensable component of many financial services activities, from stock pricing to portfolio management. In order to achieve this, banks rely on algorithms and models that calculate statistical probabilities. A world where data is king means that ever-more-powerful computers are necessary for the accurate computation of probabilities. Many banks are turning to a new generation of processors imbued with quantum theory in order to crunch and grind data at superfast speeds.
Quantum computing technology can help banks analyze large or unstructured data sets more effectively and present them in a logical, structured and legible form. Through these domains, banks can gain more insight about their customers, which will enable them to make better decisions and improve customer service, for example by offering more timely or relevant offers such as a car loan based on browsing history. Quantum computers are especially promising in areas where algorithms are fueled by a large amount of real-time data such as equity quotes, which have a high level of random noise.
The use of Big Data allows firms to empower their employees to make better financial decisions on behalf of their clients. Nevertheless, finding relevant data points among a plethora of documents is a challenge since this data is disparate, fragmented, and unstructured, and resides in formats that are difficult to analyze and assimilate, thereby diminishing any value of such pertinent data points.
A sound financial decision is made possible and invigorated by the latest market segments and client data. Artificial intelligence tools like natural language processing enable employees to quickly discern insights from unstructured data by using AI to identify patterns within millions of documents. As a result, valuable insights that were previously overlooked can be uncovered.
For example, The AIEQ, the world's first AI-powered stock exchange-traded fund, was built using NLP by fintech company EquBot and the ETF Managers Group. Every day, AIEQ collects data on nearly 6000 US companies, including unstructured data whose formats are difficult for analysts to understand. This includes posts on blogs and social media posts.
On the buy-side of the capital markets, (responsible for purchasing and investing in securities for the purpose of earning a return on investment) organisations can integrate financial services AI to parse and scan for investment opportunities, analyse sell-side research reports (the sell-side of capital markets is involved in the issuance and sale of securities such as an IPO to generate funds for the client-company), process confidential client information, and generate actionable client insights. Then analysts can incorporate AI recommendations while drafting research for portfolio managers, suggesting investment strategies, and streamlining meetings and quarterly reviews. A company can also use AI to anticipate the needs and behaviours of customers, and use that insight to deliver automated, customer-centric service experiences.
As a sell-side analyst, with machine learning algorithms, you can examine the unstructured data of a potential investment, glean relevant information such as the founders' backgrounds, the amount of money raised, and existing deals, and synthesize that information to figure out what the potential return on investment is.
AI can also be used to assist employees in generating deal ideas by identifying potential private equity firms to work with or by analysing past transactions of private equity firms. Institutional firms in capital markets can use AI as a force multiplier to derive more from their data and deliver better experiences.
The use of AI and Big Data in companies can improve revenues in several ways through enhanced personalisation of services utilizing structured and unstructured data, cost optimization through automation, reduced error rates and optimal resource utilisation as the AI takes care of routine and tedious tasks, providing employees with the time and bandwidth to focus on problems that require more expertise and unlocking new and previously un-realised opportunities that had previously gone unnoticed.
Those banks who do not make AI central to their core strategies and operations - what we refer to as AI-first banks - run the risk of being overtaken by traditional banks and financial institutions, subdued by new entrants such as neo banks or digital-only banks, and abandoned by their customers. The shift to AI and big data is indispensable and non-negotiable for banks.