# Spider's Web: Money needs artificial intelligence to multiply

Author: Norbert Biedrzycki
Published: 2020-01-29
Language: en
URL: https://norbertbiedrzycki.pl/en/spiders-web-money-needs-artificial-intelligence-to-multiply/
Summary: Money needs artificial intelligence to multiply. Apart from money, a bank's most valuable currency is knowledge about its customers.

My article in Spider's Web, 24 January 2020: Money needs artificial intelligence to multiply.

Marketing, customer service and sales used to rely on separate data-gathering technologies. The digital revolution has made it possible to consolidate these distant areas and bring them closer together. Today, assessing a customer, studying their preferences and potential, preparing a product and selling it all happen in a shared digital space, using tools that are integrated and compatible with each other.

Data, the "food" of artificial intelligence, flows in from many directions, but it is all digital in nature. Importantly, most processes happen in real time. For a long time, the main variables for building good customer relationships were age, income, job position, marital status and previous history with the financial institution. They are still in use and banks take them into account. But they are only part of the analytical puzzle.

Today what counts are insights into the customer's current activity (online activity, of course). Companies know what customers want and who they really are by analysing their behaviour on the bank's website, during calls to the helpline and in contact by email or phone. This information grows in value because it can be processed in practically unlimited ways and, most importantly, the processing happens in real time.

## **The voice assistant asks about a loan**

One of the most radical changes we can expect in the coming years concerns customer service. Voice bots, which we already know from our private lives, will make a career in customer service. Some are even beginning to talk about the coming "Alexa-isation" of our lives – a reference, of course, to the growing popularity of devices such as Alexa and Siri.

According to forecasts by RBC Capital, by 2020 nearly 130 million devices directly connected to Alexa will be in use worldwide.

Nothing stands in the way of software based on the same model being used to serve bank customers – all the more so because bots trained by specialists are getting better and better at human-sounding speech. Bank of America has introduced the chatbot Erica, which advises the bank's customers by voice and text message. The bot works 24 hours a day and lets customers carry out all the transactions they use every day.

It is easy to guess that the cost of employing a dozen or several dozen specialists working around the clock would be considerable. In Silicon Valley they say that people do not scale like machine algorithms and the chatbots built on them.

## **When a bot sounds like a person**

In line with the self-improving nature of machine learning systems, the competence of such devices will grow. A bot you can talk to will not just answer questions about the weather or traffic. It will talk about loan interest rates and the benefits of a deposit. It will tell us how much we still have to repay and remind us about an overdue instalment.

Today, such devices work well in the first contact with a customer: they sort helpline calls and route customers to the right departments. In the future, assistants will be delegated more complex duties. For that to happen, bots' communication skills must of course develop further. Building complex sentences, picking up the customer's intonation, sensing their deeper problem – all this is still in the future. But bots are learning faster and faster.

## **Money for a fingerprint**

Banking, as an industry that collects and processes huge sets of information, faces a serious challenge – all the more serious because today's users are hypersensitive about security. Media reports of recurring leaks of sensitive data collected by companies and websites reinforce this distrust. IT specialists in banking and insurance therefore have their hands full and keep searching for solutions that could rebuild the average consumer's recently shaken trust.

In this context, the trend towards changing the way we access online accounts becomes interesting. In short, it is about logging in and all authorisation procedures. Traditional solutions are nothing more than ordinary logins using a string of characters – letters and digits. But as hacking techniques improve, this is becoming a critical weak point in IT systems.

Security codes and passwords fall prey to bots designed to steal this kind of data. Hence the need for new solutions. There are many signs that the technique with a chance to revolutionise authorisation will be the fingerprint, and soon facial recognition. Such a unique, individual personal sign seems so far to be the best and simplest protection against break-ins. Just as voice control of devices may push traditional search engines and touch interfaces out of the market, a manually typed string of characters will become an anachronism compared with a mechanism based on the touch of a human hand.

## **Send us a selfie and get a loan**

And what about authorisation based on a photo of your face? Is that possible? It is starting to be. After all, individual features captured in a photograph are also unique and impossible to forge. That uniqueness can be decoded through reliable analysis using software capable of identifying the finest nuances in an image.

Scanning a photo, or scanning a face through the camera in our laptop, therefore has a good chance of becoming another verification procedure. If artificial intelligence can recognise faces – a skill increasingly valued by the police, for example – this can also be used in completely different situations.

We can imagine that in the future a loan will be granted on the basis of signals "written" in a photo that we upload to a designated place. Analysed by algorithms, the photo will send the bank information about our health and general life situation, and assess our worth.

## **Investigation and monitoring in real time**

On the subject of security, it is also worth mentioning the use of artificial intelligence tools to monitor the security of banking systems. Intelligent software working in real time is designed to track the smallest anomalies that indicate hacking attacks.

Around the world, hundreds of thousands of attempts are made every day to break in and steal valuable data or funds. Old-style software would not be able to detect so many attacks. Only artificial intelligence tools provide adequate protection and a sense of security.

It is not only about attacks. Banking struggles with the constant, natural problem of approving ambiguous transactions that carry risk and are open to different interpretations. This can concern different areas: loans, loan repayments, calculating interest, and a whole mass of accounting operations.

Analysing specific cases and approving operations the traditional way takes many hours of work by people and traditional systems. With machine learning, authorising such transactions takes only seconds. The benefits for the efficiency and security of the systems seem clear.

## Artificial intelligence in the insurance market

The solutions discussed here are, or will be, used in the insurance market too. Here, too, artificial intelligence will find large resources of data on customer behaviour and needs. The category that is emphasised even more here than in banking is risk. Algorithmic analysis covers every possible situation involving a threat to the customer's life or health, as well as their material status (property).

Here, too, the photo analysis technique I mentioned above can be used. Selling life insurance will require extensive material for analysis. From this point of view, efforts to obtain data in less conventional ways seem interesting.

The customer's car, for example, is becoming a goldmine of knowledge. Information such as the vehicle's mileage, the driver's tendency to cause accidents and even driving technique can be valuable material for any company preparing a personalised insurance product.

Equally important is the use of complex algorithmic analysis to estimate the likelihood of illness associated with a particular lifestyle. Whether insurance companies will in future be interested in tools for analysing DNA remains to be seen, but it cannot be ruled out.

## **Tailored services and automated purchasing**

Artificial intelligence tools enable one extremely important thing: precise identification of an individual person. Thanks to this ability, the customer stops being an element of a larger set and becomes an entirely individual case.

That is why the overriding innovation worth mentioning here is customer-centred personalisation of all services and products. This approach leads to one more solution. In the near future, customers will use interfaces that can analyse data they enter themselves. This is another sign of the automation of banking services – automation that is an obvious consequence of using the intelligence of algorithms.

## **Shopping for the average customer and for traders**

Following this line further, it is worth adding that automation will increasingly accompany the purchase of specific products. While today the salesperson is still an essential part of the customer relationship, in the future customers will be able to rely on apps.

Their task could be, for example, buying financial and insurance products in series and relieving customers of decisions that take more time. It is worth stressing that such serial purchasing will be a blessing for business investors, whose daily life involves large sets of products and information. Professional traders can expect artificial intelligence to take a whole mass of duties off their shoulders, related both to purchasing and to a huge arsenal of analyses.

## **The power of algorithms and the mystery of the black box**

Artificial intelligence in banking and insurance has a guaranteed future. Algorithms will simplify many operations, making things more convenient for customers and saving them time. The sense of security customers gain along with automation can be another added value in the effort to make their wallets fuller.

A side effect of these phenomena, however, will undoubtedly be issues of employment in the financial sector. One illustration is JP Morgan, which launched an application (platform) for extracting data from documents related to lending operations. Analysing 12,000 loan documents would have taken the bank's employees 360,000 hours of work.

Machine-learning applications worked on the task for... a few hours. The benefits are clear, but the interpretation of this event from the point of view of employment policy can also be interesting.

Another important question is whether these algorithmic operations will at some point require explanation – a deeper account, both for customers' sense of security and for bank management's confidence that they know what is "going on" on board. Here we touch on the famous and much-discussed "black box" problem: everything that happens inside AI-driven devices that in many cases escapes human understanding.

The revolution in the financial market brings into focus, as through a lens, all the most important questions about the role of artificial intelligence in our lives, both private and business. The question of its possibilities and its impact on the convenience of customers of financial products has essentially one possible answer: they are enormous and incomparable to anything before.

The financial industry is one of the biggest beneficiaries of algorithms beginning to dominate many areas of business. That does not change the fact that it is precisely people in this sector who realise how many questions about protecting funds, system security, processing customer data and regulation remain unanswered. Money needs artificial intelligence to multiply. At the same time, it needs rules and regulations so that it does not melt away in technological chaos.

Link to the [article (in Polish)](https://www.spidersweb.pl/2020/01/sztuczna-inteligencja-bankowosc-ubezpieczenia.html)

Related articles

– [When will we stop being biological humans?](https://norbertbiedrzycki.pl/en/when-will-we-cease-be-biological-people/)

– [Artificial intelligence is the new electricity](https://norbertbiedrzycki.pl/en/artificial-intelligence-is-new-electricity/)

– [Robots are waiting for judges](https://norbertbiedrzycki.pl/en/robots-awaiting-judges/)

[– Only God can count as fast: the world of quantum computers](https://norbertbiedrzycki.pl/en/only-god-can-count-that-fast-the-world-of-quantum-computing/)

– [Machine learning: computers are no longer infants](https://norbertbiedrzycki.pl/en/machine-learning-computers-coming-of-age/)
