by Markus Eichinger
January 17, 2018

Artificial Intelligence (AI)? Some may think of Hollywood movie-inspired dystopian future worlds, but what was considered a technical utopia a few years ago, has now become reality – although it comes in a subtler form.

In fact, AI has already arrived in our everyday world: Just think about your e-mail spam filter, natural voice recognition, online shopping recommendations – or simply about the predictive keyboard on your smartphone.

In this second part of my series on payment megatrends (see an overview of all articles at the bottom of this page) which will strongly influence the industry in 2018 and beyond, I will focus on Artificial Intelligence and why it will be a huge game changer for payment.

Definition: What is Artificial Intelligence?

Artificial intelligence is an umbrella term used for a set of hardware and software technologies, some have already been around for many years, some are brand-new. In a nutshell, artificial intelligence describes the ability of a technical system to replicate “human” behaviors such as learning and problem solving.

There are two subcategories:

  • Strong Artificial Intelligence in this respect aims at creating a machine whose thinking skills cannot be distinguished from those of a human being – this basically means, creating a machine that can pass the Turing Test. No technological system has yet passed this test, and there are various predictions on when this will be achieved. Back in the 1950s, its famous inventor Alan Turing predicted the year 2000, whereas newer predictions range between the year 2020 and 2029.
  • Weak Artificial Intelligence, however, is a category for applications that focus on solving real-world problems with machines rather than with humans (which is why this kind of AI is more relevant for business applications).

Here’s an example:
Megatrend AI - this is easy to read for a child, but hard for a machine

You’d surely agree that both pictures illustrate the hand-written number “4”. Every child can easily solve this task, but when it comes to “explaining” this to a machine, things get complicated. Algorithm-oriented ways of describing the “4” might work for a small sample size, but when it comes to handwriting recognition, this approach no longer works.

Instead, Deep Learning (or Machine Learning) technologies come into place, and they are evolving with ever increasing speed. They could be described as an AI function that imitates the way the human brain works in processing data and by doing so, creates patterns from training data that are able to predict behavior for new data.

Neural networks are like human brains: Experience makes them wiser

Deep learning technologies are commonly based on neural networks that are trained with existing sets of data and can be seen as the brains of modern AI. A neural network consists of a multitude of neurons, i.e. collections of connected units or nodes – analogous to the biological neurons in a human brain.

These artificial neurons are designed to perform simple decisions, based on input data. The power of the network then lies in the combination of multiple layers of neurons and their weighting.

Explaining how neural networks work in detail would surely exceed the possibilities of a blog article. Instead, I’d like to recommend this free online book for a good and easy-to-read introduction or the following video for a quick overview.

This short explanatory film has been created for an AI course at Texas A&M University

However, the quality of a neural network is not that much driven by algorithmic excellence, but rather by the amount and quality of available training data. At Wirecard, we use datasets ranging back up to 10 years to train our AI-based risk systems.

Reduce the risk: the use of Deep Learning technology in payments

Risk decisions in payments are never 100% clear, therefore they are the perfect playing field for Deep Learning technologies.

How are risk decisions made? As a merchant or acquirer, you do not want to accept a payment where there is a high probability for a fraudulent action or a high likelihood for a chargeback. On the other hand, rejecting a “good” payment normally results in conversion loss and thus, lost revenue for the merchant.

To decide on acceptance or decline of the payment, many different factors are available to be taken into account, for example:

  • Time & date – e.g., is a credit card used at normal shopping times?
  • Merchant category codes – is it used at a supermarket or for online games?
  • IP-addresses – where is the device used for online shopping located?

Social information – the use of social media data is quite a new application in payment risk detection

Neural net learning consists of multiple hidden layers and mimics the behavior of the human brain. Deep learning included multiple neural networks put one after the other.

Neural net learning consists of multiple hidden layers and mimics the behavior of the human brain. Deep learning included multiple neural networks put one after the other (Source: Sunil Kappal, Deep Dive Into Machine Learning, published at DZone)

By training a neural network with historic risk data, the network will learn and improve and therefore be able to better predict risk for future payments. However, it is important to point out that the most important driver for deep learning quality is the grade and amount of training data – not the actual algorithm!

Artificial Intelligence with Deep Learning will become a key differentiator for payment companies in the future #AI #Payment

Pay with your voice or your face: Upcoming applications of Deep Learning and Artificial Intelligence

The payment industry has a long history of working with data to make qualified risk decisions. For sure, risk analysis is the “bread and butter” application of Deep Learning technology within a payment company.

In the upcoming years we expect many new applications, driven by AI technology that will significantly influence the industry, both in payment and commerce in general.

Just to list a few:

  • Chatbots and AI-assisted shopping

Chatbots will enter the field of commerce and help consumers to find and research products more easily. Widely used applications such as WeChat in China will provide the reach for merchants and are already offering a platform for shops and payments.

It will be AI-powered chatbots that will resolve the disadvantages of the smartphone’s limited display size and browsing capability and thus open up all-new sales channels for merchants that require flexible and well-integrated payment solutions.

  • Natural voice processing

2017 was the year of natural voice processing services such as Amazon Alexa or Google Assistant. They not only found their way into home appliances such as Amazon Echo, the Sonos speaker system and even cars, but they also significantly improved in quality and intelligence.

While creating the whole shopping experience within the “voice channel” is still a bit of a cumbersome process, putting items into a shopping basket via voice certainly is not and works just fine. This will pose new challenges on designing an optimal checkout process and also on how to integrate payment in the right way.

“Alexa, could you put milk in my shopping basket?” In 2018, it has become perfectly normal to buy something online by just saying what you need – thanks to Artificial Intelligence (Source: Wikipedia)

“Alexa, could you put milk in my shopping basket?” In 2018, it has become perfectly normal to buy something online by just saying what you need – thanks to Artificial Intelligence (Source: Wikipedia)

  • Smart Data for Offers & Advertising

The way of how we work with data to create offers and advertising for customers used to be largely based on market studies, focus groups, samples – saying: on metrics that are not tailored to the single individual. We can see that AI technology makes it possible to target offers better to consumers with the sample size one, and thus open up new possibilities for merchants to reach their customers.

Payment services and especially mobile payment services will play a key role here, as they already exist and are readily available channels to process data, mostly in real-time.

  • Facial Recognition Technologies and Biometric Identification

Most biometric identification technologies are based on AI-supported technology for decision making. In 2017, Apple launched a sophisticated facial recognition technology in the new iPhone X and we expect more applications of this “contactless identification” to evolve within the near future.

This will revolutionize checkout-processes completely and make payments online and in-store more secure. We also expect that sophisticated identification technologies will make physical cards less predominant and drive the demand of virtualized payment instruments.

One example of biometric face recognition for payment purposes – this is a still image taken from our video about Wirecard’s “Connected Store”.

One example of biometric face recognition for payment purposes – this is a still image taken from our video about Wirecard’s “Connected Store”.

Conclusion: AI’s implications for the payments industry

The battle for the “best AI” is already being fought in full swing within the “Big Four” of the tech industry: Apple, Facebook, Amazon and Google. But also for the payment industry, AI will open up completely new opportunities to develop products that will improve the top line and bottom line – i.e. revenues and profits – of a retailer’s P&L.

I think it is crucial for payment companies to think far beyond payment data and to develop in-house capabilities and competences in the field of Artificial Intelligence and, more specific, in Deep Learning technologies.

AI-based technologies will change the online and physical conversion process significantly and lead to higher conversion rates and improved security – thus, we expect that AI will impact the whole industry significantly in the years to come. In my opinion, AI will at some point become the key differentiator for technology and payment companies. Given the explanation above of how AI works, it’s never too early for payment companies to “start learning”.


These are the other articles of the Payment Megatrends series: