The speeds at which today’s AI technology is advancing is far beyond what many have expected. The role of human decision-making is being eclipsed by computer and machine learning in almost every industry. We see it in autonomous cars and trucks as well as in actual robots pre-programmed to take on tasks once thought to be reserved for humans alone. Ralph Haupter, President of Microsoft Asia and Corporate Vice President, Microsoft stated the following: “I believe 2018 is the year that this will start to become mainstream, to begin to impact many aspects of our lives in a truly ubiquitous and meaningful way” There is no uncertainty in the fact that AI is progressing at the speed of light. It is yet to be seen how this will impact industry on a global industrial level but the applications to payments have not been insignificant in the past several years. This paper seeks to explore the intricacies of this technology, its evolution, and the potential impact of AI in payments in 2018 and beyond.
Artificial intelligence has been seeping into the mainstream for years. It seems that 2018 will usher in a strong driving force of AI technology. We look at some of the more common uses and applications below.
Personal assistants that leverage AI have been around for some time. Many are familiar with Apple’s Siri, a voice-activated virtual assistant connected to pple Inc.'s iOS, watchOS, macOS, and tvOS operating systems. Siri is the result of work originally done at the SRI International Artificial Intelligence Center. Built on a speech recognition engine provided by Nuance Communications, Siri employs advanced machine learning technologies to help people complete simple tasks. Siri was just the beginning of what would become the virtual assistant hype. Today, many Americans have voice-activated personal assistants in their home. A report by Juniper Research projects that voice-enabled smart devices will be in more than half (55%) of U.S. households by 2022.
This is just the beginning. While the prevalence of these voice-activated assistants is increasing, they are also getting smarter. Backed by machine learning, these devices will soon have the ability to learn intimate details about their “owners”, mapping out daily routines, tracking stocked grocery items, and streamlining everyday tasks that humans, thus far, have had to manage autonomously. Artificial
The Internet of Thing (IoT) is proliferating daily. Experts predict that spending on IoT applications could generate $64.1B by 2020. Currently, the top uses of IoT include predictive maintenance, self-optimizing production, and automated inventory management—which should continue to drive market growth in the next two years.
AI is a natural intersection for the IoT as the number of connected devices grows in the billions with a steep upward trajectory. The ability to manage these connected devices—and the data they collect—requires technology like AI. The convergence of AI and IoT means increased efficiencies in industry as well as everyday life.
As more and more digital business models and “as-a-service” offerings emerge, IoT and AI must work together to process and utilize large amounts of data. The result is the ability of businesses to offer mass personalization at scale, driven by massive computing power for execution. The teamwork between these two technologies will continue to drive investment in cloud and big data technologies.
Biometrics have become increasingly incorporated into everyday use cases over the past few years. For example, Apple’s iPhone X utilizes facial recognition technology as a means for users to unlock their phone. It’s one example of physical biometrics being used for security purposes. There are a variety of physical biometrics methods currently in use, including:
Physical biometric solutions leverage AI technology, which uses coded versions of measurable biometric information (as in the examples above) to ID and verify that the person submitting biometric information matches the biometric data points on file.
Because our biometric information is unique, it makes it increasingly difficult to spoof or scam the system. This biometric information can be collected and stored within databases and used for comparison by asking users who are trying to authenticate to offer a sample (aka an image of their face or a thumbprint on a home button). Analyses
Big Data has become a necessity for big business. In order for businesses to harness this data— and make good use of it—they need the power of AI. A key finding of the 2018 NewVantage Partners’ annual executive survey, is that a staggering 97.2% of executives report that their companies are investing in building or launching big data and AI initiatives. A majority of respondents from that survey (76.5%) also agreed that the magnitude of data is driving AI and cognitive learning-related initiatives within their organizations.
The access corporations now have to meaningful data—and the sheer magnitude of that data—is ripe for pattern detection and better understanding of behaviors via AI algorithms. Big
Analyses of subsets of data can be replaced with real-time analysis of big data thanks to the computing power and AI. This enables the ability for real-time outcomes, driving new, more personalized offers for consumers.
The applications of AI described above are just a tip of the iceberg. These general activations of AI have very granular applications as they relate to payments. As AI permeates various industries via the mechanisms described above, it is also manifesting in the payments space, to facilitate personalization, enhanced security and fraud prevention, and to streamline resource-heavy tasks.
The fact is that 2,000,000,000 people use messaging apps. It’s been several years since they surpassed the popularity of social networking apps (the global monthly active users for the top 4 messaging apps surpassed the global monthly active users for the top 4 social networking apps in 2015), with no signs of slowing. Gen Z makes up a large portion of this active user base, and they are spending increasing amounts of times in these messaging apps. One report says over half of Gen-Z’ers (52%) spend three or more hours every day on messaging apps.
This group, slated to become the biggest generation of consumers within the next two years, accounts for $29 to $143 billion in direct spending. It’s a wake-up call to merchants to get with the program—and to get where their largest group of consumers is.
With Gen Z, a generation that grew up with on-demand services like Uber and Postmates, the expectations of shopping experience are elevated. They are not interested in the “old school” shopping experiences of previous generations. Even webrooming and showrooming are proving too laborious for this generation. The ability to interact with chatbots and voicebots, however, caters to this group’s on-demand needs while facilitating commerce in a secure way.
By using AI-infused technology, merchants can engage and interact with consumers in a more natural, conversational manner. This requires thorough journey mapping, to understand the variations and turns that natural communication can take. It also demands user experience design (UX) and user interface design (UI) for a seamless, natural experience for consumers.
Beyond experience, merchants need solid AI and bot technology to serve as the foundation upon which conversational commerce is built. Merchants will need a way to integrate with the messaging/chat platform of their choice (Facebook, Alexa, etc.) and additional layers (NLP, controller, business and fulfillment) to ensure the automation of essential services— like inquiries and transactions—are natural and seamless.
Frictionless payments is the buzzword of 2018, and with good reason. In 2017, global retail ecommerce sales spike almost a quarter (24.8%), reaching $2.304 trillion. Ecommerce represented 10.2% of total retail sales worldwide last year. Global mcommerce sales are also on the rise, increasing 40.3% in 2017 to $1.357 trillion. It’s a mark that consumers are growing increasingly comfortable making purchases remotely, if the purchase experience is optimized. Unfortunately, cart abandonment rates paint the picture that many ecommerce purchase experiences are not optimized. In Q4 of 2017, 77% of online retail orders resulted in abandonment.
Fortunately, AI has the power and potential to reverse the abandonment trend and to help merchants better cater to their customers making remote purchases. One application is Instacart, which uses machine learning to learn and map out grocery store layouts, enabling its delivery workers with the most efficient routes to shop for customers. This streamlines workflow, producing faster deliveries for the end users and promoting repeat purchases.
Amazon is another great example of AI applications to remove friction from the customer experience. Amazon Go customers have the ability to avoid arguably the biggest friction— and pain—point of all: the checkout line. By using computer vision, sensors, and deep learning technologies, Amazon has recreated shopping by allowing customers to select their items and leave the store. The technology detects which items have been removed from shelves and communicates with the user’s Amazon account to complete the purchase and issue an electronic receipt.
The use of AI for fraud prevention is not new. Card brand MasterCard and companies like RBS WorldPay have been relying on AI to analyse transaction data and detect fraudulent patterns to prevent card fraud for several years. The ability to trace a payment card’s journey of use, including device and endpoint access, can help companies identify suspicious activity, points of compromise, and fraud.
AI and machine learning engines are able to collect, process and analyze large amounts of data, identifying and learning the patterns of fraudsters. These technologies are also able to pick up on traces left behind by fraudsters, which may have otherwise gone unnoticed to the human eye. This synthesized data can then be used to create clear predictors of fraud threats.
Machine learning technology can be applied this way to decrease false positives and to increase the accuracy of fraud detection and screening. Companies that process millions of transactions per year are hard-pressed to achieve 100% accuracy with simple rules-based fraud detection and human intervention. Computing power paired with AI enables these companies to quickly process transaction data, detecting anomalies which can then be handed off to human personnel for further investigation. This cuts down tremendously on false declines as well as internal costs associated with manual review.
Compliance and regulation have long been a thorn in the side of merchants, financial institutions and fintechs worldwide. According to estimates by Citigroup, the largest banks have doubled staff in charge of handling compliance and regulation. This increase has cost the banking industry upwards of $270 billion annually, totaling roughly 10 percent of operating costs.
The cost and resource burden is clear. The introduction of AI into compliance and regulationrelated tasks could significantly counter these burdens. By replacing or augmenting tasks typically reserved for human decisionmaking with AI automation, banks are finding that they can more efficiently and cost-effectively manage compliance and regulations.
Much of the work that could be designated to AI automation includes data processing tasks, which are currently prone to human error and slow processing rates. Such repetitive tasks can be streamlined via AI pattern recognition, which enables the technology to get more intelligent over time, further reducing the burden. Humans can then focus on analysis of processed data along with building more accurate and sound strategies around this analysis.
Opus developed an algorithmic and statistical scoring engine, leveraging our payments domain expertise.This engine creates a machine learning vector-based model for feature identification, modelling, and classification.
Our machine learning platform can be used to automate decisions processes for multiple payments business processes, including payments fraud detection. The machine learning engine includes multiple algorithms that have been customized for payments use cases, available for both in-premise and on-cloud deployment. Despite the technological complexity of the model, our visual-first, intuitive design makes it easy to integrate with a business’ core systems, databases, and websites. It is aligned to user-assisted learning and time series analysis to aid in fast and efficient payments fraud prevention. Compliance
Opus has the ability to develop custom bots for the enterprise, leveraging the technology of Ailene, our specialized bot technology for payments.
As illustrated earlier in this white paper, bot technology is becoming integral to successful ecommerce and mcommerce merchants globally. The ability to interact and engage with customers on their preferred channels is not negotiable for merchants that want to compete in a crowded marketplace.
Our bot technology uses Amora, our machine learning and AI engine, to create both chatand voice-bots that can automate commerce and customer service channels online. The intelligent understanding of Amora allows our bots to engage naturally in a human-like manner with your customers in real-time. Ailene is able to pick up on customer mood and other conversational nuances to reply appropriately, as if she were human.
Ailene integrates easily with all core enterprise systems and data as well as customer engagement channels (Google Assistant, Alexa, Facebook, Slack, etc.) to streamline and automate the customer experience.
Amora’s advanced machine learning models allow our bots to analyze conversation data and make intelligent business decisions in real time. The result is the ability to automate both inquiries and transactions, reducing the burden on internal human teams.
Using an NLP hub powered by Amazon Alexa and Google Dialogflow, Ailene can interact with customers seamlessly, capturing intent while engaging in everyday conversation. She can also understand and translate financial vocabulary, allowing for a more automated customer service and transaction workflow. Additionally, Ailene can seamlessly transition customers to your human workforce in situations where escalation is necessary.
The evolution of payments is not slowing or stopping. AI and machine learning will continue to impact everyday transactions as well as the payments of the future. Payments are trending towards embedded and frictionless experiences, so merchants need to effectively integrate emerging technologies into their operations and workflows to ensure they can provide an optimized experiences that meets—and exceeds—increasingly sophisticated consumer preferences.