Posted Tuesday, October 31, 2017 by Eran Feinstein
Artificial Intelligence, or AI, is an umbrella term comprised of numerous computer systems which enable machines to think, learn, and react as a human would. These systems include, Machine Learning, such as Deep Learning and Predictive Analysis, Speech to Text, Natural Language Processing (NLP), Machine Vision, Image Recognition, and Expert Systems, to name just a few of the leading technologies.
Advanced algorithms receive a host of data, from speech, text, images, and behavior, and make intelligent decisions based on previous interactions and behavioral patterns. Purchase predictions, suggested TV content, Alexa, biometric border exits and customer service chatbots all utilize some of the above technologies.
AI technologies, such as chatbots and text recognition systems are already enabling P2P money transfers and streamlined payments. The PayPal bot integrates with social apps such as Facebook Messenger and Slack to allow users to type a payment request within the app for transferring money between contacts. Similarly, companies are integrating payment bots within Messenger to streamline payments, such as Hipmunk, which enables customers to book and pay for a hotel directly from within the messenger app.
Amazon Go, the brick-and-mortar Amazon alternative, offers a checkout-free shopping experience, currently available to its employees only (beta trial). Shoppers log into the Amazon Go app upon entering the store. Store-implemented sensors and computer vision technologies calculate which products are in the customer’s shopping basket, and the customer’s Amazon account is automatically charged upon leaving the store. No need to stand in line, wait for each product to be scanned, or even swipe/tap a payment method.
The above examples of AI implementation in payments are just the tip of the iceberg, as AI has the potential to revolutionize the customer shopping experience and payment security.
Below are a few more ways AI is changing mobile payments:
1. Grab and Go Applications
Focused on saving customers time, Grab and Go applications will utilize a combination of AI technologies to improve the customer experience. Mobile customers will be able to pre-order products and pay for them via their mobile wallets, with these products ready for them upon arrival.
Imagine purchasing a new pair of jeans at your favorite shop. Usually, you’ll have to go to the store, look for the jeans you want, and stand in line to checkout. But NLP, speech recognition, and mobile payment apps will converge to allow you to simply place an order via text and have the jeans waiting for you in the store. You can then grab them and go, checking out using your mobile phone, saving you precious time. It will be even easier than ordering them online (as long as you are out and about town, anyway).
Some ATM’s already enable pre-ordering cash. As opposed to inserting the card in the ATM, typing in the code, and choosing an amount, people can get QR codes generated for a specific transaction in advance. These QR codes allow users to simply scan the code at the ATM to receive the amount they ordered, without all the hassle.
NFC powered eATM’s enable card-free cash withdrawals. Integration of NLP, speech recognition, and biometrics will take this a step further. Users will be able to pre-order cash, as above, and cash out via a simple phone tap, or even biometric authentication process, such as face recognition or eye scanner.
2. Machine Learning for Fraud Prevention in Payments
Traditional fraud detection programs are rule based. These systems require a significant amount of transactions to be manually reviewed and authorized. In fact, a 2016 survey from CyberSource, a leader in the industry, found that 25% of all transactions require human intervention. With ecommerce consistently on the rise, and as companies are expected to offer same-day service and shipping, there is a need to improve the capabilities of fraud detection systems and lower the volume of manual authorizations.
Machine learning-based fraud detection systems analyze purchasing patterns from large datasets. Learning algorithms evolve as fraudsters change their tactics, further improving coverage rates. These advanced systems, such as the one under development by Fraugster, will be able to lower the percentage of manual authorizations required and improve efficiencies.
3. Machine Learning will decrease false declines
False transaction declines, in which a non-fraudulent transaction is flagged as a security concern, currently cost companies more than actual fraud does. According to BI Intelligence, fraud losses were projected to reach $6.5 billion in 2016, while falsely declined transactions were projected to cost retailers $8.6 billion.
As Machine Learning algorithms evolve, the predictive capabilities of security systems will improve, lowering the number of false transaction declines and increasing retailer profitability from higher approved transaction volume.
So What’s Coming for AI Payments Overall?
- Artificial Intelligence will be increasingly integrated in all aspects of our lives, including in how we make purchases and execute payments.
- Natural Language Processing will enable text-based in-app purchases even from inside brick-and-mortar shops, eliminating the need to order and stand in line to pay.
- Authentication via image recognition and biometrics, combined with app-based cash ordering, will shave minutes off of ATM withdrawals, while decreasing ATM fraud.
- Retailers and customers alike will benefit from Machine Learning-based fraud prevention systems which will achieve higher accuracies in fraud identification, lowering percentage of transactions transferred to manual authorization, as well as the number of falsely declined purchases.
- As with Amazon Go, many of these technologies are currently being tested in limited trials, which means the future may be nearer than we could have ever imagined.