Savvy, forward-thinking business owners and operators of all kinds understand the importance of keeping up with the latest trends in automation, how such trends will impact payments, and what obstacles they will face as they go with the implementation flow. One prime example is a move to artificial intelligence (AI) and machine learning (ML) implementation There is no better time than now—the beginning of a new year—to explore this trend and the roadblocks along the path to deployment.
AI Becomes a Reality
End-users are beginning to introduce and use AI and ML in several ways. For example, merchants in all markets—not just retailers—continue to look for means of creating a frictionless consumer payment experience. Some are beginning to introduce AI-enabled chatbots or other AI tools to simplify and guide consumers through the payments process.
In another vein, businesses are striving to maximize payment security through AI, with powerful algorithms. These algorithms form the base of complex fraud detection systems and provide early warning signals of cyber-attacks.
Just as significantly, AI and ML are gaining ground in the area of recurring payments, which can fail for a number of reasons, including expired account credentials, insufficient account balances, unavailable network connections, and more. AI and ML are being harnessed to minimize these factors, for example, by checking systems for more current information and re-trying transactions, pinpointing upcoming expirations, and the like.
Similarly, AI is now being recognized as a means of better serving and reducing payment friction for consumers whose payment information merchants keep on file for frequent, but non-recurring purchases. With AI in place, overly aggressive fraud filters that might otherwise be triggered by multiple purchases and lead to failed payments can be “stopped in their tracks” if algorithms determine that overriding them is the proper course of action. This, of course, improves transaction approval rates—a boon to business’ bottom line.
Not surprisingly statistics show that the benefits of AI and ML are becoming well-recognized. For example, 65 percent of finance-oriented executives surveyed by Brightline Technologies said they “believe in” the shift to AI in payments and banking. Forty-one percent of respondents to the survey cited greater efficiency as the biggest benefit for AI in payments and banking, and 20 percent said they perceive AI as decreasing payments fraud.
Artificial Intelligence, Real Challenges
But AI implementation is not without its challenges. The principle of “garbage in, garbage out” applies here—in other words, AI algorithms are only as functional and powerful as the data they are fed. Without accurate, complete, unbiased data, algorithms cannot “learn” to fully identify patterns or make the best predictions, minimizing the benefits merchants achieve through AI-powered technology.
What’s more, AI algorithms can only solve specific problems and cannot deviate from their intended purpose. For example, an algorithm that’s been designed and trained to detect fraudulent or otherwise suspicious payments would not be able to pinpoint an imminent cyber-attack. Consequently, merchants that undertake wide-ranging AI implementation—or implementation of AI-powered tools in multiple areas—must be prepared for a significant effort.
Integration, too, is a challenge. Consider this: Nearly half (45 percent) of respondents to the Bottomline Technologies survey indicated that despite their positive impression of AI and AI-powered technology tools, integration with existing technologies is the biggest impediment to large-scale adoption. Indeed, the success of AI implementation has much to do with how well the solution(s) integrate with businesses’ existing infrastructure—not only in terms of technology but also in terms of people and business processes. Resources—financial, training, and otherwise—must be dedicated to integration, or AI projects will fail.