News and resources on payments systems, innovations and initiatives worldwide.
AI will drive productivity in the payments industry

AI will drive productivity in the payments industry

The payments industry is increasingly being exposed to the technological transformation that is occurring outside of the financial services sector. But is the space ready to drive productivity in payments with artificial intelligence?

The second day of AFP 2019 in Boston started with a discussion on how treasury practitioners must use every tool available to ensure that payments practices are simple, smart and secure.

Jessica Cheney, CTP, VP of strategic solutions, Bottomline Technologies; Shannon Rosbrook, senior vice president, Citizens Commercial Banking and Evan Karatzas, founder & CEO, Proximity Lab kicked off the conversation.

AI definition

Cheney highlights that the three panelists are far from being AI experts, but goes on to explain the importance of having a conversation about the technology’s role in the financial services industry.

“AI means something different to everyone and we are using it a lot more than we realise. There are as many definitions of the term ‘AI’ as there are terms associated with it,” Cheney adds but goes on state that Deloitte’s definition is perhaps the most accurate.

Deloitte defines AI as a “suite of technologies, enabled by adaptive predictive power and exhibiting some degree of autonomous learning, that dramatically advance our ability to: recognise patterns, anticipate future events, create good rules, make good decisions and communicate with other people.”

Further to this, Cheney explains that AI is a catch-all term for describing computer-oriented machines and is not one specific object. “You can’t go buy AI and use it. AI is an enabling technology.”

She goes on to explore definitions of machine learning, neural networks, natural language processing, robotic process automation and analytics before pointing out that while some AI types are described as “weak” or “strong”, all AI is still capable of powerful acts.

However, often decisions supported with the use of AI require us to understand the reasons behind it, which results in the technology having to explain itself, opening up the reasoning to human scrutiny. “We need to be able to explain and defend the conclusions that AI reaches and validating models that are used, taking bias out of decisions,” Cheney says.

In addition to this, the intention of the implementation of AI is not to replace the human workforce, AI will assist and improve processes.

AI application

Rosbrook directs the conversation to the application of AI in financial services and how Citizens Bank is using the technology to increase productivity in the payments industry. “Three to five years ago, there was a narrow view of what AI meant and for corporates, it was not on their radar. Would we use Alexa to initiate a wire? No. It’s a different ball game now.”

She explores how AI tools are benefiting clients in a procedural way, for example, speeding up processes and reducing risk, but in addition to what is “happening under the covers, we are also seeing client facing technology coming to the fore,” Rosbrook says.

On fraud prevention, Rosbrook says that this is “low hanging fruit as every bank has some form of anomaly detection for payments. Tools are getting cooler now and AI provides better ways to determine a person is who they say they are.” She provides voice imprints or handedness checks as examples.

Bots, or robotic process automation, is also being used in the banking back office. Rosbrook uses her own experience managing channels team to advise banks that are implementing new corporate online platforms to use bots to run tests where the outcome is definitive.

“Humans can focus on things that are more subjective and these tests can be separated to allow individuals to concentrate on more pertinent client needs.”

A number of banks have also started to automate paperwork processes and cutting down on the huge backlog of paper with document digitisation. Rosbrook adds: “Scanning and housing documents in databases with tags gives us a library to source information from and the ability to form analysis on repayments and risk, all in one place.”

She also says that while on the client-facing side, consumers are familiar with Alexa and voice UIs, there is an opportunity for improvement in how corporates use this AI tool and the user experience can be personalised with the staging of payments.

With accounts receivable processes, AI can help with the matching of payment information with remittance information, providing more data than an EDI file and improving straight-through processing. Predictive analytics can also help to determine when banks expect to get paid, can take advantage of early payment discounts and better predict cash flow.

“We can also analyse why payments are late, when they are the best chance to reattempt payments. This will become more relevant as card payments become more relevant, but as the industry is shifting to a subscription-based model, card payments will continue to be difficult to access.”

AI decisioning

Karatzas focuses on the B2B space and the inherent expectation for processes to be “glanceable and easy to understand, when technology should play a more advisory role.” He predicts that over the next couple of years, corporate payment processes will become more conversational and banks will be able to navigate and decipher answers independentally, to then pose questions to technology.

“We need to move beyond balances and revenues and profitability and allow me to look at my business in new ways. Help me to connect the dots between the choice that we are making as a business. We should be asking: how does our trending revenue compare to other companies like mine? Rather than: what is our revenue?” Karatzas says.

This benchmarking plays into the subject of explainable AI, full disclosure of processes and transparency of decisions. Once these problems are solved, Cheney highlights that this will change the way customers are engaged with and through the use of machine learning and analysis of data, more insightful and personal services can be provided.

Rosbrook later mentions that in order to do this, a huge amount of data is required and occasionally, this data is maintained in disparate systems.

Now that the industry has access to an increased amount of compute power with the shift to the cloud and there are more people in the financial services workforce that have the talent and knowledge to use this technology efficiently, banks have the capability to explain their guesstimates for loan approvals, for example.

“Cultural mindset is a burden for AI and with the data being derived from statistical outcomes, what is deemed to be the truth may in fact be wrong. Conservative banker minds are difficult to rely on and we cannot trust that the data is reliable either. This is a shift that we will have to move through,” Rosbrook opines and states that it will soon be the bank’s obligation to track how decisions are made.

However: “never fear because millennials are here.” Rosbrook concludes with a positive note on how younger generations will be more accepting and open to new technology, and this transformation will occur.

Comments: (0)