Generative AI: Chatbots that work?

In a recent blog we debate the current wave of scepticism about AI. In brief, and outside its obvious applications in fraud detection and management, some senior executives have expressed doubts about the added value AI brings to payments.

Generative AI: Chatbots that work?

However, a new white paper from Mastercard and new developments at Marqeta cast fresh light on where AI, or at least one version of it, might prove genuinely useful.

Generative AI, much-mentioned of late in both the trade and mainstream media thanks to OpenAI’s Chat GPT, is – to quote Google founder Larry Page – “the ultimate version of Google: a search engine that understands everything on the web … it understands what you want, and gives you the right thing.”

While ChatGPT itself has been overhyped, applications that are more limited in scope look more promising.

“Marqeta estimate generative AI can reduce the cost of integrating new payment functions by 75%.”

In a development that drew less attention than it deserved, last month Marqeta announced the introduction of Docs AI, a tool that enables developers to navigate their way through the company’s document library thanks to intelligent interpretation of their written requests for how to embed functions such as card issuance, processing, BNPL and more into client web portals and apps.

Marqeta estimate that developers can reduce the time they require to code and test new integrations by around 75% thanks to generative AI – an important cost saving given the expense of software development today.

Relevant chat: and faster, cheaper coding

In the Q3 2023 edition of their “Signals” series, Mastercard set out their vision for where generative AI might prove useful in the short and longer term.

One application sure to be of interest to CFOs grappling with high development costs is the potential to automate coding and software development.

Since a proportion of modern coding typically involves patching together existing off-the-shelf “widgets” and plug-ins, this appears long overdue.

Another interesting application proposed by Mastercard for the near term is more responsive and relevant chatbots.

Generative AI makes this possible, Mastercard argue, thanks to better understanding of consumers’ needs and smart access to relevant data inside an organisation.

In other words, instead of a set of stock responses, chatbots powered by generative AI would be able to give responses tailored to customer interests – and based on internal data the customer wants to access.

Looking further ahead, Mastercard see generative AI as automating large parts of existing accounting and wealth management functions – music to the ears of small business owners and investors everywhere who are currently paying high fees to access such services.

Payments Cards & Mobile Opinion

While generative AI is still in its infancy, it seems to have genuinely useful applications that may be realised within the next two to three years – most particularly in orchestrating app coding and speeding up the software development process.

The development of more responsive and effective chatbots would also be great news for consumers.

Finally, it seems current AI applications are most effective when applied to more limited tasks and datasets.

It would be great to see generative AI deliver on a more limited objective, rather than watch it drown in hype about its transformative power.