This Blog Post Was Brought to You By AI: A Closer Look at AI Claims in Fintech

AI: Or How I Learned to Stop Worrying and Love the Algorithm

Artificial Intelligence (AI) is at the forefront of my mind most days, by choice and by force. The term is thrown around so frequently in marketing materials, think pieces and headlines that I fear it has been rendered nearly meaningless. From optimists insisting we won’t be able to live without it to alarmist declaring it’s the downfall of humanity, we must accept a simple fact: It’s already here and it has been, for quite some time. Many things in my life, and likely yours, already leverage AI. Like the autocorrect I relied on while writing this blogpost, or the spam filters in my email that guard my inbox—and eyes—from endless Elon_Musk* and #Face.Book messages, or the Instagram algorithm that somehow correctly predicts I am in the market for another pair of platform boots.

While applications of AI can be groundbreaking—like cancer-detecting models and IBM’s Watson, some AI is just really good at playing Go or writing formal emails—AI already exists all around us, in ways we take for granted. So, rather than being something scary or even revolutionary, much of the AI technology in circulation is pretty innocuous and honestly, rather old hat. After all, before there was Claude AI, there was Clippy. However, after the explosion of ChatGPT, I noticed many financial firms promoting their shiny, new AI-powered content.

But, what exactly does AI-powered mean?

In many ways, this blog post is AI-powered. Importantly, it is human written. But by a human who sometimes cannot spell. A human who uses search engines for research. A human who uses Face ID to login to the apps that appear in this piece. And likely, it will be AI that brings this post to you, either as a search engine result or perhaps delivered to you by the algorithm powering LinkedIn’s feed. In this sense, AI-powered is more of a marketing term than a statement of quality or complexity. It is too generic to say much about the actual technology underpinning a product or service. ChatGPT and self-driving cars are AI-powered, but so is this grill. And your Roomba. The vast range of use-cases means we need to be discerning and critical when we encounter the term.

So, when a company says “look at our AI-powered chatbot,” we need to think about what that actually means, which is potentially very little. It does not necessarily mean high-tech or cutting-edge. In fact, we’ve had ‘AI-powered’ chatbots since the 1960s. AI-powered does not exclusively refer to large language models (LLMs), like ChatGPT or Google Gemini, technology known as generative AI (genAI). GenAI takes an input and creates a novel output. Your average bank chatbot doesn’t do that, but it probably doesn’t need to. That’s not to say firms shouldn’t look to more complex uses of AI, but they should also be thoughtful and creative about how they implement them. Firms shouldn’t just be thinking about how they are going to use AI, but also why they are doing it.

Klarna: Leaning into your brand

Klarna has adopted AI into its brand strategy, implementing applications throughout its platforms. While still a financial firm, Klarna has found its footing as a shopping brand. The firm’s virtual assistant (VA), AI Assistant, for example, functions as both a customer service bot and a personalized shopper. One might be tempted to think that the use of chatbots would lead to a lower-touch, depersonalized experience. Klarna proves otherwise. The VA offers a kind of luxury service: Shopping Concierge. With the ability to compile information about products, make recommendations based on various inputs and offer product comparisons, AI Assistant offers synthesized content, allowing clients to skip the research portion of deal hunting and shopping. The use of AI here allows the firm to engage with its audience without creating a more expensive product.

The firm partners with ChatGPT maker, OpenAI, for the LLM that underpins AI Assistant. Positioning itself as a disruptor, the firm’s use of cutting-edge genAI technology fits well in this narrative. The claim is something like ‘AI is part of what sets us apart from incumbents’. This willingness to publicly embrace genAI is certainly noteworthy. Corporate Insight conducted a study which found a wide range of sentiments towards AI, with 45% of respondents indicating they were “more concerned than excited” and 32% of respondents reporting they were equally “excited and concerned” with the increased use of AI in society. The same survey found that 61% of respondents indicated they “somewhat or strongly agreed” that they could only trust AI with low stakes requests like shopping recommendations. So, while AI skepticism is certainly present, Klarna seems to be on the right track with the kind of AI clients are willing to accept.

Wells Fargo: How to work with what you’ve got

In 2023, Wells Fargo launched its chatbot, Fargo, which leverages Google Cloud AI for its plain language recognition and financial insights. The bot quickly handles basic tasks like identifying account balances, displaying transaction history and redirecting users to relevant interfaces. Fargo comfortably addresses requests phrased in a variety of ways thanks to Google’s Dialogueflow product. Google boasts Dialogueflow’s ability to combine deterministic—think if “x” or “y” input then “z” output—and generative AI elements to produce a smarter but controllable chatbot. This balance makes Fargo a particularly effective VA.

Beyond the smooth and consistent plain language interactions—offered in both English and Spanish—the use of AI allows the VA to produce a range of personalized financial insights. Your bank already knows how much money you spend and what you spend it on. Content like Fargo’s Activity Forecast allows the firm to leverage the information they already have and add value to it by introducing a predictive element. Fargo uses past and scheduled future transactions to predict balances for the upcoming week and month. Tools like this are a great way to use AI to solve a customer need, like budgetary planning, without overstepping or asking too much of the customer. Users don’t need to input their expected expenses. The firm’s AI can recognize patterns from the transaction data they already have. All the user needs to do is keep using their account.

Klarna and Wells Fargo provide unique, firm-appropriate examples of how genAI can be used to produce strong AI-powered chatbots. Other firms should consider their strategies and client needs when evaluating how AI can power their products. For some companies, that might look like better predictive results for their webpage search bar, for others, perhaps a personable chatbot. AI-power is here to stay, but what that means for you, your firm, and your clients still requires some good ol’ fashioned human intelligence. Or maybe ask ChatGPT?

Check out CI’s Fintech Monitor for the latest advancements from fintech firms. And check out our Insights section for more on the latest trends and best practices from across the world of financial services and healthcare.