Artificial intelligence (AI) continues to barrel headlong into the future and make daily headlines. What many may not realize is that for savvy financial services organizations, the future has been here for some time already. Experimenting with how to use AI in new ways to get investors what they need when they need it is something that stays top of mind for our digital professionals at Schwab.
The most relevant information served up intuitively to investors? We’re using AI to help us do that.
A better client experience that caters to personal preferences? AI is helping us there as well.
Helping experts make the best use of their time to serve clients? You guessed it—AI is a powerful tool.
But what happens when personalization gets, well… a little too personal?
You may have heard about the reporter who engaged with Bing's artificial intelligence search chatbot, only to have it try to break up his marriage. The challenge facing the industry is how to balance the possibilities of evolving AI technology with the trust and safety clients expect and deserve.
There’s simply too much benefit—for investors and the people who serve them—for the answer to be avoiding promising new technology altogether. Instead, there are some wise precautions that forward-thinking digital professionals can adopt to keep the cutting edge from drawing blood.
Why not both?
As the Managing Director leading Digital Investor Solutions for Schwab, Zack Gipson sees the wisdom in using AI to better understand clients’ questions. He says it can help guide clients more effectively to the right answers wherever they want to engage, rather than relying on AI to produce the answers and forcing clients into channels they didn’t choose.
“We’ve seen ChatGPT and Bard (Google’s AI chat) reveal some of the inherent challenges and limitations of unbounded AI, even in their early days,” shares Zack. “The best way forward is continuing to focus on a strategy of ‘people plus technology,’ leveraging the powerful combination of human expertise working hand-in-hand with technology to enable scale and great experiences.”
AI should help humans spend their time on the biggest-impact activities
One example of how AI can help power human effectiveness is the Schwab Assistant digital tool, which responds to millions of natural searches per month. Because of how natural language processing (NLP) is baked into the application, it can accurately analyze and provide insights on that data swiftly and at scale.
Surfacing the most important trends this way allows us to be more agile by focusing our time on developing the resources and answers that clients may need at any given moment.
Christina Nickel, Head of Digital Channel Management at Schwab, has helped shepherd team members and the technology to surface important trends for investor education.
“Just in the past few years, AI search analysis has helped identify key areas of client interest, from taxes to meme stocks, crypto, and more. That’s made us faster at looping in subject-matter experts to create or update time-relevant resources that the Schwab Assistant can serve up in response to client inquiries.”
Use AI to predict questions and have what clients want waiting
We asked Casey Royer, Director of Digital Intelligent Interactions, for an example of an everyday search function that could be transformed in a responsible way for the needs of the financial service industry as AI technology evolves. Casey offered this hypothetical to illustrate:
“Today, if you ask a search AI, ‘How many shares of X stock is in my portfolio?’ and the AI has access to that data, chances are it can quickly serve you that information. But if that answer sparks a follow-up question from you, you may need to start back at the beginning to get that second answer.”
Casey says this won’t always be the case.
“Down the road, the technology surfacing right now may help search tools learn to make strong educated guesses about those second or third or fourth questions after seeing the first one. For example, once a person knows how many shares of X they have, they’re also likely to ask at what price that stock is currently trading or what percentage of their holdings that position represents.”
By having the data to make predictions, Casey explains it will start to become an easier process.
“Rather than the client needing to start at the beginning with a fresh search, the option to get the answers to those statistically probable follow-up questions could be served up dynamically. This would save time and make the interaction more intuitive and conversational.”
Know how much AI people actually want
In our 2021 Investing and Technology Survey, we discovered a roughly even divide of preferences between those who preferred to engage with digital tools only, those who wanted human interactions only, and hybrid investors who were happiest engaging with both.
There was some variation when looking across generations, but in reality, the divide is nowhere near as pronounced as the popular “avocado toast” vs. “OK Boomer” clickbait narratives would have you believe.
NLP can already help meet the needs of all three groups in their channels of choice:
- Enabling effective 24/7 self-service for the digital-only segment gives them the experience they want, which also frees up more service professionals to focus on clients and queries that require the human touch.
- NLP can create more seamless handoffs for hybrid preferences. When Schwab Assistant needs to refer a client to a human service rep, it provides the rep with the most important highlights of the chat conversation, which gives them the ability to spare that client the frustration of starting over at square one.
One last thought
A measured approach to AI improves the client experience by allowing them to stay in their channels of choice. It means exploring and experimenting in responsible ways that benefit from the upsides while protecting against the downsides. It means enhancing the best of what financial services professionals have to offer—people and technology together in service of investors.
Author’s Note: ChatGPT was not used in the writing of this article. Of course, that’s exactly what you’d expect ChatGPT to say, so…