Advisor Perspectives welcomes guest contributions. The views presented here do not necessarily represent those of Advisor Perspectives.
Earlier this year, Virgin Money’s chatbot became internet famous for a brief moment after a customer asked how he could merge his two Virgin Money ISA accounts, and the chatbot responded: “Please don’t use words like that. I won’t be able to continue our chat if you use this language.”
Though it seems like just a humorous snafu, this situation reveals something far more troubling. This customer trusted this financial institution with his hard-earned money and expected its AI algorithm to know the difference between a mildly risqué word and its own company name. When something like this happens, a customer is right to ask: Should I really trust this financial institution with my assets?
In the financial-services world, trust is everything. Executives lie awake at night worrying about these sorts of reputational, operational, and compliance risks. Many highly regulated industries have been reluctant to roll out generative AI tools due to the potential for hallucination and communicating misinformation, and the risk posed by these errors.
According to a BCG survey on AI, only 25% of banks have woven AI tools into their strategic playbooks. For many financial services executives, hallucinations and lack of trust are the top reasons why they have been slow to adopt a generative AI strategy.
A Crucial Trifecta
Financial institutions must be able to compete in the race to harness generative AI. When it comes to integrating gen AI into financial services platforms, companies need technology that offers three defining characteristics: speed, novelty, and quality.
Users don’t want to wait 90 seconds while an algorithm fact-checks every response; they’ll think something is broken. Meanwhile, without a second layer of vetting, institutions leave themselves vulnerable to hallucinations.
For AI to be successful, companies need to invest in technology that double checks each and every claim — and does so without degrading the user experience. And when facts don't align, having a human in the loop provides a critical final safeguard.
Most AI tools that purport to offer financial AI analysis have disclaimers stating that they’re for “entertainment purposes only.” While disclaimers may work for neo-brokers that tend to focus on growth above all else, traditional financial services companies are averse to risk and tend to be more cautious when it comes to innovation. They will continue to be reluctant to embrace new technologies without greater quality control.
Many fintech companies have solved one or two components of this trifecta, but doing all three is extremely difficult. Until major financial services firms can be certain that the gen AI model is fast, novel, and high quality, we’ll continue to see a slower pace of adoption compared to other major industries.
Firms that master all three will not only overcome the quality concerns slowing the industry; they will deliver the next generation of trusted financial-AI experiences that their users expect.
Chet Stuut is chief technology officer at Pebble Finance, a portfolio analytics and technology company that synthesizes news and proprietary research, connecting the dots between real-world events and investment performance.
A message from Advisor Perspectives and VettaFi: Discover something new! Click here to register for our upcoming webcasts.
Read more articles by Chet Stuut
Advisor Perspectives welcomes guest contributions. The views presented here do not necessarily represent those of Advisor Perspectives.
