Is ChatGPT Ready for Banking? Yes and No

ChatGPT has captured the imaginations of many bank and credit union executives. The constant hype has drawn their attention. But what is the potential down side of this enticing new technology? What risks would it pose if adopted wholesale for customer service in banking?

As the old saying goes, “When all you have is a hammer, every problem looks like a nail.” ChatGPT may be an impressive AI hammer, but not every tool is perfectly suited for every job.

ChatGPT is one of the latest technology crazes to hit almost every business. Banking is no exception. Some bank and credit union executives see this generative AI model as a panacea of sorts. True, it can give some very clever responses to customer requests, sparking many people’s imaginations regarding the number of different processes it could potentially automate.

Inevitably one is customer service. However, there are a number of different concerns that should be addressed concerning why a generative AI model like ChatGPT might not be ideal for customer service roles.

Lack of Control over Responses Given to Banking Customers

Generative AI works by identifying patterns within the massive bank of data it’s been trained on to respond to queries on-the-fly. Other common chatbot models draw from a predetermined library of responses. By contrast, every response from a generative AI solution will be unique because the system takes slightly different approaches for each input.

Such capability can seem fantastic if you’re looking for some help writing a poem or adjusting the tone of an email. However, it’s less impressive if you need your system to be able to tell someone the interest rate on a specific credit card. When it comes to customer service at banks and credit unions, it’s better to make sure that your users are getting a consistent response every time for the same question, no matter how it’s asked.

Risks of ChatGPT Roulette:

ChatGPT's open-ended nature can go even more awry when you consider that it may give an answer your bank doesn't want it to provide. It could produce a response that is confusing, incorrect or even offensive.

This issue is avoided with other types of AI models which simply fetch pre-written, pre-approved responses that are shown based on the detected need of the inquirer. Conversational AI solutions, for example, leverage a large language model to provide the breadth of information required to help customers and drive toward resolution. It also ensures that those responses are pre-approved, accurate and consistent.

Read more: Ally Taking ChatGPT Slow, But Could Be Using It By Yearend

ChatGPT Presents Compliance and Security Challenges

AI language models require a large amount of data to train the algorithms that power the program. These databases contain a wide range of different materials that fuel the interactions between bot and human. But there can be weaknesses, and even deep concerns, when using it.

For example, generative AI like ChatGPT can cause various undesired issues including potential data and privacy breaches, bias and discrimination, and even intellectual property violations, according to Network Computing. Due to the nature of how these AI models operate, it’s not always possible to ensure that they will follow all necessary ethical or safety considerations without extensive monitoring of their activity.

There are also many concerns with providing a ChatGPT-run chatbot with sensitive company information, or allowing users to share their own private information. Steve Mills, the chief AI ethics officer of Boston Consulting Group, said in a recent CNN Business article that “you should not put anything into these tools you don’t want to assume is going to be shared with others.”

ChatGPT uses its own conversations to further train itself, meaning that this data could potentially be shared with others by the chatbot if asked for it.

Read more:

ChatGPT Is a General Solution, But Banking Is a Very Specific Industry

ChatGPT has a wide range of “knowledge” of many different topics, but this knowledge doesn’t often extend very deeply. While the AI model is trained on an incredibly large amount of data, a good portion of it doesn’t relate to matters your customers will be asking about. Furthermore, it likely won’t be optimized for many of the specific products and features relevant to banking, meaning that there will still be much training to do before such a chatbot would be ready to serve consumers adequately.

ChatGPT and other generative AI tools are not enterprise solutions, so they would need extra work to be able to fit into your existing workflows. On their own, these tools are not going to provide much value to a banking institution, and will need retrofitting in order to be used in an effective manner. Organizations like Salesforce and Microsoft are working on ways to include ChatGPT into their own solutions, and similar measures would have to be taken by other frameworks to make generative AI effective for financial services.

Read more:

Does ChatGPT Have Any Home in Banking?

While there are many reasons why you should reconsider ChatGPT as a customer-facing chatbot, there are still many different ways that you can take advantage of generative AI to create new efficiencies within your financial institution right now. Two examples:

• Support for internal operations. Because of ChatGPT’s ability to analyze and synthesize vast amounts of information into concise form, it can be a valuable research and training resource for employees. This could make it handy whenever they are looking to gain working knowledge about a certain topic, such as upcoming regulations.

Findings that previously took hours of search engine scrolling can be surfaced in seconds. Information that ChatGPT provides should still be reviewed for accuracy, but the technology can be a great way to condense research time significantly.

Want to go deep on AI best practices for banks?

Attend our AI Masterclass — Unlocking the Power of Artificial Intelligence in Banking — at The Financial Brand Forum 2024 on May 20-22 in Las Vegas. Led by Ron Shevlin, chief research officer at Cornerstone, this three-hour workshop will be jam-packed with lessons learned from industry leaders and real-world case studies.

For more information and to register, check out the Forum website.

• Create training materials or FAQs. Using ChatGPT to analyze large volumes of corporate data, such as the hundreds of PDFs around company policies, how products work, and other general information, can offer a quick and easy way to create FAQs and other training materials for new employees. This could eliminate the need to dedicate significant employee time to such rote chores, allowing the training staff to focus on more strategic tasks. Be careful to not give ChatGPT any sensitive information, per the security concerns explained above.

In addition to these examples of leveraging generative AI in your current workflows, you should also take advantage of the ways that your existing enterprise solutions are implementing the technology.

The recent public interest in AI is a good catalyst for banking organizations to start adopting the technology themselves, but make sure you’re not simply jumping towards the newest, hyped solutions and instead spending time finding a solution that better suits your needs, safely.

About the author

Jake Tyler is the AI and automation market lead for Glia Technologies.

This article was originally published on . All content © 2024 by The Financial Brand and may not be reproduced by any means without permission.