Artificial Intelligence in Banking: Possible Impact of OpenAI’s ChatGPT
If you haven't heard of it already, OpenAI's ChatGPT is an artificial intelligence-powered text tool capable of providing cogent answers in seconds to just about any question you throw its way. It is viewed with a mixture of awe and apprehension.
Despite the unease, ChatGPT is acquiring new users faster than any of the most popular consumer apps created up to now, which makes it a tool that all industries must be ready to consider.
The chatbot launched for public use only in December but has already amassed 57 million monthly active users in its first month of availability. It is expected to have surpassed 100 million monthly active users in January, according to new research from investment bank UBS.
By comparison, it took social media app TikTok roughly nine months after its launch to build the same user base, and roughly two and a half years for Instagram to acquire 100 million monthly active users. Google Translate took six and a half years to reach that milestone.
A Formidable Tool
The potential capabilities of ChatGPT are impressive, to say the least. As described by Entrepreneur.com:
The program is equipped with a transformer-based neural network with extensive data and information up to 2021. This means that when a human user inputs a command or question, the AI intakes it, processes it, analyzes it through its deep neural network and produces an output in a text pattern that mimics human diction.
ChatGPT can have entire conversations, admit mistakes, challenge incorrect premises and write essays, scripts, articles and headlines. Because of its intelligence and massive database, ChatGPT can serve many industries — even banking.
Obviously, the next question is - - what roles are foreseen for ChatGPT in banking?
Possible Bank Deployments for ChatGPT
Entrepreneur.com anticipates ChatGPT's impact on banking in these primary areas:
ChatGPT's Limitations
That doesn't mean there aren't pitfalls of which to be aware, as further reported by Entrepreneur.com:
- Cost: While ChatGPT is free, the cost of implementing it into a banking system would not be. Banks must consider how much it would cost to integrate, train and deploy, maintain ongoing operational costs and keep up with regulatory costs. ChatGPT is a great resource, but it might not be accessible to smaller banks with a tighter budget.
- Security: OpenAI, the creator of ChatGPT, has taken immense measures to ensure the security and privacy of ChatGPT users. However, banks need to take their own course of action in addition to ChatGPT to safeguard the security of its users. This can be seen as an extra task and worry for banks, making the process less efficient.
- Limited human interaction: While automation certainly has its perks, it does reduce the need for human labor.
The article also notes an issue that restricts the usage of ChatGPT from a regulatory standpoint.
ChatGPT's database only has access to information until 2021, so any new laws or trends cannot be added to the AI's considerations.
As we all know, banking regulations are subject to change -- with the possibility of implementation at a moment's notice. This will limit ChatGPT's usability from a regulatory and operations standpoint.
And, we might add, there is the danger that over-reliance on the ChatGPT technology as a consumer "front end" may lead to errors usually caught by the human eye. For instance, one could conjecture by the sentence construction and writer's credit given (see at right) that our primary source article for this blog entry, Entrepreneur.com, actually deployed ChatGPT to create their article. At one point in the article, there is a numbering error that, to a nonhuman editor, would not have registered -- perhaps an "Easter Egg" to keep us optimistic but cautious about the utility of ChatGPT!
Byline for Entrepreneur.com website article entitled “What Are ChatGPT’s Potential Impacts on Banking?”
When we examine ChatGPT's potential in payments -- specifically checks -- its utility will be relatively limited, if useful at all. The key for payments is utilization of artificial intelligence and machine learning models that are specifically trained to the payment. For check processing, these models analyze the image of the checks and extract data accurately to ensure the payment is processed correctly. For check fraud, these technologies will analyze the images of checks to identify deviations from previously cleared checks such as check stock, writing styles, and signatures that do not match, while also looking for an amount discrepancy on the item -- things that ChatGPT is not built for.