Chats and Balances: Human Oversight Will Dictate ChatGPT’s Utility for Trade Finance

Since its release on 30 November 2022, ChatGPT has taken the world by storm. It reached 1 million users in just five days from launch and currently has more than 100 million users.¹ [March 2024 update: 180.5 million active users]

I started experimenting with a ChatGPT bot in January 2023 and tested it with a number of questions on trade finance. I wanted to share some of the results and evaluate whether it could be used for trade finance, but first a brief explanation of what ChatGPT is and its uses for today.

ChatGPT, a member of the generative pre-trained transformer (GPT) family of language models, was developed by OpenAI, an organization in artificial intelligence (AI) research. Its backers include prominent personalities Reid Hoffman (co-founder of LinkedIn) and Peter Thiel (co-founder of PayPal), and IT companies Microsoft and Infosys.

One of the key features of ChatGPT is its ability to recognize context and respond in a way that is relevant to an ongoing conversation. This is achieved through use of advanced natural language processing (NLP) techniques, such as contextual embedding and attention mechanisms. These techniques enable ChatGPT to identify the meaning of words and phrases in the context of the entire conversation, allowing it to generate responses that are both coherent and relevant.2 The model was trained on a vast array of conversational data, including social media conversations, online forums, and customer support interactions, to enable it to generate realistic and contextually appropriate responses.

Today, ChatGPT is being considered for use in various industries and applications. One of its most significant applications is in customer service where it can be used to power chatbots and virtual assistants. By using ChatGPT, these chatbots can provide more human-like responses, improving customer satisfaction and reducing the workload on human support agents.

Another significant application of ChatGPT is in content creation. The model can generate coherent and contextually relevant language, making it ideal for generating articles, reports, and other written content. This has the potential to revolutionize content creation, enabling writers and content creators to generate high-quality content quickly and efficiently. An example of ChatGPT’s use in this regard is this article itself. I asked ChatGPT to help write this article and what you read here is a combination of ChatGPT and my authorship, with responsibility for the final article fully mine.

ChatGPT can also be used in education and research. The model can answer questions and provide explanations, making it useful for students and researchers. It can also generate new insights and hypotheses based on its understanding of text data, enabling researchers to explore new areas of inquiry and gain new insights.