How Finance and Banking Professionals Can Use ChatGPT

ChatGPT and AI in Finance

A question I have often heard over the past few weeks is, “How can ChatGPT help finance professionals?”

I’m someone who keeps tabs on finance subreddits, scans chatboards, and talks to our customers frequently. I’ve been hearing stories of how ChatGPT is being used almost daily at this point, so it seemed like a great time to summarize some of the most common ways it’s being used by finance and banking professionals. If you are unfamiliar with ChatGPT, I’ll cover that too.

What is ChatGPT?

For those who haven’t yet heard about ChatGPT, or have heard about it but have no clue what it is, let’s start with some background. I turned to ChatGPT itself to write an introduction.

Me:

What is ChatGPT and how would you explain it to someone who isn’t familiar with it?

ChatGPT:

ChatGPT stands for “Generative Pre-trained Transformer for Chatbot”. It is a type of machine learning model that uses natural language processing to generate human-like responses to textual inputs. The GPT model was developed by OpenAI and has been used in various applications, including chatbots, language translation, and text generation. The model is trained on large datasets of textual data and uses a deep neural network to generate responses that are contextually relevant and grammatically correct. The “pre-trained” part of the name refers to the fact that the model is initially trained on a large corpus of data before being fine-tuned for specific applications.

Me:

Can you provide a different way to describe it?

ChatGPT:

ChatGPT is an AI-powered chatbot that can communicate with humans in a natural language. It has been trained on a large dataset of human conversation and can understand and generate text in response to user inputs. ChatGPT uses Natural Language Processing (NLP) techniques to interpret the meaning of the user’s message and provide a relevant response. 

NLP is a field of study focused on making machines better understand human language. With NLP, ChatGPT can analyze and respond to natural language queries and conversations in real time, making it an ideal tool for customer service and support.

It is designed to provide helpful and informative responses to a wide range of topics and questions. Users can ask ChatGPT about various things such as general knowledge, current events, science, technology, health, entertainment, and more.

Me:

How could a finance or banking professional use ChatGPT?

ChatGPT:

One of the key benefits of NLP for finance and banking professionals is the ability to process large amounts of data in a short amount of time. For example, ChatGPT can be used to analyze customer feedback and complaints, which can help organizations improve their products and services. It can also help organizations identify patterns and trends in customer behavior, which can be used to develop more effective marketing and sales strategies.

I continued the conversation and asked for more specific examples, but they were all fairly generic and repetitive. It seemed I had hit the wall on ChatGPT’s usefulness in this conversation, so the rest of this article summarizes what I am hearing and thinking on this topic.

Use Cases for ChatGPT in Finance and Banking

While ChatGPT was originally designed for text-based conversation, it has numerous potential applications in finance and banking. If there is an overall theme of what I read and hear from people already using it, it’s faster data analysis and faster collection of a wide variety of information sources into succinct, useful summaries. Some are using it for accuracy in Excel and models, particularly those who are less experienced in creating models or checking formulas.

Investment analysis and prediction

Analysts can use ChatGPT to analyze financial statements and other data to make investment decisions. ChatGPT can also help identify trends and patterns in market data that may otherwise be missed. ChatGPT works incredibly quickly, so it is able to do this much faster than a human can. If nothing else, it provides a helpful, quick starting point for a human to use to dig further into.

Compliance and risk management

ChatGPT can be used to analyze legal and regulatory documents, ensuring that organizations are complying with laws and regulations. It can also be used to identify and assess potential risks, allowing organizations to mitigate those risks proactively. Again, the speed at which ChatGPT can do this is much faster than a human’s, and AI might pull from a wider variety of sources. One of the most useful things about ChatGPT is its ability to have threaded conversations that build on each other. For example, when risk is identified, ChatGPT can respond with relevant risk-mitigation strategies. The conversational aspect of ChatGPT is part of what makes it so helpful in day-to-day work.

Forecasting

ChatGPT can be used to generate more accurate financial forecasts by training it on historical data and using it to predict future trends. Theoretically a company can use ChatGPT to identify patterns and correlations in data, which can help make more informed decisions about financial models.

Identifying key variables

ChatGPT can help identify which variables are most important to a financial models. Training it on data and analyzing the results through a conversation can help better identify which variables have the most impact on a model, and help adjust the model accordingly.

Data entry

ChatGPT can be trained to automatically enter data into Excel models. This potentially saves a significant amount of time and reduces the risk of errors in models.

Data analysis

ChatGPT can help analyze large datasets and find patterns that are not easily visible with traditional analysis methods. By training it on data, it can quickly identify trends and patterns, which can inform financial models and decision-making. This might be my favorite Excel-based use case.

Reports and report outlines

ChatGPT can be trained to generate outlines and reports that summarize the data in financial models. By automating the report generation process, it saves time and reduces errors.

I’ve heard of people using ChatGPT for data analysis and report generation, but I was personally unclear as to how one would go about it. So, of course, I asked ChatGPT for specifics.

Me:

Can you tell me more about how ChatGPT would analyze large datasets and find patterns? Would I upload an Excel sheet or a data set? I am confused about how to get ChatGPT to do this.

ChatGPT:

To use ChatGPT to analyze financial data, you would typically first need to prepare your data in a suitable format, such as a CSV file, which can then be uploaded to the platform or environment where the ChatGPT model is being run. Depending on the specific tools or libraries you’re using, you may need to do some additional preprocessing to convert the data into a format that is compatible with ChatGPT.

Once you have your data in the appropriate format, you can then use ChatGPT to train a model on your data, which involves providing the model with examples of the patterns or relationships you want it to identify. The training process involves adjusting the model’s parameters and weights to optimize its ability to recognize these patterns and relationships.

Once the model is trained, you can then use it to analyze new data and generate insights. For example, you might input a set of financial data into the model and ask it to identify any trends or patterns that might be relevant to your analysis. The model might then generate a report or a set of visualizations that help you understand the relationships between different variables in your data.

Overall, the specific steps you’ll need to take to use ChatGPT for financial analysis will depend on the tools and libraries you’re using, as well as the specifics of your data and analysis goals. However, with the right preparation and approach, ChatGPT can be a powerful tool for identifying patterns and trends in financial data and generating insights that can inform financial models and decision-making.

I still felt confused, so in response, I asked, “Can I upload a data to chat.open.ai in order to have ChatGPT help me analyze some data?”

ChatGPT:

Unfortunately, it is not possible to directly upload data to chat.openai.com or use ChatGPT in a direct-to-consumer manner at this time. chat.openai.com is designed as a platform for exploring the capabilities of OpenAI’s natural language processing models and is not currently set up to handle data uploads for analysis.

However, OpenAI does offer a variety of tools and APIs that can be used to work with their natural language processing models, including ChatGPT. One option is to use OpenAI’s GPT-3 API, which provides programmatic access to the GPT-3 language model. With this API, you can input text data, such as financial reports or news articles, and use GPT-3 to generate new text, answer questions, or perform other language-based tasks.

Another option is to use OpenAI’s platform for training and deploying custom language models. This platform allows you to train and deploy your own language models, including ChatGPT-based models, for a wide range of language tasks, such as text classification, sentiment analysis, and language translation. With this platform, you can upload your own data and use it to train a model that is tailored to your specific needs.

Both of these options require some technical expertise and development skills, but they offer powerful ways to use ChatGPT and other natural language processing models for a wide range of applications, including financial analysis.

Now I get it! I think one aspect of ChatGPT that’s important to understand to unlock it’s potential is to actually use it for a conversation, with each question and response building on the last they way I did in the example above.

In addition to hearing people chatter about ChatGPT’s usefulness for finance and banking, I also like to scan Reddit conversations to find real examples of how professionals are using the tool. Here are a few Reddit conversations I’ve seen about ChatGPT and finance recently:

• This thread identifies use cases including
  • Structuring a pitch deck covering some pre-prompted information about the company
  • Building a financial model from scratch; DCF for example
  • Auditing a financial model to check for and comment on inaccuracies
• This thread points out: “In finance it can accelerate/handle the fundamentals of accounting or market analysis. However, if algorithms were truly able to learn how to predict the stock market then we would have already heard a lot more about AI engineers striking it big in trading.”
• In this thread a user shares the investing criteria it provided to ChatGPT and the resulting recommendations the engine made. There is great commentary here about the limitations of ChatGPT including how some of the information sources are outdated and some aspects of the investing criteria were missed in the recommendations.
• I love this thread because it’s an example of using ChatGPT for a threaded conversation, using one response to ask further questions and get deeper, more relevant and specific information that’s related to the conversation.
• This thread points out the limitations of ChatGPT including limited sources and repetitive answers through the course of a conversation. I’ve not experienced that, but I think it depends on the topic.
• Here’s an example of a user trying to use ChatGPT to answer questions on the CFA test. Spoiler alert, ChatGPT got a lot wrong!

One theme on Reddit involves the jobs ChatGPT may replace, including trading, accounting and other roles in finance, particularly entry-level positions. A comment in one of the threads above summarizes my own thinking on the topic: “It will become an essential tool, but all those jobs require human brain…” and “I would also add that finance is a wide category of careers. The best way to protect yourself is to never stop learning or growing. If you work hard and smart no technology is a threat. It’s a tool.”

Learning how to leverage tools to make us work faster and better can only be a good thing. For now, I think any profession, in any industry, can be using ChatGPT as a tool for efficiency, efficacy, and broader knowledge summary. I don’t think it can be used as a replacement for the work we need to do and the tasks we have on our plate.

Challenges and Limitations of ChatGPT

While ChatGPT has numerous potential applications in any industry, including finance and banking, there are challenges and limitations to be aware of. One challenge is the potential for bias in the training data used to develop the model, and the recency of the data and information that was used to train it. If the data does not represent the entire population, or is outdated, ChatGPT is likely to make inaccurate recommendations.

The other big risk is with reliability. The internet is full of examples where ChatGPT got answers very, very wrong, so it’s a good best practice to check ChatGPT’s outputs to ensure accuracy. For example, I asked ChatGPT who the CEO of Corporate Finance Institute is and it replied, “I’m sorry, but as an AI language model, I do not have access to real-time information on the CEO of the Corporate Finance Institute. However, as of my knowledge cutoff date of September 2021, the Corporate Finance Institute did not have a publicly listed CEO on their website or on any other reputable sources. It’s possible that the information has since been updated, so you may want to visit the Corporate Finance Institute’s official website or contact them directly for the most up-to-date information.”

I also asked it about our co-founder and Chief Content Officer, Scott Powell, and it replied (inaccurately), “I’m sorry, but as far as I know, there is no one named Scott Powell who is associated with the Corporate Finance Institute. The Corporate Finance Institute is a global financial training organization that offers online courses and certifications in various areas of finance, such as financial modeling, valuation, and investment banking. Their team consists of experienced finance professionals, instructors, and support staff, but I could not find any information about a specific individual named Scott Powell who is affiliated with the organization. If you have more information or context about the Scott Powell you are referring to, please let me know and I can try to provide more accurate information.”

A great reminder to always check the output of ChatGPT and only use it as a starting point.

The Future of ChatGPT for Finance and Banking

ChatGPT and other NLP tools have the potential to transform the way we work, not just in the finance and banking sector, but in any industry. While there are big limitations to consider, especially right now, the benefits of using ChatGPT are potentially significant and worth exploring. If you are curious about getting started, head on over to ChatGPT and ask it a question about what its future use cases might be!

Additional Resources

Free Financial Modeling Guide

Big Data in Finance

Why Online Training for Finance Teams Makes Sense

See all eLearning resources