AI in finance: What ChatGPT and generative AI means for CFOs

With ChatGPT and generative artificial intelligence (AI), the pace of progress with AI in finance has been remarkable.

From mobile phones to the metaverse, we’re used to technological breakthroughs.

But even with this rapid progress, AI’s leap into the public consciousness has surprised many.

In particular, the emergence of ChatGPT has shaken the world, with people only beginning to understand what it means for the future of business technology.

As we continue to embrace technology, CFOs and financial leaders need to understand the growing potential of generative AI and ChatGPT.

With the ability to automate mundane tasks, analyse large data sets, and even generate reports, AI could change how your business operates.

In this article, we explain how generative AI tools and their potential applications work. By understanding what they do, you might see opportunities to use them to increase efficiency, reduce costs and drive business growth.

Here’s what we’ll cover

What is generative AI and ChatGPT?

Generative AI is a type of artificial intelligence that can create new content independently, without explicit instructions or data input.

One example of generative AI is ChatGPT, which stands for Generative Pre-trained Transformer.

It’s a language model that uses deep learning to generate human-like responses to user inputs.

By analysing vast amounts of data, ChatGPT can learn to mimic human conversation and generate responses that are both coherent and relevant to the conversation at hand.

Why are financial professionals excited about AI?

AI tools are becoming increasingly popular in financial management because they can automate routine tasks, improve accuracy, and provide valuable insights.

With reams of financial data generated daily, traditional financial management methods no longer feel sufficient for the work you need to do today.

With AI and automation, you can analyse vast amounts of data, and identify patterns and trends you might otherwise miss.

With more accurate financial reports, you can make more informed decisions.

Additionally, AI tools could automate routine tasks, create financial reports and handle customer enquiries, freeing your team to focus on more strategic tasks.

Angus Gregory, CEO of software provider Biomni, says passing information from one colleague to another is essential in maintaining smooth operations within a company.

This is especially true in the finance sector.

Angus says: “In finance, day-to-day activities are heavily based on knowledge sharing, be that business-specific processes, compliance information or developments in the industry.

“Many of these tasks could be automated, an area in which AI excels.

“Using ChatGPT in the financial sector could significantly improve organisation, productivity and overall growth.”

As AI technology advances, it’s increasingly likely that your competitors will adopt AI tools as part of their financial management strategy.

Microsoft is a company that is betting the farm on AI’s potential to transform financial management, investing $10bn (R183,938,800,000) with ChatGPT’s creator, OpenAI as a key part of its business strategy.

You can connect the dots and see that Microsoft’s investment in AI is partly driven by a belief that AI can help financial professionals make better decisions, improve operational efficiency, and drive growth.

Speaking at the World Finance Forum in London, Microsoft UK CFO Mark McCardle said AI is a “game changer” and the most intelligent scientists in the world describe AI as “more transformational than the invention of the internet”.

“AI will change how we work, though, in finance, we’re not yet using it to its fullest potential,” he said.

“However, we’ve advanced in machine learning when it comes to predicting the future by using information from the past—algorithms can forecast better than I can.

“At Microsoft, we use AI for standard forecasting, where we understand the customer pipeline through revenue forecasting of our different solutions.

“Every week, we review this information with market leaders and discuss where our customers are going. It’s where the science comes together with human review—going with what computers tell us and rationalising it.”

ChatGPT—a “super assistant” for financial teams?

Let’s focus on what your finance team might already be doing with ChatGPT.

ChatGPT can generate texts, create summaries, assist with complex assignments or research tasks, automate repetitive tasks, and pull out requested information—all of which are invaluable for streamlining the daily work of financial teams.

Angus Gregory outlined specific examples, such as:

• Analysing huge datasets in a matter of seconds—you could assess financial markets, identify patterns, and predict the outcome of certain decisions quickly and effectively.
• Helping with governance and complying with legal requirements when used within a framework and provided with the right prompts.
• Generating digestible summaries of documents and policies, as well as presenting large amounts of information in an organised manner for reporting purposes.
• Supporting ‘blue sky thinking’ and brainstorming sessions around topics such as growth or cost saving.

Daniele Grassi, CEO of fintech company Axyon AI, says AI can see things financial professionals can’t see because of its ability to analyse and cross-reference large quantities of data at a granular level quickly and accurately.

He says: “AI has a superior ability to identify non-linear patterns in data, such as fundamentals, technical indicators and macroeconomics.

“It can see how those patterns have developed over time and how they interact, using them to predict future returns.

“Additionally, AI technology can be used by financial services businesses, such as asset managers, to automate specific tasks, giving managers more time to spend on more significant aspects of their portfolio management.”

Paul Ronan is the chief technology officer of FE fundinfo, an investment fund data and technology firm. He has a background in finance and technology, spanning around three decades.

With around 180 technologists, his team has already been using machine learning for a few years to handle large amounts of data, with more plans in motion to use AI to help clients.

Paul says that with ChatGPT, you need to use caution in implementing it, but it does have some obvious benefits.

He says: “You can use ChatGPT’s ability to learn naturally. It can help you understand people’s needs and direct them to the right resources.

“This is especially useful in businesses where it can be hard to find the information or support needed to do a job well.”

Humans don’t like to read lots of information, and that could be particularly useful in the finance space, especially in an area like compliance.

Paul adds: “Some financial and compliance-related documents are often lengthy.

“By combining machine learning and chat features, AI could go through these sections of these documents and highlight important information.

“AI can act as a super assistant, providing frontline support and simplifying text processing.”

Streamlining communication

Because it replies like a human, you may already have seen how ChatGPT could streamline internal and external communication by providing quick and accurate responses to enquiries.

Internally, it can assist with team communication and collaboration by providing instant feedback and suggestions.

With some development work and the ChatGPT API, you could create a chatbot to help with customer enquiries, providing 24/7 availability and quick response times.

ChatGPT could save you time and resources by automating responses to common enquiries, freeing staff to focus on more complex issues.

Daniel Kroytor, founder and director at TailoredPay, said: “Many companies are implementing AI chatbots to power communications and understand the customer experience.

“Doing so allows teams to use advanced sentiment analysis to spot gaps and train chatbots to solve them.

“The fact that AI chatbots work 24/7 makes it easier and more accessible for consumers to engage when they need it most.”

What it might mean for your people

ChatGPT could change how your finance team works (you might be seeing this already):

1. It could encourage collaboration

Firstly, it can encourage collaboration among team members by providing AI-driven insights that foster discussion and collaboration.

By identifying patterns and trends in data, ChatGPT can provide valuable insights that inspire new ideas and encourage team members to work together towards shared goals.

You could improve communication and collaboration, leading to better decision-making and performance.

2. It might require upskilling and reskilling

Tools such as ChatGPT could encourage upskilling and reskilling among financial professionals.

With the increasing use of AI-powered tools in financial management, it will be essential for financial professionals to adapt and learn new skills to work alongside these tools, and you might need to provide training and support.

3. It could enhance strategic thinking

By automating routine tasks such as generating reports and handling customer inquiries, ChatGPT could free up staff to focus on more strategic tasks.

The limitations and drawbacks of ChatGPT

However, it’s worth noting that ChatGPT has several potential yet important limitations.

Limited domain knowledge, inaccuracies, and a lack of context understanding

Firstly, ChatGPT’s domain knowledge may be limited compared to specialised financial tools.

ChatGPT may not have access to the latest information or in-depth expertise in a particular area.

While it can analyse vast amounts of data and provide valuable insights, it may not have the same precision or accuracy as dedicated financial management software.

ChatGPT may also provide misleading information or lose track of context in conversations, leading to misunderstandings and incorrect replies.

Your finance team can’t afford mistakes of this nature, as you depend on accurate and reliable information to make critical decisions.

Paul Ronan says: “ChatGPT can hallucinate and make mistakes, which is concerning, especially in the financial sector where incorrect advice could have serious consequences.

“We expect these bots to be accurate, but they’re still learning, so there’s a challenge in ensuring reliability.”

Integration with existing systems

ChatGPT may not be fully compatible with existing systems, requiring additional resources and time to integrate effectively.

A better option is customisable financial management software that meets the specific needs of your business, meeting industry-specific requirements and compliance regulations.

Before adopting them fully into your operations, you must carefully evaluate the benefits and limitations of AI-powered tools such as ChatGPT.

Security

Feeding financial data into ChatGPT or any other AI model could pose a security risk. Financial data is often sensitive and confidential.

Criminals could use it for fraudulent activities or identity theft if it falls into the wrong hands.

Minimise security risks by encrypting financial data before feeding it into an AI model. Encryption is a process that converts the data into code that unauthorised individuals can’t easily read.

Additionally, it’s important to use secure communication channels when transmitting the data to the AI model.

It’s also important to have strict access controls and authorisation mechanisms to prevent unauthorised access to financial data.

You can ensure only authorised personnel have access to the data and that you grant access on a need-to-know basis.

Cybercriminals’ use of ChatGPT

You should be aware of cybercriminals using ChatGPT to attack you, says Jimmy Fong, chief commercial officer of fraud detection company Seon.

Jimmy says: “AI is getting more complex. We are seeing how this develops and what cybercriminals use as attack vectors.

“For example, we see the use of ChatGPT for application fraud and applications for online financial products.

“Bad actors can use ChatGPT to write these applications in very specific ways—whether in the style of someone over 55 or a student’s age.”

What could generative AI look like for your finance department?

Because of ChatGPT’s limitations, excitement for financial professionals could be less around what AI tools such as ChatGPT can do now and more about what it means for the future.

Let’s use an imagined example.

Emma is the CFO of a mid-sized company that’s experiencing rapid growth. Her workload, and that of her team, has increased significantly, and she needs a financial management solution to help her access and analyse financial data quickly and efficiently.

Emma decides to try out a financial management platform with AI natural language processing capabilities, so she can interact with financial data using natural language without navigating complex reports or spreadsheets.

For example, Emma can ask the platform questions like: “What are our top-selling products this quarter?” or: “How much cash do we have on hand?”

The platform will provide her with real-time insights in a conversational format.

The platform could generate customised financial reports and dashboards based on Emma’s specific requirements, allowing her to access the information she needs quickly and easily.

Thanks to AI and language processing capabilities, Emma accesses and analyses financial data quicker and more efficiently than ever, allowing her to make informed decisions and drive her company’s financial performance.

That’s just an example of what could be.

But, as Paul Ronan says: “Machine learning is already helping identify patterns and support questions in client platforms.

“As we gain confidence in AI, it could provide quick pivots on views and support advice for various fields like finance, tax, and compliance. With controlled narrative and clearer communication, AI can be a powerful tool for assisting professionals and building campaigns.”

Paul says large learning models such as ChatGPT have around 175 billion parameters (adjustable values that help the model make predictions).

This means they learn from data and adapt to find patterns and relationships. More parameters make the model more complex and capable but require more training resources and time.

These huge parameter numbers mean ChatGPT is a generalist able to answer anything, but it’s not a specialised expert in a particular area.

Paul states: “To have confidence in AI’s domain expertise in areas such as finance, we must move towards sparse expertise models that are easier and cheaper to train.

“These models will be specialised, allowing for mass adoption and increased confidence in the quality of their responses.

“The focus should be on creating expert systems specifically trained for industries like finance, pharmaceuticals, or others, rather than general models that are experts in every field”

OpenAI’s API is open for business

You may have looked at getting access to OpenAI’s API as a cloud-based service. You can see the implications for finance, with the world already accessing AI language models.

You might see potential in using OpenAI’s API to create applications that automate customer service tasks, such as answering common enquiries, verifying customer identities, and assisting with financial transactions.

Or you could look at OpenAI’s language models for fraud detection and prevention, risk analysis, and investment decision-making.

OpenAI’s API provides developers with tools and resources to create customised language models tailored to your business’s needs.

Emmanuel Methivier, business program director, member of global digital catalysts at Axway, says: “The API is designed to be highly customisable and tailored to specific use cases.

“This means that developers can create applications that leverage the API’s advanced capabilities in various contexts, from chatbots to content creation tools.

“Overall, ChatGPT’s combination of GPT’s advanced language capabilities, the API’s flexibility and customisability, and its broad range of language-related functionalities, make it a truly revolutionary tool in artificial intelligence.”

Matt Hammond is a software architect and founder of Talk Think Do, which innovates, maintains and modifies cloud-native applications for businesses.

His company is part of Microsoft’s Open AI Service, which means it can access OpenAI models with added enterprise benefits.

He says: “While there are challenges in applying AI in real organisations, particularly due to compliance concerns, we see this as an exciting strategic push to use our expertise in building big complex systems which can integrate AI technology.”

“By partnering with a company like Microsoft, we can use additional features on top of OpenAI’s services, such as controlling the regions where data is held. This is crucial for businesses subject to GDPR.

“Additionally, the stronger security measures and access controls make it easier to identify users and filter out unwanted enquiries.”

The future of AI in financial management

Although it’s very early days when it comes to the enterprise, AI has the potential to play a significant role in the future of financial management.

Natural language processing

You could use natural language processing (NLP) to develop conversational interfaces that allow users to interact with financial data and make decisions based on insights generated by the model.

Employees could easily access and understand financial data, even if they don’t have a background in finance.

It may be that generative AI for finance comes in a different form than ChatGPT.

For instance, Bloomberg recently revealed work on a new AI model called BloombergGPT, specifically designed to take on NLP tasks within the financial industry.

The model was trained on a large dataset of English financial documents and validated on existing finance-specific NLP benchmarks, outperforming existing models of a similar size.

Predictive analytics

Machine learning algorithms are already used to analyse financial data and generate predictive insights.

You could train an AI model to predict future cash flows, identify potential risks and opportunities, and recommend strategies for optimising financial performance.

Mark Troester is the vice president of strategy at software development firm Progress.

With AI prediction, he says: “To deliver truly impactful business outcomes, you need to deploy machine learning capabilities that use data over time to iteratively train your models and improve the accuracy and quality of the output.

“Focus on deploying AI technology solutions that are insightful, actionable and valuable to you.”

Automation

AI can automate routine financial management tasks, such as data entry and reconciliation.

You can reduce the time and resources required for manual financial management tasks, allowing your finance team to focus on higher-value activities.

Dan Miller, EVP at Sage Intacct, tells Diginomica that finance teams in a time of non-stop volatility are consistently searching for tools to become more productive.

Very often, the answer is automation.

He says: “What finance teams want from us, through our partner ecosystem, as well as from our direct team, is counsel on what else they can do to be more efficient. They look to see how to squeeze more out of their work.

“If they’re not using planning today, how do they plan more effectively? If they are not using analytics in a big way, they’ll want more visibility. They want help, and that means further automation.”

Risk management

AI-powered algorithms can analyse financial data and identify patterns and anomalies.

You can flag unusual transaction types as potential fraud indicators, such as large cash withdrawals or wire transfers to unfamiliar accounts.

AI could also monitor employee behaviour, such as changes in spending patterns or excessive expense claims, which may indicate fraudulent activity.

Jimmy Wong says: “ChatGPT should be a valuable tool to help fight against cybercriminals who are also wielding it against the banking and finance industry.

“AI models are already implemented across many risk management software suites, generating fraud scores for users and their behaviour on domains, but this is a narrow scope compared to the security holes you could potentially plug.”

In financial management, Sage Intacct uses General Ledger Outlier Detection, advanced algorithms and machine learning that can automatically identify and flag unusual transactions within your financial data.

By analysing historical patterns and trends, this system helps streamline financial management and enhance accuracy, enabling finance teams to spot potential errors or anomalies easily.

As a result, you can maintain better financial control, improve audit readiness, and increase overall operational efficiency.

Dan Miller says: We’ve seen significant adoption of our general ledger outlier detection. We released some new capabilities into that, which can highlight many anomalies.

“Our account payable automation solution also went to general availability within our February 2023 release, which can pre-fill vendor information based on past behaviours.”

Personalisation

You could use AI to develop personalised financial management solutions tailored to individual business needs.

You could train an AI model to provide customised financial reports and dashboards to meet each business’s specific reporting demands.

Overall, the future possibilities of AI in financial management are exciting, with the potential to improve processes and provide CFOs with greater insights into their financial performance.

The importance of AI trust and ethics

With AI increasingly being used to automate financial processes and provide insights into financial data, trust is essential.

Financial decision-making can involve risk, and you must trust that the AI systems you use are making accurate and unbiased decisions based on reliable data.

You must also be able to trust that AI systems are secure and that you protect financial data from fraud or other malicious activities.

Building trust is critical to use AI in financial management effectively.

Dan Miller says of Sage’s treatment of data: “We very much believe that we are a trusted adviser, a trusted processor of our customers’ data. It’s not our data; it’s our customers’ data.

“We build technology to use that data for the customer’s benefit. That is a very different way of thinking than some of the sensational things you hear about in the media, where people are figuring out how they use data for their benefit, as opposed to how to use it for the customers’ benefit.

“How we’ve constructed the Sage Intacct license agreement is important. How do we protect data, anonymising the data when appropriate, when we’re looking at building models?

“Where it’s not anonymised because it’s specific, we’re looking at how we make sure that customers give us specific rights to be able to do that on their behalf, so they can see that benefit come to fruition.”

There are also ethical concerns that businesses must manage.

Paul Ronan says: “The issue with AI models like GPT is that they act as a ‘black box’, meaning it’s unclear why they make certain suggestions.

“This raises ethical concerns, especially when we can’t debug or reason about the conversation like we can with humans. Compliance and legal teams worry about using such models in frontline applications.”

Dr Kurt Rohloff, co-founder and CTO of Duality Technologies, warns that AI tools may not always provide fair responses, especially as generative AI’s responses and accuracy depend partly on the completeness and quality of data.

He says: “If there are any implicit biases in data collection fed to the generative AI systems, then these biases will propagate into the output of the generative AI systems.

“Unfortunately, data used in training AI systems is often lacking and low-quality for historically underrepresented minority groups, and generative AI might unintentionally reinforce historical, social biases.

“Part of the pursuit of responsible AI includes efforts to expand the diversity, fairness and completeness of the data that feed it, in a way that protects personal data privacy.”

Growing businesses need dedicated financial technology tools

While ChatGPT has its advantages, CFOs and businesses with complex financial needs will benefit more from dedicated and trusted financial management software now and in the future rather than unproven software.

Financial management software is specifically designed to handle complex financial transactions and provide advanced accounting functionalities that may not be available in a language model such as ChatGPT.

Here are a few reasons why dedicated financial management software is better suited for finance teams in growing businesses.

Customisation

It can be customised to meet the specific needs of a growing business, including industry-specific requirements and compliance regulations.

Integration

It can integrate with other business applications such as customer relationship management (CRM), enterprise resource planning (ERP), and payroll systems, allowing for a seamless data flow between departments.

Scalability

As businesses grow, they need a financial management system that can grow with them.

Look for software that can scale your business and handle increased transaction volumes, number of users, and complex accounting requirements.

Security

Financial data is sensitive. Financial management software should have robust security measures in place, including encryption and data backup, to ensure the safety of financial data.

Reporting

You’ll want advanced reporting capabilities, allowing businesses to generate customised financial reports, track key performance indicators (KPIs), and make informed decisions based on real-time data.

Integrate AI holistically

AI and automation shouldn’t be limited to just one part of financial management. Use it throughout. This way, data and processes can work together more effectively, leading to better decision-making and overall efficiency.

The use of AI in the enterprise has steadily grown, but recent advancements have made it more accessible and powerful than ever before.

This accessibility has led to significant improvements in language processing and image generation, with development accelerating at an unprecedented rate.

We are now at an inflexion point where you might seriously consider using AI to achieve your goals.

As the AI landscape evolves, you can use these advancements to deliver high-value solutions to your customers. With increased accessibility and power, AI technology represents an exciting opportunity if you want to stay ahead of the curve.

Final thoughts

As CFOs navigate the complex and ever-changing landscape of financial management, it’s critical to understand the potential benefits and limitations of AI-powered tools to make informed decisions about their implementation.

Using AI language models and dedicated financial management tools, CFOs can gain valuable insights into their financial data, streamline financial processes, and make more informed decisions.

However, it’s also important to recognise that AI-powered tools have limitations and must be used with other technologies and human expertise to achieve the best results.

With a clear understanding of AI-powered tools’ potential benefits and limitations, the finance function can head towards a more efficient, data-driven, and strategic future.