Artificial Intelligence (AI) is gradually revolutionizing various industries, including the field of accounting and finance. With the emergence of AI tools and Large Language Models (LLMs) like ChatGPT, Google Bard, and BERT, professionals in these fields can benefit from enhanced capabilities and streamlined processes.
In a recent webinar sponsored by Datarails, the FP&A solution for Excel users, three distinguished finance leaders came together to discuss the impact of AI on corporate finance. They highlighted how AI technology is transforming the way finance and accounting teams work with data and make decisions.
(Skip to the end to watch the webinar recording)
The first panelist is Sloane Kolt who also leads Datarails Labs, the department that is part of the FP&A software startup that developed FP&A Genius, a Chat GPT-like chatbot for finance. With over 10 years of corporate finance experience in various industries, she brings a unique blend of financial and operational expertise to her work. She is passionate about harnessing AI applications in corporate finance to streamline processes and unearth valuable insights.
Glenn Hopper also shared his expertise. He has spent the past two decades helping startups navigate growth, scale operations, and prepare for funding and acquisition. As a seasoned finance leader in telecommunications, retail, technology, and legal sectors, Glenn envisions transforming the role of Chief Financial Officer (CFO) from being a historical reporter to a strategic forward-thinker.
Nicholas Boucher, a Finance Leader and Educator also participated in the webinar. He brought over 14 years of finance leadership experience from renowned firms like PWC and Tails across multiple countries. His expertise spans financial analysis, budgeting, business partnering, financial storytelling, excel, and audit. Known for his generosity in sharing knowledge, Nicholas has coached hundreds of finance professionals and has a significant following on LinkedIn. He has also launched the first Chat GPT for the finance training program.
The panelists emphasized several key areas where AI is making a significant difference in corporate finance. Improved financial modeling and investment banking management are among the notable benefits that AI brings to growing companies. By leveraging AI algorithms and automation, finance professionals can effectively manage complex financial models and optimize investment strategies.
Furthermore, AI is being used for demand and revenue forecasting, anomaly and error detection, decision support, and proof-of-concept (POC) revenue analysis. These applications of AI enable businesses to make more informed decisions, detect potential risks, and improve overall financial performance.
The introduction of ChatGPT, a cutting-edge AI language model, has also had a profound impact on corporate finance. This technology allows professionals to seek real-time insights, consult for expert advice, and access top-notch information in a more efficient manner.
When inquired about their usage of AI tools such as ChatGPT and others, it was found that approximately 30% of the 140 participants haven't yet tried any AI tools. Following that, around 23% reported spending less than 20 minutes to 2 hours using these tools, while 22% dedicated more than two hours. Interestingly, nearly one-fourth of the respondents already utilize AI tools on a daily basis.
Unveiling the AI Toolbox: Finance Professionals Guide to Identify the Right AI Tools
Hopper acknowledges that there are two paths to be discussed: AI at scale and individual use. Their primary focus has been on AI at scale, particularly with generative AI and drag-and-drop AI, and machine learning tools becoming more prevalent for individual use. However, they emphasize that AI is a broader field that encompasses more than just Large Language Models (LLM).
He also mentioned that they haven't personally witnessed the use of generative AI at scale today. They have had conversations with startups incorporating AI-powered chatbots into existing tools and exploring new finance-focused businesses utilizing AI. These projects are still in the early stages, but it seems on the cusp of seeing more production-level applications. As an example, Data Bricks recently acquired Mosaic ML for $1.3 billion, enabling them to offer a service allowing companies to train their own models for enterprise use.
However, when it comes to using generative AI at scale within an enterprise, Hopper admits they haven't come across a trustworthy tool yet. Nevertheless, they believe such a tool is on the horizon and eagerly anticipate its arrival.
Understanding Generative AI and Large Language Models (LLM)
Hopper explains that many people are familiar with Venn diagrams that illustrate the different levels of artificial intelligence. At the top level is the broad field of artificial intelligence itself, which involves programming computers to perform tasks typically requiring human input. Machine learning is a subset of AI that uses algorithms for classification and prediction. Within machine learning, there is a specific type called neural networks or deep learning, which employs neural networks for predictions or classifications.
Neural networks are also used in generative AI, specifically in large language models. These models have gained attention not just in language generation but also in generating images. They take inputs and create entirely new content, generating text or images based on the given prompt or input. Large language models undergo extensive training on vast amounts of data, including billions of documents, words, and parameters. It is important to note that this is just one type of AI among many others, but it currently holds significant fascination and appeal.
To simplify the difference between AI and non-AI software, Boucher specified that there lies in their ability to adapt and improve. Non-AI software is programmed to perform a specific task and remains unchanged unless the code is altered. On the other hand, AI software has the capability to continuously analyze and learn from large amounts of data, allowing it to improve itself over time.
When it comes to finance, if a software system can analyze and comprehend increasing volumes of data and evolve to provide better insights, it can be considered AI software. However, if a software program consistently reacts and responds in a predetermined manner without the ability to self-improve, it cannot be classified as AI.
In today's landscape, it can be challenging to distinguish between AI and non-AI tools, as many software sources incorporate AI into their offerings. The question arises: does continuous usage and feeding of data into the software result in improved outcomes, or does the output remain the same regardless of training or data provided?
Robotic Process Automation (RPA) has played a significant role in the automation of accounts payable processes, offering numerous benefits. The similarity of invoices and processes across various companies allowed providers to develop algorithms that self-trained on invoices, showcasing the power of training in RPA. However, as technology progresses, the use of RPA expands beyond finance.
The breakthrough moment comes with advancements in non-finance areas such as language learning, speech understanding, and machine translation. This broader conversation between humans and machines is facilitated by chat interfaces, which enable more natural communication. People now rely more on chat interfaces because they eliminate the need for coding or explicit instructions, making the interface between people and machines effortless.
Examining the Implementation of AI in the Finance Sector
Boucher further explains now that individuals can interact with AI, regardless of their coding background or not, the process of utilizing AI has become more accessible even for those without programming knowledge.
He also shared his chart that gave emphasis on the benefits of AI to various fields such as:
Productivity
Tutorial for Tools
Scenario Analysis
Create Finance Procedures
Write High- quality Emails
In relation to ChatGPT, the focus lies on productivity enhancement. In finance, where a love for mathematics is common, writing skills may not be as strong. Thus, having AI assist in improving writing abilities is a significant advantage for finance professionals.
Another application gaining traction is leveraging AI for scenario analysis. AI can perform calculations much faster than the human brain, given appropriate prompts. As algorithms are trained on data and patterns, the scope for obtaining additional insights from scenario analysis expands. Utilizing ChatGPT eliminates any excuses for inadequate writing proficiency when it comes to drafting procedures.
Hopper also emphasized the importance of automation and data-driven decision-making in the finance industry. He has written a book on using data in finance and accounting and has been focused on integrating systems, collecting data, and utilizing it for models and forecasts. The emergence of generative AI also allows language models to write code. According to Hopper, he has been working on various projects involving AI, including fine-tuning finance and accounting specialists and using language models to analyze financial statements and advocate for exploring and understanding AI technology to leverage its potential in the future.
The Transformative AI's Impact on Jobs and Job Descriptions
According to Kolt, the impact of AI on jobs and job descriptions will lead to a significant shift from manual work to strategic work. This shift will prioritize tasks that involve being present in important discussions, acting as business partners, and supporting teams in decision-making processes. Job descriptions will also reflect this change, placing greater emphasis on soft skills, collaboration, team elevation, and understanding of finance principles. This transformation will redefine how professionals work with others, comprehend opportunity costs, and make decisions for their organizations.
Hopper believes that AI is significantly impacting various job sectors, including accounting and finance. The discussion emphasizes how automation is prevalent in daily functions like expense reports and accounts payable/receivable. While fewer people are needed for data entry, the increasing data volume requires new approaches to roles in the field.
He also stresses the importance of soft skills and human interaction, along with the ability to effectively leverage data to create value. He also highlighted that what AI does is not only with the democratization of data but also the growing accessibility of data science tools for data analysis. Additionally, they highlight the need for analytical skills and the ability to ask relevant questions, as traditional knowledge of credits and debits is being supplemented by these tools.
According to Hopper, “We must actively embrace and engage with this technology, rather than passively receiving its benefits. We have to own it. We have to understand it and drive its adoption and know how to use it as a tool if we want to be able to kind of ride this wave and keep up with it.”
Key Considerations for Finance Teams Before Implementing AI
The experts believe that it's important to consider the responses received so far on this topic. Trusting platforms like Bard or ChatGPT blindly for accurate responses every time is not advisable. As finance and accounting professionals are naturally risk-averse, and it's crucial to understand the barriers to implementing AI and how to overcome them.
One concern is the phenomenon of hallucination, where language models can confidently provide inaccurate information, even making up references. Due to the lack of corporate governance policies in this emerging field, Hopper understands there can be undesirable outcomes. Additionally, there may be pushback from leadership, which is understandable given the current uncertainties.
However, it's essential to strike a balance between skepticism and embracing the future. To prepare for using AI in an enterprise setting, it's crucial to have a certain level of data maturity. Without sufficient data maturity, implementing AI with proprietary information may not be feasible. Data security and management are also significant considerations when working with AI.
It's worth reiterating the importance of not uploading proprietary information into platforms like ChatGPT to ensure data security. Instead, the focus should be on addressing skill gaps and understanding the basics of AI, machine learning, and language models. It's necessary to assess how these technologies can integrate with existing systems and conduct a thorough security assessment of data usage.
Boucher also emphasized the importance of viewing AI as a new tool. Just like any new tool, it requires learning and adaptation. According to him, transparency is key; we do not hide the fact that we are using ChatGPT, as it is essential for your boss and team members to be aware.
It is crucial for your team to inform you when they use ChatGPT. As this technology is still new, the quality of its outputs is uncertain. Learning together as a team allows the company to discern when it is appropriate to leverage ChatGPT or other AI tools and when their usage may lead to errors or flaws.
However, it's important to note that the way we communicate with these machines may differ from the past. We must learn how to ask and talk to machines effectively. It's crucial to understand that AI models are trained with probabilities. For example, if you ask a machine what 1+1 equals, it will provide the most probable answer, which might be 1.9999 or 2.0001. While this may seem incorrect to a finance person, it's important to recognize that you are not working with a spreadsheet but an algorithm designed to provide the most probable response.
The way we communicate may be new, but it is here to stay. Learning how to effectively use ChatGPT, including proper prompting techniques, will give you a competitive advantage in the job market. AI is a tool that is rapidly evolving, and mastering its usage will allow you to work more efficiently than others.
Maximizing the Potential of AI for Accounting and Finance Professionals
The individual believes that the fear of the unknown can be quite overwhelming for some people, especially when it comes to AI. Kolt mentioned that AI, such as ChatGPT and Bard, actually makes tasks easier and more accessible. They emphasize the importance of understanding how to work with these tools and explore different AI models available.
AI extends beyond generative language models and includes predictive analytics, which can enhance forecasting and data analysis in finance and accounting. Professionals in the finance industry are encouraged to explore these tools as they offer valuable insights and are free to use. While acknowledging the associated risks, experts emphasize the significant potential of these tools in terms of time-saving and increased productivity.
As individuals gain expertise and explore the capabilities of AI, they will discover both its limitations and possibilities. There is no need to be apprehensive, as starting with AI is not as daunting or challenging as it may seem initially.
On the other hand, Hopper’s thought it is important for people to recognize the significance of the current situation, especially now that it is receiving more media attention and attracting a wider audience. We are currently in a period of relative calm before a major transformation takes place.
Many individuals are still new to the concept of AI. While waiting for AI solutions to become widely available, it presents an excellent opportunity to start exploring and grasping the fundamentals. Understanding what AI is, how machine learning works, and how algorithms are trained will be crucial for navigating the future. This knowledge will prevent us from being caught off guard, similar to how we prepared ourselves when the internet and smartphones were introduced. We are currently in a preparation phase, akin to being in the on-deck circle in baseball, getting ready for the upcoming changes. Finance and accounting teams should prioritize their focus on preparing for what lies ahead.
Strategies for Effective Prompting of AI
When asked about the most effective way to prompt an AI or LLM, Bucher gave a quick demonstration of a framework called CSI which is the abbreviation for Context-Specific Instructions.
The framework consists of the necessary components for an effective prompt. It should include the context, such as being an accountant, providing relevant information about the client's overdue payment in 2 months, and specifying the desired instruction. They then ask for assistance in drafting a communication to their client, using ChatGPT. This straightforward prompt will be used to showcase how the framework can be applied.
Boucher also added the second strategy which he called CSI + FBI which means Context Specific Instructions + Format Blueprint Identity where the blueprint could include the style, tone, and examples. The format could be either a table, an e-mail, a procedure, or a memo like every format you can think about a code.
In this context, Kolt also added a few additional points to remember when using prompt engineering:
It's better to instruct the LLMs on what to do rather than what not to do. Simply stating "do not" followed by an action often proves ineffective. Being very specific and avoiding vague language is crucial in crafting prompts.
Separating instructions from the context being provided helps the LLM understand better. Providing clear instructions separately, it enhances the LLM's comprehension and allows for more accurate responses.
Utilizing examples can be a powerful tool, especially when trying to create a large corpus of data. Providing examples gives the LLM something concrete to work with and enables it to perform statistical analyses based on the provided examples, allowing it to generate responses in a similar style or manner.
Unlocking Productivity with AI
Hopper shares their perspective on AI productivity, adding to the previous discussion. He shared a mantra borrowed from Clifford Stoll, a renowned computer scientist and author. Hopper finds Stoll's quote inspiring and believes it drives the value they can bring.
"Data is not information. Information is not knowledge. Knowledge is not understanding and understanding is not wisdom." He further explained that this quote resonates with them and emphasizes the importance of progressing from raw data to wisdom. Hopper believes that AI plays a crucial role in facilitating this journey, making it easier for individuals to reach a state of wisdom.
When asked what can they suggest as areas for improvement or focus for finance professionals is AI takes on more tasks in the profession, Hopper replied that with the abundance of available data and the increasing availability of AI models, finance professionals should focus on enhancing their understanding of statistics, model building, and basic machine learning algorithms. They emphasize that having a solid foundation in statistics is crucial for effectively utilizing machine learning tools, as machine learning relies heavily on statistical knowledge and probabilities. They view this as an essential area for finance professionals to refresh and strengthen their skills, as it will enable them to leverage AI tools more effectively in their work.
FINAL WORDS
The rapid advancement of AI technology has opened up new breakthroughs and challenges for finance professionals. By understanding the various AI tools available, they can harness the powers to streamline processes and enhance decision-making. The implementation of AI has been proven to be transformative, especially with AI algorithms and machine learning models revolutionizing tasks traditionally performed by humans. While there are concerns about job displacement, it is important to recognize that AI also creates new roles and opportunities for finance professionals.
Strategies for effective prompting of AI, such as training models with high-quality data and providing clear guidance, can significantly unlock productivity at work. Finance professionals can focus on higher-value tasks and strategic decision-making, leading to greater efficiency and improved results. AI has the potential to revolutionize the finance industry, offering numerous benefits to accounting and finance professionals. By embracing and understanding generative AI and LLM, and considering key factors before implementation, finance teams can unlock the full potential of AI and drive transformative change in their organizations.