Future-forward organizations are learning to make the most of generative artificial intelligence (AI) tools, like ChatGPT, to improve productivity and capacity in the workplace.
Released to the public by OpenAI in November 2022, ChatGPT — a chatbot version of a generative pre-trained transformer — became the fastest-growing consumer application in history, reaching 100 million active users a month in just two months’ time.
As a nationally recognized thought leader on AI, Byron Patrick, CPA, CITP, has been adapting and evolving solutions for ChatGPT to share with fellow CPAs. Patrick’s first application of ChatGPT was to enhance the use of Excel, using AI to create formulas for problem-solving. He discovered that ChatGPT could generate complex formulas he never knew existed to perform calculations. The result: time-saving solutions that met the problem at hand.
The technology and innovators behind generative AI
As a category of computer science, machine learning is a form of AI that consumes data and uses it to predict outcomes. Bound by the data inputs to produce outcomes, generative AI moves beyond machine learning to “take action” and create new content from the input provided.
Generative AI can, for example, serve as a thought partner, “participating” in brainstorming conversations or as a practice buddy for framing and practicing difficult conversations with humans.
Applications such as Alexa and Siri respond to trigger words without putting them into context. By contrast, large language models make generative AI applications powerful, Patrick explained.
The large language model can use deep learning to understand language and put it into context to create new responses. Generative AI applications can be trained by using information inputs to “learn.”
Major innovators in generative AI include companies like OpenAI, Microsoft, Google, Anthropic, Meta, X, AWS, IBM, and Oracle.
Protecting your data and privacy
Generative AI learns how to interact with you based on your preferences and style, and this can be a safety concern.
Is your data private when using generative AI? Yes, it can be when using OpenAI and Microsoft.
The data is not shared with third parties when using a paid version of these generative AI apps.
In paid versions of a generative AI app, the user can turn off the “training” feature, disabling the model from learning patterns and relationships from data and algorithms. This stops the creation of new content based on this learning.
Patrick recommends turning off the training function in a paid generative AI app because it may include components the user does not understand. If you use a free version, your data inputs may be used to train the app.
Government entities have special needs related to security, privacy, and compliance. For government use, OpenAI services are available on Azure Government. Copilot is available on Microsoft Government Cloud. ChatGPT Gov is available on the government cloud ChatGPT.
Generative AI use cases and tools for CPAs
Professional uses of generative AI models include:
Email editing
Turning screenshots into Excel formulas
Creating or reverse-engineering macros and complex spreadsheets
Simplifying financial analysis and reporting
Explaining complex regulations, like tax code, in simpler terms
Generative AI tools can also be used to streamline audit and accounting processes in different ways.
A tool called Deep Research can perform in-depth, multistep research from the internet, taking about 20 minutes to “think” about the topic posed and return an answer.
Audit GenAI can evaluate disclosures in financial statements and make suggestions for improvement. As a result, the quality of audit planning and audit disclosures is improved.
Accounting GenAI can assist with financial analysis. Data for the general ledger can be fed into a generative AI tool and a report produced, speeding up the process by doing a first pass of the general ledger produced or by providing revisions to a report that has been produced.
Although AI can provide valuable support, and holds promise for the future, the human element remains critical. Generative AI tools are as effective as the human judgment that determines when, where, and how to apply them.