Demystifying the uses of generative artificial intelligence (AI) tools, particularly ChatGPT, to improve work-life balance was the focus of the first session of the Virginia Society of CPAs (VSSCPA)’s Women’s Leadership series on March 8, presented in conjunction with the NYSSCPA.

The VSSCPA took the occasion of International Women’s Day to launch the four-part series, starting with this tech session webinar, to discuss what generative AI can and cannot do, as well as safe ways to explore the technology to one’s advantage.

Leading off the session was Ashley Francis, a CPA who specializes in helping ultra-high net worth clients with complex tax and estate compliance and planning. The founder of Kitchen Table Automations, a newsletter and series of classes concerning AI and automation, Francis self-deprecatingly said that she sometimes struggles with such office necessities as a printer before her co-presenter, Byron Patrick, interrupted to remind the attendees that she has spoken about technology at major conferences. Patrick is a CPA who serves as vice president of client success at the B3 Method Institute, which helps leaders design accounting practices.

Patrick acknowledged that the flood of information regarding AI can be overwhelming by engaging in some hyperbole. “Last week, I told Ashley, 'I'm done,'” he said. ”I'm not listening to any more of the AI news. I'm not reading any more of the AI articles because it is going so fast that I was approaching paralysis. ...I can no longer even take action on this stuff because tomorrow [there is] something new.”

Francis agreed, then exhorted the attendees to “dip [their] toes in the water when it comes to this stuff, and you will find value and [you will] benefit from it, so it doesn't have to be so scary, even though it can really seem” to be that way. “And I also don't recommend reading any of the books that will tell you that this is the end of the world, because that won't help,” she added.

The pair suggested starting out by understanding prompts to help to write emails. They went so far as to define what a prompt is, with Francis explaining that it is “having a conversation with this technology. So, when I say prompt, I mean you talk to it, you're saying something to it and it's going to say something back.”

To illustrate what is meant by a conversation in this context, they used the analogy of a tuna sandwich.

“If I walk up to you and say ‘tuna sandwich,’ you are probably going to guess that I want a tuna sandwich, but you might not know that I'm allergic to tuna or I hate tuna sandwiches or I'm there to complain about too many tuna sandwiches in the world,” said Francis. So, you won’t really be able to help me.”

“So, that's why when we talk about prompts, we're talking about conversations and [choosing] our words to our long distance imaginary friends in our computer,” she continued.

From this, she developed a simple framework for starting to use prompts in generative AI: “I am,” “you are, and “I need.” She also said that she sometimes “catfishes”—telling AI something that she is not in order to get a specific result or specific information.

Patrick picked up on that with a specific use for the framework: writing emails. 

“I'm an accountant in a bookkeeping firm, and I need to send an e-mail to my client, a staffing business, to ask them to [send] a list of things they need to provide for the year-end close of books.” he said as an example.

“Can you please draft an e-mail in a warm but professional tone?” he would ask the chat bot, instructing it to include a bulleted list of items, make it fewer than 200 words, and request a due date of one week from March 8. “There's a lot of instruction in there, but we're trying to be very specific,” he said. This type of letter can be refined, he said, by giving the bot more instruction, such as updating the list to reflect more specific needs.

“This doesn't have to be completely right,” he said. “This could be 85 percent of the work, and I just need to copy and paste it [and] modify a few things. I don't know about you [but], if something else can write 85 percent of an e-mail for me, my day has been changed.”

Francis then cautioned that, “if you try to get 100 percent with any of these Gen AI tools, you will slowly lose your mind and it's not worth it. This is supposed to help us, not make us lose our mind.”

Francis and Patrick built on that example by offering examples of other professional and personal uses for generative AI tools, especially ChatGPT. One, which Francis uses in her business, is the vacation planner bot. It is used to plan around estimated preparation and year-end needs for certain types of clients by listing all of the filings and project dates, taking into account planned vacations and office closures. Based on that information, the bot can be asked to draft an e-mail template, the tone, length and frequency of which can be specified.

The pair concluded with a summary of best practices, and some dos and don’ts.

“Do learn how to use it,” said Patrick. “Gen AI is a judgment-free zone. You really need to experiment with it.”

He also recommended creating an AI policy. “You should set some rules because human nature is: If there are no rules, there are no rules,” he said. “So you should set some clear rules and expectations ... with respect to the work content and the way that things can be used.”

The panelists also urged the attendees to be aware of some of the nefarious uses of the technology.

“College students are definitely using it. They are lying to their professors,” said Francis. She urged firms to strictly enforce their rules to prevent new hires from engaging in such behavior, noting that high schools are now teaching students how to use AI, so it's important to make clear what the proper parameters are for its use.

“We're all CPAs here [and, as such, are] hyperparanoid about client data,” said Francis. “Do not tell it sensitive data, no client data, no entity or business secrets.” 

She called the technology an “intern parrot,” explaining that “you would not put your intern parrot in front of a client and [say], "Hey, so I'm just going to leave you two here alone and I'll just come back  and find out how things go. It will make stuff up. Don't use it for complex tax research.'”

Still, Patrick ended the session on a positive note. “Take this as an opportunity to learn how to communicate better,” he said. “It will not only improve the output from generative AI, but it might even improve the way that you work with your team.”

Future sessions of the Women's Leadership series will be presented throughout the year:

 • April 26, 2024, Part 2: Women’s Mental Health Day – Fighting Burnout
 • Aug. 26, 2024, Part 3: Women’s Equality Day
 • Nov.19, 2024, Part 4: Women’s Entrepreneurship Day