Chat GPT and other well-trained Artificial Intelligence and Machine Learning systems have the potential to completely disrupt the accounting industry in the future. We'll go over what Chat GPT is and how you can start using it right away to make your life easier.
You've probably heard of Chat GPT by now. The internet is basically exploding with people freaking out, and for a good reason.
The Open AI firm created Chat GPT (Generative Pre-trained Transformer), a chat bot. It was trained on massive datasets containing billions of words from books, articles, and websites as recently as 2021. The dataset's exact size is proprietary and not disclosed, but it is thought to be one of the most sizeable AI training datasets used to date. You can use the software to ask questions in a casual language and receive a written response that the bot generates on demand based on the data it has been instructed with.
The coolest thing about it is that you can ask the bot specific accounting or business-related questions and usually get a useful response.
Here's an example of a Chat GPT accounting question: "What is the journal entry for recording the sale of a $10,000 car?"
If you typed that in the search section of Chat GPT, you’d be impressed with the answer, context, and explanation it provides.
Its response is based on an analysis of thousands to millions of data points instead of reiterating an article that was previously written by someone and indexed by a search engine like Google or Bing.
We put Chat GPT through its paces on much more difficult accounting and audit questions found on the CPA exam. It provided answers that were surprisingly detailed. What makes it even more intriguing is that the dataset on which Chat GPT is based is a general dataset. If a similar bot was trained using accounting, audit, and tax data, the answers would become even more detailed.
CPA Exam Sample Question:
What constitutes required supplementary information for a local government's general purpose external financial reporting?
Chat GPT Answer: A local government's required supplementary information (RSI) for general purpose outside financial reporting typically includes:
-A schedule of federal and state financial assistance expenditures: This schedule details the amount and type of financial assistance received from the federal and state governments, as well as how it was spent.
-A schedule of long-term liability changes: This schedule details changes in long-term liabilities such as bonds, notes, and other long-term obligations.
-A schedule of revenue-generating activities: This schedule details the revenues generated by the government's various revenue-generating activities, such as taxes, fees, and charges.
-A schedule of general fixed assets: This schedule lists the government's general fixed assets, which include buildings, land, and equipment.
-A debt service schedule: This schedule details the government's debt service payments, including the amount, due date, and maturity date of the debt.
-A pension and other post-employment benefits (OPEB) schedule: This schedule details the government's pension and OPEB liabilities, including the amount of the liability, the actuarial assumptions used to calculate the liability, and the plan's funded status.
It is important to note that RSI requirements may differ depending on the type of local government and the accounting standards used.
One application for accountants is to use Chat GPT as a wingman in areas where you are unfamiliar. Accounting jobs are in high demand.
Accounting data, for example, is frequently used in Power BI or Tableau dashboards, but formulas for it are rarely written. Fortunately, you can rely on Chat GPT to create formulas that you may not use very often.
You could use Chat GPT to get tips on how to structure formulas.
Chat GPT's technology could be used to answer queries that a financial analyst or accountant may have previously been asked to compile. In this case, a bot from Chat GPT could be trained to answer common questions using a company's own data set. What are the top ten selling items this month? Which products are most frequently purchased together? What caused the decrease in operating cash flow this quarter?
Microsoft and other software companies have attempted to develop similar solutions, but they are relatively simple and not nearly as natural as interacting with Chat GPT. Microsoft Power BI, for example, can already be configured to answer questions about underlying data but must be manually trained and seeded with base questions.
Tax practises add significant value to their clients by conducting legal research and providing opinions on various tax strategies. It is not a stretch to train a Chat GPT bot on a legal dataset and tax case law. Once trained, it could make tax recommendations or be integrated into an ERP to identify opportunities that a tax practise may not be aware of at first.
It can also be utilised in place of e-discovery. Businesses are frequently required to prepare documentation and record requests for everything from local sales and use tax to IRS audits. The process of gathering this data can be time intensive. An AI equipped with your internal data may be able to pull out the documents you require much more quickly.
While we believe that having an AI like Chat GPT represent you in tax court is a stretch, The National Law Review recently published an study exploring the use of GPT as a replacement for law professionals. ChatGPT Is the Courtroom's First Artificial Intelligence Lawyer? (natlawreview.com)
For the time being, it does not appear that lawyers will be completely replaced. Chat GPT is still having difficulty passing the bar exam. Craig Ball: ChatGPT Demonstrates a Subpar Law Student - EDRM However, these tools have the potential to help Tax and Legal teams prepare for cases, respond to IRS requests for information, and possibly even serve as a reference point for potential tax strategies at your company. At the very slightest, we can hope that they will reduce tax lawyers' billable rates.
It's no secret that the IRS is dealing with a massive staffing shortage that they're attempting to overcome. Long term, it is possible that these jobs will become increasingly automated. Most straightforward personal tax returns are already audited for fraud and errors by machines. IRS Agents or IRS Special Agents audit more complex returns. As AI improves, it's likely that an increasing number of tax returns will be fully audited or, at the very least, flagged for suspicious activity.
-Answers can be so high-level that they are useless. For example, when we asked why we were getting an error when using the Alteryx to SharePoint connector, some of the responses were helpful, but others included restarting your computer. It's possible that restarting your computer will help, but it's unlikely to solve the problem.
-Some answers may be incorrect but appear to be correct. We were using it to create Python scripts that appeared to run but did not compile. It then took some time to figure out why it wasn't working, which may not have been much faster than writing it from scratch.
-It takes some technical knowledge to recognise when something is incorrect. Many of the responses appear reasonable and practical. However, if you don't know Python, it may be impossible to go back and figure out why a Python script didn't run. Chat GPT is a great place to start, but you'll need your own personal depth of knowledge to ensure the results are what you need.
