ChatGPT Shows the Magic of AI — But Beware the Ghost in the Machine: 3 Caveats for Employers

You may have heard about ChatGPT over the last few weeks, a new chatbot with an uncanny ability to mimic human dialogue and decision making. This artificial intelligence (AI) — which was developed by OpenAI and is currently free to access — has a simple premise: You “talk” to a chatbot that can assist you with any number of tasks. For example, you can ask it to explain a subject, write an essay or article, run a calculation, help code in Python, or simply follow a conversation. The technology then pulls knowledge from its dataset and produces a response that looks surprisingly accurate — at least at first glance. The release of ChatGPT has led many employers (and employees) to become excited about the ways they can use this service to streamline their processes. But questions remain about the reliability of the data it spits back at you. Notably, ChatGPT is currently in its “initial research preview” — which means the technology is still being tested and researched before being fully released for commercial use. Given these limitations, you should carefully consider the issues that may arise before using such technology in the workplace. Here are three caveats for employers to keep in mind as this technology continues to develop:

Recognize ChatGPT’s Limitations

Because of its ease of use and capabilities — and despite being in limited beta — ChatGPT quickly made headlines for its ability to automate tasks and produce usable, written material. It has come as no surprise that business professionals spanning many disciplines have already experimented with the new technology, whether they’re in marketing, human resources, accounting, software engineering, legal departments, education, or other fields.

As exciting as the technology may be, however, ChatGPT has limitations — and employers should be cautious about relying on it for serious, consequential matters. Indeed, its developers have themselves cautioned against doing so.

Sam Altman, the CEO and co-founder of Open AI, tweeted on December 10: “ChatGPT is incredibly limited, but good enough at some things to create a misleading impression of greatness. It’s a mistake to rely on it for anything important right now. It’s a preview of progress; we have lots of work to do on robustness and truthfulness.” Greg Brockman, the company’s President and co-founder, responded: “We believe in shipping early & often, with the hope of learning how to make a really useful and reliable AI through real-world experience and feedback. Correspondingly important to realize we're not there yet — ChatGPT not yet ready to be relied on for anything important!”

Further, ChatGPT was trained on a dataset (a wide range of texts, including books, websites, and articles from many different sources) that was current up until 2021, and its knowledge is based only on information that was available at that time. It cannot browse the internet or access new information, so its answers – particularly about developments in labor and employment law – might not be up to date.

Moreover, questions remain about ChatGPT’s ability to assess the credibility and accuracy of the information it provides you. The bot itself encourages users to independently verify any information they receive from it.

Given all of these concerns, here are the three initial caveats employers should consider before plunging in with both feet.

1. Use Caution When Creating Employment Documents

Although ChatGPT may seem like a great tool to help you draft employment documents, you should consider using it only as a starting point and consult with your employment attorney on its effectiveness — because problems might otherwise arise.

For example, if you ask it to draft an employee handbook policy, it may give you a document that seems competent on its face but actually lacks important nuances. An anti-discrimination policy may cover the basics but likely won’t account for local variances or company culture or exhibit the human understanding needed to navigate discrimination issues.

These problems will also likely arise when asking ChatGPT to draft other employee-related documents. The final product may be a mile wide but only an inch deep. Similarly, the chatbot may generate a competent-sounding confidentiality agreement that fails to cover your company’s specific environment, and hasn’t been battle-tested by the applicable courts.

All of these warnings lead to the inevitable conclusion: There is no guarantee that the documents you ask the chatbot to draft will be legally enforceable or provide you with the protection your business needs.

2. Consider Confidentiality and Privacy Issues

When logging into ChatGPT, the tool expressly warns users that: “Conversations may be reviewed by our AI trainers to improve our systems,” and cautions, “Please don’t share any sensitive information in your conversations.” Considering the popularity of ChatGPT and the lure of using the tool to “cheat” on work assignments, employers are rightly concerned that their employees might ignore those warnings and unwittingly feed it confidential business information or trade secrets.  Additionally, these potential disclosures could expose your business to liability under a rising tide of privacy laws, such as the California Consumer Privacy Act (CCPA).

As a result, employers are wise to consider updating and redistributing any applicable confidentiality and trade secret policies to ensure they cover the use of any third-party A.I. tools, and should train employees as necessary. Employers might also consider blocking access to such tools on any work devices. Of course, you may also want to explain your rationale for restricting access, so employees understand the limitations to the technology and why you have taken this step.

3. Be Sure to Check Your Calculations

ChatGPT is also capable of running numerous calculations. For example, it can calculate payroll deductions for state and federal income tax, Social Security, and Medicare. The technology can also produce Excel formulas, calculate wages based on hours worked, consider overtime rates, and more.

But relying on ChatGPT for these calculations can be dangerous, as there is no guarantee of accuracy. Also, the chatbot may not consider nuances in the law. Will it incorporate state and federal requirements, as well as local minimum wage rules? As every math teacher reminded us: Make sure you check the work.

Getting these calculations wrong can be costly. For example, miscalculating employee paychecks can be a huge source of liability under numerous laws. And failing to properly pay employees and withhold deductions can prompt a visit from the Department of Labor, IRS, or other government agencies. So, before you use ChatGPT as a calculator, make sure that any calculations are accurate and based on correct inputs.

ChatGPT is the Tip of the AI Iceberg

New technological advancements typically prompt a discussion about job automation and the other numerous ways in which AI can transform your operations. ChatGPT is no different. Expect this latest exciting leap into the future to prompt excitement, concern, discussion, and questions from all corners of your workplace, and be prepared to address them in a thoughtful and considered manner.

As AI develops and improves, you should begin to consider what types of work can be automated, if you haven’t already. For example, can AI eliminate certain “grunt work” so your employees can work on more interesting “bigger picture” projects? Technology like ChatGPT may ultimately be used to quickly perform complex but undesirable tasks, such as generating analytical reports that would otherwise take hours or days for an employee to create.

But even if you can use AI to automate work, should you? You’ll have to carefully weigh the risks and benefits at the current time. And at least for now, most tasks will still require human oversight, which may limit your ability to effectively leverage them for meaningful work in the near future. But you should keep close watch on developments in this area, because a future where AI plays a key role in your company’s work is right around the corner. 

Conclusion

We will continue to monitor developments on ChatGPT and artificial intelligence, so make sure you are subscribed to Fisher Phillips’ Insight System to keep up with the most up-to-date information. Please contact the authors of this Insight or your Fisher Phillips attorney should you have any questions.