To save costs and time, HR professionals are increasingly using AI to create employment documents akin to handbooks, offer letters, and separation agreements.
Getty
EEarlier this yr, Carly Holm, CEO of executive search firm Humani, received a call from a New York-based client. One worker had filed a workplace harassment claim against one other, and the situation quickly escalated. But when Holm asked the client for a replica of his worker handbook as a part of a routine compliance review, he stumbled. “They kind of sheepishly said, ‘Okay, here it is, ChatGPT wrote it,'” Holm said ForbeS.
For Holm, who has reviewed AI-written policies for dozens of her clients, the revelation got here as no surprise. However, on this case, there was a giant problem: the manual created by the AI contained no anti-harassment policy in any respect.
“If the workplace does not have appropriate policies in place such as a zero-tolerance policy for sexual harassment, workplace violence, etc., the investigation will focus on the employer and that will have consequences,” she said. “So they realized, ‘Wow, we should have brought in professionals to get these policies right.'”
According to interviews with 4 HR and payroll service providers, corporations are increasingly turning to AI to create vital, often legally binding employment documents – from company handbooks and workplace policies to contracts like offer letters and separation agreements. And in several cases, these documents created using AI tools like ChatGPT miss vital clauses and crucial issues, putting employers at serious legal and financial risk, said human resources consultants who’ve worked directly with such corporations Forbes.
“They looked at the manual again and found a section about overtime, but it was blank.”
While some corporations hire HR experts to review their AI-generated policies before publishing, others seek skilled help when problems arise. For example, a California-based company with 200 employees used AI to create its corporate handbook, said Daniel Grace, director of worldwide HR consulting firm Iris Forbes. But after it was distributed company-wide, managers discovered that it contained neither an additional time policy nor a legal requirement requirement within the state. When employees worked several hours of additional time, the corporate was faced with an enormous and unexpected payroll bill.
“They looked at the handbook again and found a section about overtime, but it was blank,” Grace said.
Another example: In the U.K., an HR team used Microsoft Bing’s AI Copilot to create a severance agreement, Grace said. The legally binding document, which was missing key information, was then forwarded to the worker’s legal counsel.
“Their lawyer basically just threw it away and said it was useless. It really soured their own negotiations and they ended up having to pay a higher settlement amount,” Grace said.
Manuals and workplace policies are sometimes a whole lot of pages long. These comprehensive documents are each tailored to reflect the corporate culture and have in mind local laws that specifically apply to the businesses’ employees. But while AI tools can assist create a primary draft of those documents in seconds, their underlying training data sets are sometimes outdated in the case of ever-evolving labor laws, said Ben Houghton, VP of engineering at Iris. Then there are a variety of privacy concerns. If confidential company data is entered into the free version of ChatGPT and Co., it could possibly be used Train AI models or have turn into public.
Paul Cortissoz, CEO of HRSoul, a Florida-based human resources consulting firm, said Forbes Over-reliance on AI for these policies can have unpleasant consequences. He said Forbes a few distant SaaS company that hired him after an worker filed an internal criticism. The company’s handbook, created with the assistance of ChatGPT, provided for five days of paid leave as a part of its bereavement policy, but within the state where the worker lived, employees were entitled to as much as 10 days of paid leave.
“That was kind of the pain point for them, where they said, ‘Oh, you know what, maybe we need to bring in someone who really knows what they’re doing,'” Cortissoz said, adding that his company was asked to do it have been repeating a series of AI-generated worker handbooks since then.
While AI is used for a spread of HR tasks – writing job descriptions, sourcing talent, screening and interviewing job candidates, awarding bonuses – it just isn’t yet sufficiently integrated to create precise company policies on what the Companies that use them to cope with serious problems expose legal and financial repercussions.
“Let’s face it, people use artificial intelligence tools to speed things up… and they really shouldn’t… These things have teething problems,” Grace said. “You create these documents. Nobody proofreads it, they publish it, and it causes more problems than if they had taken the time to successfully create the manual from the start.”
MORE FROM FORBES
