A.I. is not going away.  Its use will undoubtedly continue to expand.  In many contexts, A.I. presents opportunities.  In others, it creates challenges. 

The use of A.I. within the employment relationship itself, is complex.  Whether its use within that context creates opportunities or challenges is likely to differ between situations, and between organisations.  Irrespective, many employers have begun the journey of assessing and defining the acceptable, and unacceptable uses of A.I. within both the employment relationship, and within interactions and processes that impact workplace relationships. 

Today, many of us will read an employee communication and cynically or otherwise, will make an instant assessment about whether it was generated by the employee, or by A.I.  Maybe that’s OK, but what’s our reaction and how does it impact our approach when we form the view (rightly or wrongly) that it’s been A.I. generated?  If we’re honest with ourselves, I’d suggest that most people will typically discount, and potentially distrust, something they assess as A.I. generated. 

Against that backdrop, in your organisation: 

• Is it OK that a candidate’s application for employment was written by A.I.?  

• When an employee files a complaint, does it matter that the language of the complaint is no longer that of the individual, and is now that of a large language model? 

• Is it helpful that instead of receiving a complaint that details individual occasions where something the employee considered unsatisfactory occurred, you now receive a lengthy submission that focuses on drawing conclusions as to what the alleged conduct amounted to, cites legislation, and sets out a list of demands as to what you (as the employer) are now required to do? 

• Is it helpful that a disappointed or adversely impacted employee’s first point of contact for their concerns is now A.I., and that their initial expectations about what should happen next are being set by a large language model? 

• Are you comfortable with a manager using A.I. in their decision making?   

• Is it OK for A.I. to be used within a process for selecting employees for redundancy? 

• Does it matter that an employee has outsourced the drafting of its written communications with your customers or suppliers to a proprietary or publicly available A.I. platform? 

• Do productivity improvements need to be weighed against the possible de-skilling of those employees who subsequently rely on A.I. for a particular purpose (for example, communicating with others). 

• Does the speed and convenience of relying on an A.I. generated reply carry a risk of a relationship impact on the recipient?  Was the response too friendly? Does it come across as disingenuous?  Did the response miss the customer’s point? 

• Would any of this change if an employee was required to proactively disclose the use of A.I. within their communication? 

Does your view change based on the terms and conditions of the A.I. service being used?  For example, the terms and conditions of a popular, publicly available A.I. service state that when you use their service, you understand and agree that: 

• The output may not always be accurate, and you should not rely on the output as a sole source of truth or factual information. 

• You must evaluate the output for accuracy and appropriateness for use for your circumstances. 

• You must not use any output relating to a person for any purpose that could have a legal or material impact on that person, such as employment, legal, medical or other important decisions about them. 

Courts and Tribunals also continue to react to the impacts of A.I. on their role and processes.  This includes the use of A.I. when generating materials that are used within their processes for defining, hearing and determining disputes.  The regulation of A.I. in these spaces will undoubtedly continue to evolve, but current restrictions include things such as: 

• An obligation to disclose the use of A.I. in the production of a piece of work. 

• A prohibition on the use of A.I. in the production of evidence, including a person’s statement as to what occurred. 

• A certification that any permitted A.I. generated information has been reviewed by the person, who has also independently verified the truth and accuracy of all information. 

Are similar regulations appropriate for communications and decision making within employment relationships?  Is prohibition the solution?  Ultimately, this will be a decision for each employer, but better employers are already engaging with these issues and are beginning to provide guidance and clarity to their people.  Irrespective of where you might land, the first question you really need to ask, and answer is “Am I comfortable leaving it to each employee to make their own decision about these matters?”.