The Employer’s Wage and Hour Advisor

As discussed further below, these changes may inadvertently create litigation risks for employers. Several common FLSA overtime exemptions (including exemptions for administrative employees, supervisors, and outside sales employees) are premised on the employee engaging in significant decision-making and carrying out functions that require a certain amount of authority. If this decision-making is replaced with AI tools, or employees are not given discretion to overrule the AI process, their exempt status becomes more questionable. Additionally, for nonexempt employees, compensable worktime has traditionally been based on employees personally completing various tasks for the employer. However, when AI absorbs more of these routine tasks, it becomes much harder to tell when the employee is actually working.   

Exempt Administrative and Executive Workers

For administrative employees to qualify as exempt from overtime under the FLSA, they must exercise independent judgment and discretion over matters of significance in the company’s general business operations (among other requirements). This exemption is one of the most widely used overtime exemptions in the U.S. economy. Independent judgment and discretion has long been defined by the Department of Labor as involving “the comparison and the evaluation of possible courses of conduct, and acting or making a decision after the various possibilities have been considered.” Such discretion must involve business matters of significance to the employer’s operations. Thus, the adoption of AI products could place exempt status at risk where AI significantly replaces human decision making.

Courts in the Sixth Circuit and beyond have held that administrative tasks that are routine and largely scripted by the employer do not reflect the use of independent judgment. For example, Michigan federal courts have held that, where an employee must follow the outcome of screening tools for admitting patients to a hospital or engaging in certain transactions, with little to no ability to deviate from the screening protocol, such employees are not exercising judgment or discretion. Similarly, the Eighth Circuit Court of Appeals has found that lower-level employees who carry out their duties using mandatory scripts and checklists would lack the discretion needed to qualify for exempt status. In litigation concerning the exempt status of real estate appraisers, courts similarly have found that the use of automated processes to appraise properties and policies that set the outcome of the appraisal within a narrow approved range suggest that the appraisers did not exercise sufficient judgment and discretion.   

Exempt managers, much like exempt administrative employees, must perform certain duties to qualify for the exemption. Critically, managers and executives must, “have the authority to hire or fire other employees, or the employee’s suggestions and recommendations as to the hiring, firing, advancement, promotion or any other change of status of other employees must be given particular weight.” In other words, the exempt manager must have authority to carry out or recommend significant changes to subordinate employees’ employment. 

Where an individual lacks the authority to hire, fire, promote or discipline employees (or recommend those changes), that individual is no longer exempt. To take just one example, in court decisions construing the status of assistant store managers in a retail setting, where the assistant store manager cannot make significant decisions on their own and a more senior manager is free to disregard their input, the assistant manager has been found eligible for overtime wages. Once AI plays a greater role in gathering information concerning employee productivity, performance or conduct, lower-level managers may find themselves essentially just carrying out the decision the AI proposes based on the employer’s criteria.   

Following from these court decisions, the further automation of office work using artificial intelligence could jeopardize the exempt status of a larger group of administrative and managerial employees. To the extent that an AI product generates decisions within a pre-defined range of potential outcomes, using pre-determined criteria, an administrative worker may not be exercising judgment and discretion unless they had some ability to discard the AI product’s result. Similarly, a manager who is generally required to adhere to the AI’s determinations on hiring, promotion, firing and discipline may no longer qualify as an exempt executive. Where an employer limits employees’ ability to vary from the AI product’s result, that could jeopardize the exempt status of those employees.   

Exempt Sales Employees

Likewise, AI could pose a similar problem for employees who are classified as exempt under the FLSA’s “outside sales” exemption. Sales employees must routinely travel away from their employer’s facilities to meet with customers and prospective customers, and their activities must focus on soliciting and making sales. Where a salesperson only promotes the company’s products or services, but does not themselves make sales or approve customer orders, this is considered non-exempt “promotional” work, not exempt sales. The Sixth Circuit Court of Appeals has previously held, for example, that door-to-door sales employees who solicit individual customers and describe products and services, but who depend on their employer’s back office to actually approve and close sales after a customer submits an order, are not exempt sales personnel.

AI’s application to sales processes (including lead generation, customer and receiving orders using automated processes) may risk reducing the outside sales employee’s role to a promotional one.  For example, if an LLM is trained to assist a sales employee by (1) collecting information about prospective customers, (2) automatically identifying or contacting prospects to set up meetings, (3) following up with prospects, and (4) qualifying orders for approval to close on transactions, it becomes difficult to determine whether the sales employee is still the one making the sales. If the actual orders are received by an AI process, and the sales employee role is effectively reduced to merely educating the client about the product, the exemption may no longer hold up if challenged. The more sophisticated AI services become in handling sales transactions, the more difficult it may be to maintain the exemption for certain sales employees.  

Recording Work Hours

AI products may also impact how nonexempt employees’ worktime is recorded and what time is considered compensable. The FLSA generally considers the period from the start of the employee’s first principal activity of the day until the end of their last principal activity of the day to be compensable worktime, excluding significant break periods (such as for lunch). With the advent of more automated processes for office or other non-manual labor, one question that arises is whether the employee is “working” while AI processes are running. The FLSA considers certain periods of “on-call” time to be compensable work hours, but only if the employee is either required to remain at the office or is severely limited in their activities at home while on call.  

If an office worker’s tasks become substantially dependent on AI, they may spend larger portions of their day essentially “on call” – not engaging in actual tasks, but waiting for a certain set of outputs from the AI processes. This otherwise non-productive time would remain compensable under the FLSA, unless the employee were permitted to leave the employer’s premises and engage in personal activities. Employers may struggle to separate such non-productive “on call” time from time spent actually performing other tasks. For instance, employers could arguably require employees to take a series of 30-minute or longer breaks throughout the day to limit labor costs, if no other productive work were completed during these waiting periods. Such a fragmented work day would be a sea change from the longstanding 9-to-5 schedule of the American office.

Complicating the picture further, AI tools will likely be used to monitor employee productivity. Employers may be tempted to rely on such tools to record compensable worktime and only pay employees for the time the AI records as “working time.” Yet, the FLSA and state wage and hour laws present potential conflicts with such practices, given the inevitable disputes about whether the AI is accurately capturing all of the employee’s work activities and how long it takes to perform those activities. Moreover, employees will argue that they are essentially be “engaged to wait” while the AI finishes its analyses. 

While the potential uses of AI seem limitless, employers should carefully consider all of the possible effects of AI on their employment practices before implementing any changes. For instances, where AI will significantly replace employee decision-making and authority, employers will need to determine if the benefit of the AI is worth the possible changes to overtime classifications it could cause.   Additionally, extreme caution should be taken before using AI to determine an employee’s compensable worktime. As technology continues to evolve, employers are well-advised to consult with their human resources professionals and employment counsel before implementing any fundamental technology shift into workplace such as AI.