Mintz On Air: Practical Policies — Take a Human to Work Day

The secret to a thriving AI-assisted workplace might be surprisingly human.

In this episode of the Mintz On Air: Practical Policies podcast, host Jen Rubin is joined by Corbin Carter, a Mintz employment attorney and counselor who advises employers on workforce management, compliance, and employment disputes.

Together, they explore the growing role of AI in workplace decision-making, including:

• The importance of human accountability for decision-making in the workplace
• How a human in the loop is not only necessary to implement and enhance AI, but is critical to hiring, performance, and employee relations decisions
• How human interaction builds communication skills, collaboration, and culture and reduce legal risk in the workplace

This episode offers practical insights for HR professionals, people managers, business leaders, and in-house counsel navigating the evolving role of AI in the workplace.

Mintz On Air: Practical Policies — Take a Human to Work Day Transcript

Jen Rubin (JR): Welcome to the Mintz On Air: Practical Policies podcast. Today’s topic: Take a Human to Work Day. Be that human manager in the AI workplace loop. I’m Jen Rubin, a Member of the Mintz Employment Group with the San Diego–based Bicoastal Employment Practice, representing management executives and corporate boards. I am always looking to provide creative and business-centered solutions to workplace legal problems. Thank you for joining our Mintz On Air podcast. If you have not tuned in to our previous episodes and would like to access our content, visit the Insights Center at Mintz.com, or find us on Spotify.

Today I’m joined by my employment colleague Corbin Carter, a solutions-oriented employment counselor and litigator who guides clients through all aspects of the employment life cycle. Corbin advises businesses across a wide range of industries including financial services, technology, life sciences, health care, and real estate. Corbin’s practice covers everything from offering day-to-day employment advice to leading the management-side, defense, and prosecution of employment-related claims at the trial and appellate levels.

Corbin is kind of bicoastal like me — he spends time in New York and out of his Dallas, Texas office. I don’t know if we call that quasi-coastal, inter-regional, or dual city. But it doesn’t matter what we call it, we just enjoy working together no matter where we happen to sit on any particular day. Thank you, Corbin, for joining the pod today.

Corbin Carter (CC): Thanks, Jen, I appreciate it. It’s good to be back. I’m going to go with “co-located.” How about that? 

JR: Okay. You got it. “Co-located.” Good enough!

Mitigating Potential AI Employment Liability: A New Basics

JR: Our listeners are probably wondering what I mean by “take a human to work.” As you’ve probably gleaned, the topic of today’s episode is focused on how human managers could and should mitigate some of the potential liability and problems associated with having technology quite literally take over our workplaces.

I thought it might make sense to revisit a world in which humans make decisions and choices and exercise good judgment (hopefully most of the time) — and some of the things we can do to make our workplaces more productive and conflict free by going back to basics. But a new basics. Because let’s face it, our world and workplace look and feel really different in 2026.

I’d like to set the stage with the legal framework and the challenge AI brings to that legal framework, and some practical solutions to help human managers manage better as humans in an AI-enabled workplace. 

Corbin, you and I have discussed AI’s impact on how employers recruit, retain, and train employees. Starting at the most basic level, what happens when you automate the recruiting, retention, and training framework for the workplace?

CC: We’re still living in a highly regulated employment environment that changes very slowly. The baseline reality is that technology will always outpace regulation. Most employment laws — Title 7, the Age Discrimination and Employment Act, the ADA, the Americans with Disabilities Act — these were drafted with the fundamental assumption that humans, and basically humans alone, are making employment decisions.

These statutes focus on the intent, the actions of human decision makers. Now we’re seeing a patchwork of new laws trying to catch up. States like Colorado and Illinois and municipalities like New York City have enacted AI disclosure requirements. And there’s a growing body of guidance from the EEOC, the federal agency, about the need to audit and de-bias AI engines that are used in hiring and other employment processes.

But the gap between what the technology can do and what the law requires remains substantial. A great example of this tension playing out in litigation is the Workday case. It’s a popular case often discussed, Mobley v. Workday, Inc. It’s pending in California federal court right now. The plaintiff there alleges that Workday’s AI-driven screening tools, which are used by hundreds of employers, maybe thousands, systemically discriminate against applicants based on race, age, and disability.

The core legal question there is whether Workday, as a vendor providing AI screening services, can be held liable as an employment agency under the various anti-discrimination laws. And recently the court in that case rejected some of Workday’s arguments that its liability as an agent has to turn on the liability of its employer customers, meaning only derivative liability can exist.

That’s their argument. And instead, the court held that claims under California’s anti-discrimination statute could proceed because Workday’s potential liability is potentially based not just on derivative liability, but directly on its own engagement in regulated activities on the employer’s behalf. It’s really interesting. There’s also an IBM case where similar claims have been raised. These cases represent a new frontier of AI employment liability.

Employers need to understand that using the third-party AI tool does not insulate them from discrimination claims. At least for now, the business will always remain accountable for employment decisions. That’s the fundamental principle. But who is “the company” in practice? Who’s the employer? It’s the actual managers and supervisors who implement policies, who make daily workplace decisions. Under some employment discrimination statutes, individual supervisors can face personal liability in certain circumstances.

This is key. Because when AI makes or influences a decision — say, screening out a candidate or flagging an employee for performance issues — the manager who acts on that recommendation becomes the face of that decision. They can’t point to an algorithm and say, “Well, the computer did it.” Courts and juries are going to look at the humans who press the button as the decision makers.

That leads to an obvious question: Why do we need a human in the loop? The first point is about discernment. AI can parse keywords and credentials, but what it can’t do — at least not yet — is exercise the professional judgment that comes with human interaction. Think about reviewing resumes. A human recruiter can appreciate that perhaps a candidate took time off to care for a family member, that a career pivot actually demonstrates adaptability on the candidate’s part, or that a nontraditional background might bring fresh, new perspectives. They can read between the lines as a human, notice that a cover letter reflects genuine enthusiasm versus a template-scripted response, pick up on subtle signals during interviews that indicate cultural fit or leadership potential. That sort of nuance is lost when an algorithm is doing the initial screening. 

Also, workplace disputes arise in the context of misunderstandings. This is sort of our bread and butter as employment attorneys, right? Misreading, failures to account for human quirks. That’s where disputes arise. Consider an employee who suddenly starts missing deadlines or seems disengaged. A good manager is going to ask what’s going on. They may learn something about the employee’s family health crisis or difficult commute situation, or perhaps burnout.

That conversation between employee and manager can lead to a reasonable accommodation process, or some sort of temporary flexibility that keeps a valuable employee. An AI system might just flag declining performance and go no further before recommending termination or discipline or something similar. 

Also, interpersonal conflict can stem from communication style differences from different cultural backgrounds or different past experiences that require human mediation and understanding to resolve. If a machine is doing the reading of employees and candidates, that history and context goes missing. It creates a paradigm that I think breeds more workplace disputes. AI operates on data points and patterns. It doesn’t know that Sarah in Accounting has been the glue holding her team together for quite some time.

Or that Juan in Sales had one bad quarter because he was handling a major crisis that saved a key relationship elsewhere in his book. Those decisions can’t be based purely on algorithmic outputs without human context. I think AI sometimes leads to the evaporation of that context. Employees feel dehumanized; that erodes trust and engagement. That’s often the seed for a discrimination complaint, wrongful termination, or some sort of other legal action by employees. 

Also consider the downstream effects. Employees who feel they’re being managed by algorithms rather than by people can become disengaged. And disengaged employees are more likely to make mistakes. They’re less likely to go above and beyond. And often they’ll start looking for other opportunities. Employers know very well that turnover is expensive. Estimates suggest replacement costs for employees can be 50% to even 200% of annual salary when you factor in things like recruiting and training and lost productivity. That’s even before you get to litigation costs if things go really poorly.

I think having a human manager who builds relationships and addresses concerns early is a significant risk mitigation strategy. 

JR: I can’t dispute a single thing you just said, Corbyn. There’s no doubt about the importance of having good judgment, discerning what the true facts are, picking up on the human engagement, understanding the nuances of how we communicate with one another. It’s interesting to me because those nuances lead to learning how to better communicate, in fact.

Of course, how we build trust is the foundation for a productive workplace. We as humans thrive on trust; I think AI thrives on data. Those things seem to me to be completely at odds. 

However, with respect to that, there are good reasons to have human-on-human management, but we also need to understand that this technology is here to stay. And it’s not only here to stay, it arguably massively enhances our productivity. The paradigm that rejects AI technology outright just doesn’t make sense. 

CC: You’re absolutely right. It’s here to stay. To be clear, this is not an anti-AI diatribe. It’s just trying to square these human elements with what is inevitable: AI. The key is learning how to integrate these new technologies with real human beings.

How Mangers “In the Loop” Can Harness AI

JR: Let’s pivot to how can managers can harness AI with a view to understanding its social and cognitive impacts on the workplace.

CC: You know, I think the concept of “human in the loop” is premised on accountability and not just accepting what AI delivers in some unquestioning way — but also ensuring the proper functioning of the workplace.

The EEOC and many state and local governments have been clear on this point. Employers now can expect — and government agencies are saying — you can’t avoid liability by just claiming that an algorithm or an AI tool, rather than a human, is making discriminatory decisions. The human in the loop isn’t just good practice, it’s becoming a legal necessity.

That human need to understand what the AI is doing, to be trained to evaluate its outputs, to override it when appropriate — that’s all important.

What does an accountability framework look like in this context? When AI makes a wrong call, whether it screens out a qualified candidate or misinterprets an employee communication, someone might have to answer for it legally. And “the algorithm did it” is not going to be a defense that flies with regulators, juries, courts, or your own employees.

Employers need to think about who is responsible for validating AI outputs, who is going to sign off on AI-influenced decisions, and how they document that human oversight that’s occurred. This is becoming AI workplace dogma. Managers also have to be trained on balancing empathy in that accountability platform, because that’s what’s going to build trust.

Employees who feel their managers care about them as people, not just as productivity robots, are significantly more engaged. They’re more loyal, more willing to raise concerns before they become problems. That’s not something you can automate. A manager who notices someone’s human issues and asks how they’re doing, who takes the time to explain the “why” behind difficult decisions, that manager is going to build the kind of trust that can prevent employment disputes. Trust is the foundation of effective feedback and performance management. It all goes hand in hand. Employees are more likely to accept tough feedback from managers they trust. When they get negative feedback from a faceless system, they’re going to leave, they’re going to lawyer up. 

A big, important part of this involves managers walking the floor. I know that’s sort of an odd phrase to use, but whether that’s physically or digitally, those managers might discover better what their employees need. This type of accountability also builds trust. A good manager who’s walking the floor in a remote environment, making regular check-in calls, actually doing face-to-face interactions, they’re going to pick up on things no algorithm can capture. They’re going to notice body language. They’re going to hear hesitation in someone’s voice. They’re going to observe team dynamics in meetings and develop that intuitive sense for when something feels off. Those skills — observational skills — are developed over time. They require human interaction. 

But it’s also two-directional. Employees learn to trust managers they see engaging authentically, asking questions, responding to concerns. It’s a two-way street, and you can’t build that kind of relationship through AI-viewed dashboard metrics. 

Finally, it’s important that the accountability paradigm take into consideration what’s been described as this thinking paradox. You hear this in AI conversations a lot. How do you learn how to think if you don’t have to think? That might be a profound long-term AI concern. We’re raising a generation of professionals who can get instant answers without having to think through a lot of the questions. That’s powerful. And it’s potentially atrophying cognitive muscles. The ability to sit with ambiguity, to work through complex problems step by step, formulating hypotheses, testing them. These are skills you develop through practice, and also through collaborating with other people in the employment context. That’s key. Good judgment requires wrestling with incomplete information.

So there are competing considerations. And if managers rely too heavily on AI recommendations, they’re going to lose the ability to make these nuanced calls when the AI doesn’t have the exact answer they want or need, or the AI gets it wrong. Teaching managers and employees the importance of thought in tandem with AI is going to be an important part of the accountability paradigm.

JR: To me, that paradigm is dependent on questioning. That goes back to the discernment comment you made earlier. Looking at an output and exercising that independent thought process, not asking another AI agent, but actually doing that independent thought. One of the keys here is feeling comfortable with unanswered questions and enjoying the creativity of the unanswered questions instead of relying on the idea that everything has to have an answer, and there has to be a way to do this, and there has to be a data set that’s going to solve the problem. Going through that thought process on your own is an important teaching point. 

The Importance of Humans Driving Workplace Culture

JR: Without question these are all important and heady topics. Let me ask you, Corbin, another paradigmatic and important question: How do you maintain a sense of human identity and common purpose in an AI-driven workplace? How do we maintain our culture, drive our culture, create that identity that instills in your employees the belief that you’re all operating toward a common mission?

CC: I’d go back to the theme I spoke about earlier. It’s human interaction. The more we interact with each other as humans, the more adept we become at reading social cues that tell us how to be good communicators. This is part of why real person-to-person interactions matter so much, face-to-face video interactions, what have you.

Those are going to teach us how to read micro expressions, adjust our communication style in real time, develop emotional intelligence — all of which are wonderful things to have in the workplace. All of this is relevant to what makes people productive and, importantly, what sort of perspective employees are going to have.

Human leaders have to be intentional about culture, especially now. It’s going to require more and more human judgment, more and more human relationship building. 

How can we make our workplaces better, more legally compliant? This is all wrapped up in this question. Why would we want an AI driving culture? That seems like an odd concept to me. Culture is something that’s built through human-shared experiences, stories, values that are reinforced in daily interactions. When AI starts making decisions about who can be hired, who gets promoted, who gets which assignment, how performance is going to be evaluated, all of those things are going to shape culture, and not necessarily in the deliberative, value-driven way we want it to. It shapes culture based on whatever patterns exist in the data it was trained on, which could include things like historical biases or priorities that don’t reflect where the organization wants to go.

Human Empathy as an Employment Enhancer

CC: Another thought comes to mind about this mission communication within organizations relates to modeling excellent communication and analytical skills and understanding the “why.” That’s all important. Leaders are the ones setting the tone for how people communicate in an organization.

If a manager takes time to explain important decisions, if they’re writing clear, thoughtful communications that actually sound like they’re coming from a human, if they’re engaging in genuine dialog with their teams, if they’re demonstrating rigorous, analytical thinking — that kind of behavior cascades throughout the organization, and people are going to emulate what they see from leaders.

Conversely, it’s a problem if managers outsource that communication to AI. You can’t just use a generic or generated template for everything. If you let algorithms summarize complex issues, you rely on those automated responses — it’s a very different modeling approach than what you probably want to take. Similar to that cognitive atrophy idea we were discussing earlier, if this is how organizations treat things, they’re also going to miss opportunities to develop their own communication skills among employees and to coach team members on effective professional communication.

Empathy is so important in this whole idea. The importance of empathy can’t be overstated. I think about situations where this matters most in the workplace — when delivering difficult feedback, handling a termination conversation, addressing a harassment complaint, supporting an employee through those issues, through other personal crises, navigating team conflicts. Those are all high-stakes moments where someone has to have the ability to read the other person, to understand how they’re feeling, what they need, and how to communicate in a way that preserves that person’s dignity. That’s going to make an enormous difference in outcomes. Empathetic handling of difficult situations reduces legal exposure because it leaves employees feeling heard and respected, even when news may not be good.

These are fundamentally human skills that require practice and can’t just be delegated to technology. 

JR: There is no doubt that empathy is an employment enhancer. There is also no doubt that managers are going to be able to use AI as a way to avoid having those difficult conversations. This concept, you called it cognitive atrophy — I love that term, by the way — that’s going to be a problem for managers who are not using that skill and that experience of human-to-human interaction, and that’s a concern.

So especially for our human resources professionals who are tuning in to this podcast, that’s something for them to think about: how they can support those managers in using those very human skills, because they need to understand that empathy is an employment enhancer, meaning it can lead to better productivity. Even if people would like to hide behind AI and data, it’s important that they be able to confront that human-to-human communication. 

Emotional Surveillance: Where We Go from Here

JR: That leads to the last question I want to ask you, Corbin. I recently read an article in The Atlantic that addressed what it calls the rise of emotional surveillance. The article explored how some companies are using AI to assess how employees feel.

In other words, are their employees engaged? Are they stressed, are they at flight risk? This is used along with personality assessments that can predict job performance and cultural fit. And this technology raises interesting questions in light of our discussion here, certainly questions that concern us as employment lawyers. But let’s answer this question: What happens when AI interprets a certain cultural communication style as somebody who’s disengaged, or misreads a medical condition with symptoms that might present as low energy?

CC: I appreciate that question — I think it’s at the heart of our discussion today. There’s such an importance to having a human in the loop. I know that’s a phrase that’s now getting overborn, but it’s so important. The human in the loop isn’t just for accountability. It’s also for the empathetic reading of people. If employers are thinking about how to assess how employees feel, maybe some human interaction could be involved there. 

The European Union has already taken the position that emotion recognition AI is too risky. It’s banned in the workplace under the EU’s AI Act. And in the US, we’re seeing increased state-level recognition and regulation of this issue. Biometric data collection laws are now on the books in Illinois. Other jurisdictions are looking at bias audits for AI tools — New York City is doing that.

So this is an area where you’re going to see continued regulatory evolution. And employers need to be cautious about adopting technologies that may soon face significant legal constraints, including ones that deal with emotion recognition.

JR: All of those things get at the core of what we’ve been talking about, which is that accountability is actually the paradigm that applies from a legal, moral, ethical standpoint. That is the human framework. I’m going to go a step further and challenge our listeners to think about whether empathy is the distinguishing competitive edge for humans in the workplace. That’s something I’m really hoping AI cannot develop to the point where it can exceed our capabilities in that respect, because that’s going to be a very different world indeed. 

So thank you, Corbin. This has been a really interesting discussion. These issues are so important for employers to consider. Like I said, and as you explained, AI is here to stay. So workplaces needs to learn how to enhance it, to use it, but at the same time make sure they double down on the human in the loop. Those are critical concepts here. 

For more information about our Employment Practice and our other thought leadership, visit us at the Insights Center at Mintz.com, or look for our Mintz On Air: Practical Policies podcast on Spotify.

Subscribe To Viewpoints