S7 E6: AI at Work: The HR Questions Companies Are Facing Now

Podcast July 21, 2026

Transcript:

Pete Wright:
Welcome to Human Solutions, Simplifying HR for People Who Love HR, from AIM HR Solutions. I’m Pete Wright. AI — it’s already showing up at work. Already. It’s been showing up at work for years now, believe it or not, whether employers are ready for it or not.

This week we are unpacking the HR questions companies are facing now, from ownership and policies and confidential data to hiring and employee communications and AI meeting tools. It is a practical conversation about how HR can help guide AI use before it reshapes trust, compliance, and workplace culture. I’ve got Kyle Pardo and Terry Cook from our own team here to help us straighten things out.

Kyle, Terry, fantastic to see you. Hot topic. It feels like a perennial hot topic. I wonder if it will ever be a cool topic. At this point, the jury’s out. Let’s get right into it. All of these points are manifested from the questions that we’re getting, so I want to make sure everybody understands these are legit. If you’re not facing them yourself now, you very well might be facing them soon. So who oversees launching AI into the workplace? Who actually owns AI? Kyle, you want to kick us off?

Kyle Pardo:
Yeah, I think there are a couple of ways to go here. I think the easiest answer is people are going to say IT, because it’s related to technology, it’s related to platforms, websites, data gathering, things like that. I truly think it’s an HR topic. It’s going to — or already has — really infiltrated every part of the organization, and it’s going to cause HR professionals to be part of the discussion on change management. This is probably one of the biggest changes any organization is going to go through in decades.

Change management in terms of how people lead, what job descriptions look like, what the culture looks like, what policies there are, what’s okay to do in the workplace and what’s not okay to do in the workplace. So while I think there is a huge component of who owns AI being in the technology space, I think HR is really going to be the people that have to be the cohesiveness and bring that all together for the company and the culture.

Pete Wright:
With AI implementation, I can understand IT wanting to own it, but just hearing you talk about it, it is an HRIS tool. It is an HR information system. It touches everything, and thinking of it as HRIS helps, I think, frame the discussion around where AI ownership lives — not just technical configuration of the tools, but where use lives. Terry, what are you running into?

Terry Cook:
Yeah, I think what Kyle said is absolutely right. HR is enforcing the AI policies. So the policies, the use, the direction that employees need to take related to AI need to come from human resources, and it does start with policies generally. As Kyle said, IT is obviously involved, but HR is definitely more involved in some aspects of it because of that.

Pete Wright:
One of the examples that comes up — we’ve got a sort of separation of employer policies and employee needs that are being met by AI, and I think this example is an interesting one. An employee reaches out to an AI platform using their personal account and is asking about a death in the family, and they ask about bereavement leave policy. Now they’re asking an AI tool, a large language model, about generic bereavement leave. What is your sense of how HR should approach a situation like this?

Terry Cook:
I think it’s one of those things with anything. People, even before AI, would Google anything on the internet to say, I think I need to make more money, because I Googled this. And the same thing, Pete, is what you’re saying. AI gets involved.

Pete Wright:
Right. If I had a dime, Terry, if I had a dime, I would have Googled better.

Terry Cook:
It gives the generic answer. Right, exactly. So I do think it just sets up the interactive process for discussion, to say, it sounds like you may have looked up something or heard about other people’s procedures or processes. Let me share with you what our handbook states about bereavement leave.

Pete Wright:
Yeah, I think you’re underscoring the gap, because somebody who just asks a generic AI tool about bereavement leave — the company’s never going to know that they need bereavement leave. There’s no system in place to indicate how do you do this. And so much more of that seems to lead us down this path of training how to properly use AI — not how to train our own AI, but how do we keep our employees understanding which channels are appropriate for which particular needs.

Terry Cook:
Yeah. One other thing, Pete, that I’ll mention on this: there are companies out there now that are using their own AI, as you said, with handbooks. So an employee can log in and ask the internal AI, what is our bereavement policy? And the AI robot — whatever they name it — can actually go through and give them the company’s bereavement policy, versus just Googling a random one through regular OpenAI. That’s something we’ve been hearing a little bit more. Kyle, I think that came up in one of our recent HR roundtables. People were saying they were utilizing it that way.

Kyle Pardo:
Absolutely. And they’re doing the same thing with benefits plans. So you can load your benefits plan documents into your internal AI, potentially, and ask, do I have vision coverage? Or how much is it going to cost for a certain procedure? I think people still caution us — it’s a tool, and you want to make sure it’s pulling the correct information. But yeah, there are a lot of resources out there.

Pete Wright:
To what extent are we using AI to write employee communications? Employee warnings, termination letters? Terry, this is a fan favorite.

Terry Cook:
It really is. So I do think a lot of people want to get started, or they get nervous about what to say, what not to say, so they try to go to AI. And I don’t think it’s a problem that people want a starting point, but they can’t count on just putting something in, for a couple reasons. One, you can’t put — or shouldn’t put — confidential information and employee names, et cetera, into AI. And two, AI doesn’t necessarily know all of the relevant federal and state laws. So you can’t just say, should I give this person a warning, and what should I give them? Because it doesn’t know your handbook policy. It doesn’t know the laws related to it.

And it also will come back, as we all know, in an AI tone. It’s not going to read like Terry Cook wrote it. It’s going to read like a computer wrote it. So I think there are a lot of cautions around it. I do think people can use it as a starting point to have a draft, but it definitely needs to be reviewed for compliance, for tone, for many things, to make sure it meets your policies. There are a lot of other pieces to it. So I would not say just use AI to create a termination letter or disciplinary notice.

Kyle Pardo:
I’d add to that, we get a lot of questions from people asking, is it okay within the company culture to use AI to write performance management documents? And it’s what Terry’s saying — what you put in is what you’re going to get out. So if I say, I’m writing Pete’s performance review for the year, give me a good performance review for this person, you’re going to get something pretty generic back out. But if you put something in and say, this person specifically worked with this client, here’s a quote from the client, here’s some data on their error rate in their work, which was better than average or better than they did last year, they also took some personal classes — whatever it is, you really put in some custom information without naming the employee and putting really personal information in there — you’re going to get a better document back out that you can then craft and make sound a little bit more like your tone.

But if people are using it just to quickly say, yep, I did the review, I checked it off my list, and I’ve done it, just by doing something generic in AI, that’s when it just doesn’t add value for the employee and for the organization.

Pete Wright:
There is a solid practice, which is — say you haven’t been writing a lot of reviews. You’re not sure how to write your own review. It’s okay to give completely generic context and say, I need help figuring out how to write a review. It should include these kinds of things. And then you get to go rewrite the review. That’s the important part. Use it as a learning tool, not as a finished product.

And I think you both have danced us right into the most complicated question, which is uploading or including employee confidential information, including identifying information, into an AI tool for the purposes of a performance review or a warning or whatever. The big question is, what now?

Kyle Pardo:
Think about it from an HR perspective. We have access to a lot of personal data, and we might think about using it for benefits open enrollment. Maybe an HR person is thinking about, what are the demographics of my organization? Let me upload a spreadsheet and get AI to analyze it for me — the age of my employees, or the gender, or the pay rates, or whatever. And if an employee thinks about it in that context and says, look, I’m just going to upload this whole spreadsheet of my employee data, I’m going to put it in there and see what information I get — that’s out there now. That’s not coming back. We talk about how you post something even on Instagram, whatever — that’s going to be there for a long time. This employee data is out there.

So I think right now, for organizations, before there’s a situation, there are a lot of steps that need to take place. A company policy that says you cannot enter client information, customers, employee information, and this is what we consider personal information. A lot of states have data privacy laws, so make sure that employees understand what that means. And then the second step, after having policy, is training employees on it — have everybody sign off that they’ve read the policy and they know what it means. And then, in the worst-case scenario, having the final piece be, what happens if somebody still does it? If the data gets out there, we can’t get it back. Do we need to notify employees? Perhaps it has Social Security numbers or something really personal. Is there a law or something that mandates that we notify employees that this is out there? Are there any other tools we can put in place to try to recapture something? Proactively, at this point, if there hasn’t been an issue: get that policy together, do the training, and then have kind of an emergency plan in place for what might happen later.

Pete Wright:
This is one of those interesting scenarios, because somebody doing the things we’re describing — uploading confidential employee information to the AI — just to be clear, there’s no recourse. Once it’s in there, there is no reliable way to confirm, no matter what your policy states, that that data has been deleted. There should always be a question. Yes, that stuff is out there. And so that brings us to this larger question. What happens to it? What is the environment like right now, where people are using AI and not thinking about these sorts of repercussions? Terry, you want to pick it up?

Terry Cook:
Sure. And I think that’s what it is — like you said and Kyle said, the problem is it’s out there. There’s no undoing anything. And so I think it’s a matter of what you do afterwards, and Kyle already started walking everybody through it. You can’t pretend it didn’t happen. It happened. And it depends on what’s out there and what you decide to do. You obviously have to train, but if there’s some big data loss, or there’s a data breach that had to be reported to the state or whoever they had to report it to, the company has to decide who released that information, and what level of discipline, per their policy, it results in. Because when you’re dealing with a lot of personal and sensitive information, you just can’t take it back. You can’t press that nice button that says undo when it comes to AI.

Pete Wright:
Oh, undo on my life, right? I love undo.

Terry Cook:
I do, I love undo. But you can’t do it in AI, and I do think that’s where people will have issues — have had issues. We read a case a while back where somebody that was supposed to be doing an executive board meeting brief of their notes thought it would be easier to just copy and paste it into AI and say, create the minutes for me. Well, now all of that confidential information about that company, all the board members, everything is out there, and you can’t undo it. You just can’t take it back. So it’s a matter of making the corrections and taking the appropriate level of discussion with people, whether that includes discipline or not.

Pete Wright:
The tricky part is that these tools — even the individual tools — have gone back and forth in their policy on their default setting. Many of them say there’s an option to go in and say, don’t use my chats for training. But the status of whether that’s set to on or off depends on when you installed the app the first time. So you have to know to go in and look at it. That’s a practical thing. You can go in and look at, does my installation say use my material for training future data?

The other piece to remember is, even if that’s turned off — and I just say this because it’s in the news and it’s hot — OpenAI and the New York Times are in litigation right now. The New York Times is suing OpenAI, and as part of the discovery process, the court has ordered that OpenAI not delete any of the chats from a large portion of 2025. So that information, even if you think it’s deleted, even if you deleted it from your account, is not deleted, because the court said no. This is a really important thing to continue to wrap your heads around — this is a very fluid situation, and even more reason to take this stuff very seriously.

And yet we have this other side of the coin. Whether it’s satisfying shareholder value, whatever — CEOs and managers sometimes have a mandate to, quote, use AI everywhere. So now we’re being told to find efficiencies, to use AI more. Kyle, what does that even mean for HR, specifically now that we’ve set up this context that the land is on fire when we turn on our AI system?

Kyle Pardo:
You can see that it’s becoming more and more in use in everything that we do. But going back to our first question about who owns AI, I think it goes back to taking a look at every job description in the company and rewriting the job description. Not even just thinking about it, but actually rewriting the job description, and putting information into two buckets. What part of all of our jobs might be transactional, might be administrative, might be things that AI can do? So we can get some of those things off of our plate, and we can streamline them, and we can write our emails more quickly, and we can create checklists — we can do a lot of things that might take us time. And then the second part of the job description is, what is human-centered? What requires a personal approach? What requires creativity and collaboration and all of those other pieces of the job?

So when we talk about AI being everywhere, people get worried about that. They think, oh, this is going to take over my whole job. So I think it’s a redesign of everybody’s job, separating out how it can help us, but where we still add that value. I don’t think AI is going to be limited. I think it’s going to keep growing. So reframing it now is probably a good time to start.

Pete Wright:
Yeah, I’m feeling a lot of that sentiment — the vastly more negative. When you look at the response data, particularly younger people are really feeling like AI is an antagonist to their life, and they’re not in favor of it. So we have this other side. We have management teams and CEOs saying, we’ve got to figure out how to find all the efficiencies, and we have new young people — who should have all the blossoming enthusiasm of a grand career opportunity ahead — saying, no, thank you. Living between those two things is a real gap. Have you run into any of this — addressing that gap yet, in terms of using AI in hiring and how it’s impacting bringing young people onto teams?

Terry Cook:
When it comes to hiring, I actually just read an article recently where it said some of the younger generations are completely turned off by AI use in hiring. They’re indicating they can tell when AI is interviewing them versus a person interviewing them, and if they feel that, they said they just drop right out of the hiring process. They want nothing to do with that company.

And the other piece of that, on the reverse side before I go further in that direction — when you’re doing recruiting, I have noticed a lot of resumes that are clearly written by AI. What they’ll do is take the job post and say, can you please create my resume to line up with this job post? And it does. It creates it. But you’ll see it as the person reviewing resumes. You see all of the keywords that you wanted in your job post, and it’s not supported at all in the experience. But yet it’s at the top of the resume, saying Terry has all of these nice words that I was looking for in a job post. So from an HR standpoint, we can see that.

I’ve also read that when you take some of your standard interview questions, people that are doing phone interviews are just typing those in ahead of time, and they have a nice answer generated by AI for the HR person interviewing them — an answer that’s going to be a good answer. How do you handle conflict? Here’s how I handle conflict. And it’s just all AI working it through. So I’ve read, from the HR standpoint, what they’re experiencing.

But as you said, Pete, there are definitely people that are turned off by being interviewed or going through a hiring process with AI. I think some companies feel like it takes the bias out. But articles I’ve been reading — and I’ve certainly experienced it differently, because AI does learn from us — said that they feel, in a high percentage, that AI can be just as biased or more when going through the review of resumes or the interviews. So it’s really been interesting to follow that in the recruiting line.

Kyle Pardo:
I’ll add to that something that happened recently. I was recruiting for a position and read a cover letter. And then I was looking at another candidate, and I thought, I feel like I’ve already read this cover letter. It was a separate applicant, and when I looked at the two cover letters, not only was the language very similar, they were structured exactly the same.

Pete Wright:
Oh no.

Terry Cook:
Uh-huh. My goodness.

Kyle Pardo:
So apparently what people have done — and this is very common — is they put their resume into ChatGPT and said, help me write a cover letter for this specific job. So they put in the job description of what they’re applying for, and ChatGPT gave them a cover letter and broke it down in the same four categories based off of what the job was looking for. So in the two cover letters, the font was the same, with the bolded font at the beginning of the bullet, with the description of what their background was.

At first I was taken aback by it. I thought, gosh, they didn’t write the cover letter themselves. These look exactly the same. But then I was like, all right, maybe they’re helping me out a little bit. Maybe they’re pointing out their background, making my job easier, because they’ve highlighted what’s on their resume that might be important to me. And so when you think about it from a recruiting or an HR perspective, it’s like, does it matter? They got me the cover letter. They covered the points that I needed. If they’re going to use that as a tool for writing on the job, do I care or do I not care? And that’s where I think people need to make that decision for themselves and what that means to them.

Terry Cook:
Well, you got a cover letter, which is amazing in itself sometimes in the recruiting process.

Pete Wright:
It’s amazing. I was just going to say that. Look at you — you got a cover letter.

Terry Cook:
Sorry, Kyle, I don’t get them anymore. I’m excited for you that you got a couple of cover letters.

Kyle Pardo:
Yeah.

Pete Wright:
And for grounding in our recent past — keyword stuffing in resumes and cover letters is not an AI thing. That happened when we started using automated response-gathering tools years ago. So maybe this made it so job seekers could further automate and refine, but this is not a new thing. So as you’re thinking about receiving these resumes, it’s not because AI exists that you’re getting these. You’re getting them because we’ve trained job seekers how to use whatever tools they have in a very complicated environment. And this is a bed of our own making, in many respects.

Terry Cook:
But it’s interesting when you say it, Pete, that you’re hearing the younger generation might not like it as much. And I think that is surprising, because we would think that people would like the fact that it’s easier. Like we’ve heard in the past in hiring, if you ask too many pre-questions, you’re not going to get the applicants, because they just want to do a quick apply. They want to get things done fast. But like I said, I found these articles I’ve been reading fascinating. Just the reasoning behind it — it’s very interesting.

Pete Wright:
Yeah, I think the gaps are fascinating, because you’re pointing out this other side, which is: I hate AI, I don’t like what it represents, and also I have to use it if I’m going to apply to the number of jobs I have to apply to in order to secure something. So there’s an ideological gap — people are having to ride this bus no matter how they feel.

One of the most popular tools — and we’re getting toward the end, but I want to make sure we cover this because I think people are using it a lot — is recording in meetings with AI tools. When’s the last time you were in a meeting and a recording bot wasn’t there? So we’ve got some legal points. I would love to know where we stand. Kyle?

Kyle Pardo:
Yeah, I’ll start, and Terry’s part of a policy writing that we just put out this week. You’re absolutely right. You go into a meeting, you see somebody’s AI note taker is on there. It’s helping them do their job, helping follow up, which is great. We love that. When the meeting’s over, we all kind of know what steps need to be taken. But at the same time, sometimes you’re in a meeting and you think, do we want all of this recorded? Do we want everyone to know exactly what was said by whom? Do we want someone to potentially upload that recording somewhere we don’t want it to go?

And then I think that leads to norms. If I say to you, could you please turn your AI note taker off — does that become offensive to somebody, that I’ve asked for that? Is that okay for me to say, if I don’t want it to be on? So I think there’s that norm piece, and then I’ll turn to Terry to talk about what the actual policy might look like.

Terry Cook:
Yeah, and I do think the policy is about what the company wants, but it also incorporates the state laws as well. I’ve noticed that people have asked in the past about smart glasses or note takers, any of these things. And in Massachusetts, for example, there’s data privacy — you have to have two-way consent to recording. So both people — anybody involved that’s being recorded — have to consent in order for it to move forward. In any of these scenarios, like Kyle’s saying, if you’re in a meeting and you don’t want to be recorded, technically, in some states, you would have to be asked or give permission to be recorded. So that could be one piece of it.

I think Kyle already brought up a really great point on the privacy of the information. Also, I think a lot of note takers are great. I’ve had people use them that didn’t tell me, and they’ve sent me the summary and it’s been fine. But there are other situations where it didn’t capture the information correctly. I don’t know if it heard wrong somehow on one side or the other, but I’ve seen the summary where it’s not accurate. So again, having that policy in place, especially at work, to say what can be expected.

Some pieces could get changed by that — if it’s an accommodation piece, for example. Maybe somebody needs a note taker because of an accommodation issue, and as an HR person, we have to keep that in mind.

Pete Wright:
But accommodations don’t supersede state and federal law around consent, right? Like, it just says, I need this as an accommodation. I am telling you that it’s going to happen.

Terry Cook:
Yeah — you have to find a way around it, though, essentially. So if somebody were to say, listen, I have to use a note taker from now on because of this condition I might have, then you as a company have to make that decision, to say, is this an accommodation that we can make because of this? So it might supersede a policy. As far as the law goes, you would really have to work within the law and find the points of the law that might be open to the accommodation piece because of a medical reason, for example.

Pete Wright:
There are eleven all-party states, if I understand correctly, and Massachusetts is definitely one of them. So generally speaking, federal law is one-party consent. If an AI recorder is in a meeting and one person happens to be in Massachusetts, then it’s all-party consent territory. So generally, just make sure people know and are okay with it.

This smart glasses question comes directly from the chat room, too. There have been talks about Meta glasses allowing instant recording of a person, or the situations when other people at a workplace are not aware they’re being recorded or photographed. Any advice? That seems to me to be illegal.

Terry Cook:
Yeah. It’s a policy thing. It could be an illegal thing in Massachusetts, for sure — any place there’s two-party consent. As you said, Pete, federal law is different. Some state laws are different, but there are several states that have two-party consent. In that instance, that would be a problem. In fact, I just had a question on our HR helpline about smart glasses in Massachusetts, so we went through all of that.

The other time recording comes up is just in general — people saying, I’m going to walk into my manager’s office and say, I’m recording this conversation whether you like it or not. Again, it’s all about your policy, and it’s about what the appropriate state law is. So in that example where the person’s asking about the Meta glasses, it depends on what state they’re in, more than likely, because of the two-party consent. It does feel like it’s not okay to do overall, in my opinion, too. To have somebody recording me without my knowledge — it doesn’t feel right. So I think it’ll be interesting to see how many challenges come up from that. I think that’s just starting in the workplace more and more right now. And a lot of people are protecting proprietary information in the workplace, so that’s another reason why they’re not allowing smart glasses in the workplace.

Pete Wright:
We’re at the point where we’ve reached the end of our conversation, but I want to give a final note to what I think is probably the best marching orders going forward. And it’s hard. I think it’s going to be hard to do. But I’m going to say it and then let Kyle talk about it. It is: let’s stop trying to catch AI. What’s your perspective on this, Kyle?

Kyle Pardo:
I think we need to figure out how we’re going to use it and the best way to do it. Like we try to do in a lot of these podcasts, there’s a lot for HR to do, and it goes back to the first point of who owns this. I think HR does. Updating job descriptions, writing policies, coming to a consensus of what you are okay with and what you are not. Are you okay with that cover letter that came in with AI, or not? So it’s starting to have those conversations. Where is it something that the CEO says we’re all going to use? Then let’s figure out how to make it work, and also think about how to make our employees feel comfortable with each of those changes. But change management — if you’re not on board with that, it’s going to keep changing. So get ready to figure out how to communicate that to your employees.

Pete Wright:
Terry, last words.

Terry Cook:
Yeah, I think it’s communication and training, as Kyle said. Those are just big pieces there, and as she said, you just have to find the way to make it work at your company, because it’s not going anywhere.

Pete Wright:
And just one comment on hope. There are people out there who are actually leveraging AI for great things. They are able to find the efficiencies that the management teams are asking for. They’re able to leverage these tools to do things that they have never been able to do before. And that’s incredible. And probably we should be looking at those people for guidance on how to write these policies.

Terry Cook:
Absolutely.

Pete Wright:
So thank you both so much. As you say, it’s going to keep changing. We’ll probably talk about this again sometime, and I look forward to it.

Terry Cook:
I’m sure.

Kyle Pardo:
Me as well.

Pete Wright:
As always. I’ve got some show notes — if you swipe up, I will be putting those in the podcast version of this show and over on the website. So definitely subscribe to the show there if you want to catch the replay. We would love to have you there. And the entire archive of these past conversations is available to you whenever you need to catch them.

AIM members can reach the HR helpline at 800-470-6277 or helpline@aimnet.org for inquiries Monday through Friday, 8:30 a.m. to 5:00 p.m. Eastern Standard Time. Next month, I’m talking with Mary McNally about benefits. You know, fireworks ensue. Fireworks ensue.

Kyle Pardo:
Another area that keeps changing.

Pete Wright:
Right. Thanks, everybody, for joining us. Thanks for sticking around. Hope you’ve had a nice lunch hour, and we’ll let you get back to work.