TL;DR: AI can support — but should not replace — human judgment in employee relations. In this quick conversation, HR Acuity CEO Deb Muller and Chief Product Officer Dianne Frommelt explain how bias in investigations and bias in AI can compound if not managed carefully. That’s why HR Acuity is taking a thoughtful, responsible approach to integrating AI into ER workflows — one that helps surface facts but leaves conclusions to people.
Three Key Takeaways from the Video: AI in Employee Relations
- AI Should Surface Facts, Not Conclusions: HR Acuity uses AI to highlight and summarize relevant information — not to draw investigative findings or make decisions that require human nuance.
- Bias + Bias = Risk: Investigations already carry the risk of human bias. Adding AI without guardrails can amplify those risks. Responsible tech requires responsible implementation.
- Good Investigators Still Matter: Strong judgment, contextual reasoning and emotional intelligence can’t be automated. AI is a tool, not a replacement for skilled ER professionals.
Video Transcript: Approach to AI in Employee Relations
00:00:12.559 — Deb Muller
Hi everyone, it’s Deb Muller, and I’m back for another episode of Two-Minute Matters. I’m the CEO and founder of HR Acuity and I’m joined once again by HR Acuity’s Chief Product Officer Dianne Frommelt to talk about an issue that’s top of mind for everyone.
00:00:26.160 — Deb
Hey Dianne, how are you doing?
00:00:28.439 — Dianne Frommelt
Hey Deb, thanks for having me back! We’re going to talk about something we think about a lot—ethical, responsible use of AI, specifically in employee relations.
00:00:38.520 — Deb
Do you think we can cover that in two minutes?
00:00:42.280 — Dianne
I think we can. I mean, we could spend two hours, but sure—let’s try!
00:00:49.320 — Deb
All right, we’re going to try. I’ll cut us off if I have to but let’s at least start the conversation.
00:00:55.280 — Deb
So you all know the rules—and we really do try to stick to them. Okay, let’s jump in.
00:01:04.640 — Deb
AI and ER. When I started thinking about this—and Dianne, you and I have talked about it—my first thought was: what makes a good investigation?
00:01:17.479 — Deb
One of the biggest pitfalls in investigations, even without technology, is bias. And that’s also one of the biggest concerns with AI. So now, we’re layering the two—bias in investigations and bias in AI. It feels like a perfect storm and not the kind we want.
00:01:43.280 — Deb
At HR Acuity, we’ve taken a really intentional, thoughtful and strategic approach to how we’re incorporating AI into our product and how we frame its use for the broader employee relations community.
00:01:59.479 — Deb
Dianne, maybe you can talk about our approach in the product and get people thinking about how they’re using AI in their own work?
00:02:08.280 — Dianne
Yeah, for us, it really came down to identifying use cases where AI presents or summarizes facts—not draws conclusions.
00:02:19.640 — Dianne
We had a ton of ideas about automating the investigations experience but a lot of them raised concerns.
00:02:28.360 — Dianne
One thing we hear a lot from customers and users is: “Can you just generate the findings and conclusions for me? Can AI go through all the interview notes and pull out the key facts?”
00:02:40.040 — Dianne
Sure, AI can do that. But can we trust it not to imply things it shouldn’t? Not to introduce bias? Not to jump to the wrong conclusion?
00:02:51.640 — Dianne
These are still things a human needs to do.
00:02:56.680 — Deb
Okay, I came up with this analogy and we’re probably out of time but I’m going to share it anyway…
00:02:59.879 — Deb
Imagine you’re accused of murder. (I watch a lot of Dateline, so go with me here.) Both sides gather all the evidence and then they just throw it into an AI bot, and it spits out a verdict: guilty or innocent.
00:03:17.360 — Deb
That’s not okay.
00:03:20.959 — Dianne
No, we still need juries. We still need reasoning, logic and human assessment to make decisions like that.
00:03:28.200 — Dianne
That’s exactly why we’re taking such a methodical, thoughtful approach every time we build AI into the product.
00:03:39.560 — Deb
Exactly. What makes an investigator great is their ability to recognize nuance, corroborate facts and use judgment.
00:03:49.720 — Deb
I’m not ready to hand that over to a bot and that’s the mindset we bring to how we responsibly introduce AI into the ER process.
00:03:58.360 — Dianne
Absolutely.
00:04:00.280 — Deb
Okay, that was more like three minutes—and we barely scratched the surface. But let’s get the conversation going.
00:04:03.840 — Deb
What do you think? Let us know and we’ll come back and talk more.
00:04:06.879 — Deb
This is Two-Minute Matters and yes, we are way over time.
00:04:10.680 — Deb & Dianne
Thanks everyone—bye!
