AI in Hiring: Treat Every AI Agent Like Your Newest Hire

AI in Hiring: What Deb Paul Wants TA Leaders to Know

Deb Paul thinks most teams are chasing the wrong number with AI in hiring. They measure speed. She measures accountability, and in her experience speed follows once the setup is right.

Deb Paul comes from three generations of engineers. She picked HR to rebel against the family trade, then landed in recruiting because it was the part of HR she could measure. She started in sourcing, worked up through recruiting at large companies, and now advises TA teams as a fractional consultant. Her Substack and her LinkedIn posts cover what she calls AI accountable hiring: the laws, the lawsuits, and what good looks like.

That data habit shapes how she talks to the business. Jonathan Duarte raised it early in the episode. The CFO doesn't care about time to fill, because time to fill isn't a line on a P&L statement. Overtime paid out because a seat sat empty is. Deb's answer is to keep asking the business for its priorities and deliverables, then translate those into recruiting work instead of the other way around.

If you're building the business case for AI inside your own team, our guide to recruiting AI tools and use cases covers the tools Deb and Jonathan discuss here, including job description writing, matching and pre-screening.

Why Your AI Agent Is Your Newest Hire

Deb's simplest rule for AI in hiring is the one Jonathan said everyone should write down. Manage an AI agent exactly the way you manage a new employee. Nobody hires a person, hands them a laptop and walks away for six months.

Your AI tool is a new hire in your team. You don't hire the person and say, 'Go do your thing and show me the results.' You hire the person, you onboard it, you give them goals, you monitor them day to day, you have a one-on-one as a leader, you mentor them, you do their performance reviews, you take responsibility for the work that they have done, and you also take responsibility for the mistakes that they have done. That is your AI agent.

Deb Paul, Fractional TA Advisor

Jonathan's reaction was immediate: "Oh my God, we have to write that down." When teams tell him they have an agent that screens or sources, his first question is how they manage it. Are they doing one-on-ones? Deb gets the same reaction from clients. By the second question, she said, they make "that face that you just did."

What managing an AI agent looks like in practice

Every TA and HR leader already runs this structure for people. Deb wants it pointed at the software as well:

• Onboard it with the real job, not a job description copied from three years ago.
• Give it a defined outcome. As Jonathan put it, you can't automate anything if you don't know what the outcome is.
• Check its work every week, the same way you'd ask a recruiter what they found on a new requisition.
• Recalibrate with the hiring manager when the results miss.
• Name the person who owns its mistakes before go-live.

When Deb asks clients what their agent is trying to find, the answer by the second question is often some version of "I just looked at the percentage match." Nobody can manage a tool on that basis.

Safe, Legal, Auditable: A Simple Framework for AI Hiring Governance

Deb's framework for AI hiring governance has three words: safe, legal and auditable. A tool that fails any one of them isn't ready for your hiring process.

She keeps it short on purpose. Each state is writing its own AI law, and Deb argues they all ask the same thing underneath:

Whatever you're doing, is it legal? Are you able to explain it? Are you able to audit it six months later and say, 'Yes, this is why we took a decision.'

Deb Paul, Fractional TA Advisor

The person in that sentence matters. "We took a decision." The tool produced a recommendation, and a named human acted on it and can explain why. Deb's sharper version of the test is a candidate's question: "Why did your AI tool reject me?" If your team can't answer it, the setup isn't ready.

Questions every TA leader should be able to answer

• Which tools in our process touch candidate data or recommend who moves forward?
• Who is the responsible person at the end of each tool?
• If a candidate asks why they didn't advance, can we explain it in plain language?
• Could we reproduce that explanation six months from now, with records to back it up?
• When did someone last check the output for bias?

If those answers are fuzzy, your vendor contract is a good place to start. Jeff Pole made a similar case on an earlier GoHire Talks episode about audits and compliance questions to ask recruiting AI vendors.

How AI Bias in Recruiting Scales Faster Than Human Bias

AI bias in recruiting is human bias at machine speed. Deb is clear that bias in hiring isn't new. The difference is volume.

If a human bias is showing up, it shows up on 5% of cases or even sometimes more. But now you add AI to it, you just multiplied it by 300% because that's how fast it is.

Deb Paul, Fractional TA Advisor

Jonathan's summary was blunt. If nobody watches the tool for six months, "then that just means you have scaled bias."

He also shared a story from LinkedIn. A director of TA went back through six months of candidates her team had processed with AI support. She reviewed every resume blind, the way she'd normally source, without looking at what the AI had suggested. It was a toss-up. Neither the humans nor the AI were reliably better at moving the right people forward. Her takeaway was that someone has to own going back and validating the results, and Deb applauded her for doing it.

The legal risk is real. In Mobley v. Workday, a federal judge in California allowed a nationwide age-discrimination collective action to proceed in May 2025 over AI screening tools. For more on how AI-led interviews performed in a large field study, see our breakdown of the 70,000-interview AI bias study.

Deb isn't arguing for less AI in hiring. She calls her approach AI accountable hiring:

When I say AI accountable, I'm not saying no AI. It has to be there. It is important, but it has to be set up right. It has to be somewhere where somebody's checking for biases. There has to be a responsible person who is at the end of the tool saying, 'Hey, I'm responsible for this.'

Deb Paul, Fractional TA Advisor

Human in the Loop or Rubber Stamp? The Feedback Loop Most Teams Skip

Every vendor promises a human in the loop. Deb wants to know what that human is measured on, because targets went up when the tools arrived. This is where a lot of AI in hiring programs quietly fail.

A recruiter who used to carry three hires a month in a tough tech role is now expected to deliver six, because the team has AI agents. Under that pressure the easy path is to approve whatever the software recommends and keep moving. Deb describes the thinking: "Let my software take the decision. Let my software learn so that I can move fast and deliver to my goal."

But am I actually the human in the loop or I'm just rubber-stamping and moving it forward?

Deb Paul, Fractional TA Advisor

Where the recruiting feedback loop breaks

Jonathan described the old sourcing loop. Run the intake, build the Boolean search, bring results back to the hiring manager, ask "Are we close?", correct, go out again. That loop taught recruiters what the role needed.

Now a tool goes out and finds everything. So where does the feedback go? To the software, or to a recruiter? Deb's version of the problem is that the hiring manager often doesn't know what the percentage match means either, because nobody asked the follow-up questions.

We do that with our recruiters. Why don't we do that with our agents?

Deb Paul, Fractional TA Advisor

Jonathan put the risk in one line for listeners: "Maybe the software is learning, but your team's not." Deb's reply was two words. "Team's not." And a matching tool run against a poorly defined job description won't fix the job description.

AI Interview Scheduling: Speed Where It Counts

AI interview scheduling is where most teams save time fastest, because coordination eats so much of a recruiter's week. Deb puts the number high.

It's pretty much sometimes more than 50% of the time going into the coordination piece, into the last minute changes piece, because we are also dealing with humans and their priorities and things that go on in their lives.

Deb Paul, Fractional TA Advisor

Her caveat comes from her Substack piece, The Boring AI Pilot Is The Dangerous One. A scheduling tool that isn't set up properly will put a candidate's availability ahead of the business priority, and that's how teams lose the best candidate in the pool. Automate it, she says, but keep human intelligence driving it.

Jonathan shared GoHire client data that changes how teams think about availability. When frontline candidates could reply yes, no or a time by text, 49% of them scheduled their interviews after business hours. They're finishing a shift or sitting down after dinner. A recruiter calling between 9 and 5 never reaches them, and the best candidates won't wait for a slow process.

That's the case for automated interview scheduling over text. It meets candidates when they're free and gives recruiters back the hours they spent on back-and-forth.

Human-Led Beats Tech-Led: The Real AI Recruiting ROI

The cheapest approach to AI in hiring rarely delivers the best AI recruiting ROI. Deb pointed to what she recalled as a recent Deloitte study, from September 2026, comparing human-led and tech-led AI transformation in hiring. By her account, human-led companies saw about 1.6 times the return.

Her explanation is that cheap tools move the cost somewhere else:

It will not give you the best ROI. You are going to spend money somewhere else, which is maybe hiring, getting external agencies to work with, maybe having a hire leave earlier because the culture didn't match to what the tools said to the new hire.

Deb Paul, Fractional TA Advisor

Jonathan added an example from a previous guest. On an earlier episode, Master Burnett described a company that cut $2 million in recruiters from its budget, assuming automation would cover the gap. It then spent $12.8 million on recruiting fees. The team that made the call was fired.

The scoreboard that matters

Deb wants teams to be accountable for every decision in the process. That covers who interviews first, how interviews get scheduled, who gets the offer, and whether the hire fits the culture well enough to stay. Getting someone into the interview loop is only half of it. Keeping the talent longer is what justifies the hire, and cost per hire alone won't show you that.

If you're mapping out where AI should and shouldn't sit in your process, the Recruiting AI Adoption Playbook walks through the rollout one step at a time.

Where AI in Hiring Pays Off Right Now

The earliest wins for AI in hiring are administrative. Deb lists documentation, notes, tracking deliverables, prioritizing work and the repetitive tasks that eat a recruiter's day.

No recruiter comes into a recruiting talent role saying, 'Hey, I love making spreadsheets.'

Deb Paul, Fractional TA Advisor

Recruiters sign up to match skills to people and people to business goals. Take the admin off their plate and their efficiency goes up, which shows up in business outcomes too.

Deb won't hand the problem to vendors. In her view recruiting efficiency comes from how you set the tool up internally and how clearly you define what your recruiters use it for. Which product you buy matters less.

What changes for recruiters

Jonathan framed it as process change. Any time a screen changes in a software program, the process changes with it. He wrote about this two years ago in an article called You Can't Implement AI Like It's Workday, because AI isn't a set of rules you learn once in a training class. Sourcers who have always hacked their tools adapt fast. An HR business partner who recruits on the side has to keep learning.

Deb's prediction is pointed. Recruiters who were strong at tying talent to business goals before AI are still thriving. Recruiters whose value was being excellent note takers are struggling, because that work is being taken away. The predictions that recruiting teams would disappear by 2026 haven't held up either. Deb sees more recruiter roles opening now than a couple of years ago.

AI Hiring Compliance: What the New Laws Already Ask For

AI hiring compliance comes down to the same three questions in Deb's framework. Is it legal, can you explain it, and can you audit it later? Responsible AI in recruiting means your team can answer all three before a regulator or a candidate asks.

The rules already on the books point the same way. Under federal law, the Uniform Guidelines on Employee Selection Procedures used by the EEOC apply to any selection procedure, including software, and set out the four-fifths rule for spotting adverse impact.

New York City's Local Law 144 requires bias audits and candidate notices for automated employment decision tools. Enforcement has been light so far. A December 2025 audit by the New York State Comptroller found the city received only two complaints, while auditors identified at least 17 instances of potential non-compliance. Employers shouldn't read that as a safe harbor.

Illinois HB 3773 amends the Illinois Human Rights Act so employers can't use AI that has a discriminatory effect in hiring decisions, effective January 1, 2026 (Epstein Becker Green summary). Colorado's AI Act (SB 24-205) treats AI used in employment decisions as high-risk and requires risk management and impact assessments. Its start date has already been pushed back once (Greenberg Traurig alert), so confirm the current date with counsel.

Adoption is moving faster than governance. SHRM's 2025 Talent Trends research found AI use in HR tasks climbed to 43% in 2025, up from 26% in 2024, while 69% of organizations still struggle to fill full-time roles. For the data-privacy side of the same question, see our post on AI privacy and compliance in HR.

GoHire's position is simple. Software never makes a hiring decision. Your team writes the questions, sets the criteria and decides who moves forward.

Frequently Asked Questions About AI in Hiring

AI in Hiring: The GoHire Talks Interview Transcript with Deb Paul

Deb's Path From an Engineering Family to Recruiting (00:00)

Jonathan Duarte: [00:00:00] Hi, everyone. Welcome to another show of GoHire Talks. I have a guest today, and I think very timely, and this is probably an episode people are gonna be watching for years to come. I'd like to introduce Deb Paul. Deb has been in recruiting, started from the sourcing all the way through, all the way up the ranks and has been in big companies.

But she's now working as a fractional advisor and consultant. And one of our big topics for today, because she's been talking about it on LinkedIn and on her Substack, is AI governance versus speed, and really understanding what you're getting when you buy one of these tools or you're considering one of these tools.

So Deb, fill in the gaps. How'd you get involved in recruiting and all the way into AI these days?

Deb Paul: It's interesting. So I come from three generations of engineers in my family- Yeah ... and the best part about all of this is I, I'm a rebel. I always try to do something different than engineering, so I picked up [00:01:00] HR.

And, the universe had planned for me to put me in that part of HR which is numbers deliverables driven, which can be measured. So I did use the tech and data mindset from where I come from to humans. I just didn't put it to machines, I put it to humans. So that's what I do.

That's how I came into recruiting because I could measure the results, I could see what the goals are, and they were all data-driven. So that kind of drew me into recruiting.

Jonathan Duarte: Yeah. Now, and this will be great because we're gonna talk a lot about this. But I want you to think, while we're going through here, think about the data piece because I think that's one of the biggest pain point for TA leaders is the analytics.

But it's also from not the TA side, but the CEO and CFO's point is the language of recruiting is not the language of the business. The business does not care about time to fill because that's, you can't [00:02:00] write... That's not a P&L statement. But what is, is that if you want to get a project funded by the CFO- Tell 'em the profit and loss, how much time is being spent, how much effort is being spent, how much overtime is being paid out because we don't have the right se- the right butt in the right seat at the right time.

So when we keep those analytics conversations, I like to frame those in business terms- ... not just the recruiting terms, which we all know. But I think that's great because it's really hard sometimes for people to get the analytics-

Jonathan Duarte: And then be able to translate that into business speak.

Deb Paul: And I think it's important, that's why, to constantly keep asking the business what your priorities are, what your deliverables are, and how is it looking, how does it translate into recruiting versus the other way around.

Interview Scheduling and Validating AI Results (03:00)

Jonathan Duarte: Yeah. Let's jump into your last Substack article, and we'll put these in the notes [00:03:00] here how to get to your Talent Realist Substack.

But the title is The Boring AI Pilot Is The Dangerous One. Yep. Talk a little bit more about that and, overall we'll just dive right in deep to it.

Deb Paul: Yeah. So for me, the key thing is everybody, the first thing in recruiting when this whole AI revolution started to come in is "Oh, just... Let's just change the scheduling.

Let's just add more AI to scheduling and let it do," because that takes a lot of the time when it comes to recruiting process. But the key piece that gets forgotten is when you're scheduling 10 candidates together, how do we schedule based on the candidate's availability versus business priority versus, hiring manager's availability and all of that?

And that boring tool actually prioritizes somebody's availability versus the business priority if we have not set it properly. And that is why I say that tool that is the most obvious one in recruiting that we want [00:04:00] to automate is also making some decision or sometimes making some mistakes where we actually lose out the best candidate in the pool.

Jonathan Duarte: Yep. Interview scheduling is one of my area of, what we've been building for the last 10 years. Where over your history, I've heard stats, and it all depends on the type of role. How much time do recruiters spend with the back and forth of scheduling?

And maybe it depends if it's professional versus frontline. I'm sure, it's a little bit different. But in your estimate and your experience, where's that time going and how much time does it take?

Deb Paul: I think the key piece that... for a recruiter or for a recruiting coordinator, it's pretty much sometimes more than 50% of the time going into the coordination piece, into the last minute changes piece, because we are also dealing with humans and their priorities and things that go on in their lives. So I think it [00:05:00] does take a big chunk More than 50% of their work to just schedule, reschedule, change the timings, and how we put this in a place where, yes, it is automated, but there is something else that's driving it, which is also the human intelligence telling it.

Yes, if the candidate is not available, how do we figure out what is the other best timings for both these people to be available so that we are driving the business interest as well?

Jonathan Duarte: Yeah. And we'll talk offline about some advanced ways to do that because there are tools out there.

A lot of people just don't know. They're still trying to do it manually. Which at this point, manually doing it without some assistance, is just it's painful. And that's all I can say. It is probably, to me,

it's the biggest area you can improve your business.

By responding to the best candidates as quickly as possible because we always know the best candidates are not gonna wait for your slow process. They're just gonna move on. And I remember I've been [00:06:00] on panel interviews where, I was one of the members to interview, and then someone else had to go out and do something else, and I jumped in on somebody else's panel.

It's that last bit is that human part that has to be there, right? But there's that how do you engage them the first time? How do you understand their schedule? We found out recently with some of our frontline clients that when we text candidates and they can just say yes, no, what time, et cetera, make it really simple 49% of their candidates were scheduling after business hours.

Deb Paul: Yes.

Jonathan Duarte: And the best way of saying this is that if you understand your candidates and they're doing this after hours, that also means that you as a human are not going to be reaching those candidates on the phone or email because they need to do it when they have a chance. Maybe it's after dinner.

Maybe it's, their shift ends. They can't pick up the phone, especially in the front line. They [00:07:00] can't pick the phone up while they're on shift. So there's, that's the simplicity that makes it easy. But as we will talk about some of the failures of jumping in and getting something done quick there's failures that we need to address and be able to measure again.

So- I saw a post on LinkedIn. I have to reach out to the, this individual.

She was a director of TA, and she went back to all the candidates that they interviewed over the last, I'd say six months- and looked at their resumes, went through the whole process "Why did we decide this and that?" And they had been using AI as well. And she said I'm just gonna blind look at the candidate just like I would source them normally," and then not knowing what the AI came up with, and looked at all the candidates that applied for those roles, and then just measured them and say what worked and what [00:08:00] didn't.

And she said it was a complete toss-up. It was we as humans still are not good at any of this. It's still a, "Hey, we were able to get somebody on the phone, so they moved forward," or a lot of things happened, and the AI didn't do much better than the human in the process of getting better people in 'cause those people sometimes, depending on the role, depending on the type of position, some people moved forward faster because of the engagement.

Other people just stalled out at the interview, and the numbers were almost equal. But she said, "At least we're trying. We're gonna try, but w-," she goes, "It's gotta be part of my job to go back or someone's job to go back and make sure we're validating this stuff."

AI Accountable Hiring and Scaled Bias (08:00)

Deb Paul: No that's absolutely what my framework talks about.

I always say that, when I say AI accountable, I'm not saying no AI. Actually, yes, it [00:09:00] has to be there. It is important, but it has to be set up right. It has to be somewhere where somebody's checking for biases. There has to be a responsible person who is at the end of the tool saying, "Hey, I'm responsible for this."

I think that's a very big piece of what today's TA world needs to be looking at. And one thing I always keep talking about this, like biases are there in recruiting. It has been there in the past, it's there today. That's normal. But then if a human bias is showing up, it shows up on 5% of cases or even a even sometimes more.

But now you add AI to it, you just multiplied it that by 300% because that's how fast it is. That's how fast we are claiming it is. Yes, there are biases in human recruiting as well, but when you transfer that, and if it's not done properly, and if it's transferred to any of those tools, you are just multiplying it way more.

[00:10:00] And I'm-- I applaud this TA leader that she took that stance. Yes, we did it differently and humans might have their own biases as well, but the tool that sh- they have been using may have done it 100 times more.

Jonathan Duarte: Yeah. And that's that thing. If you just let it go and then you got no one's watching it six months later and it's scaled faster, it's gonna be able to do it much faster than a human.

And then that just means you have scaled bias.

Deb Paul: That's what it is. And I think you and me both know there are some of these big tech companies, HR tech companies facing some of these bigger lawsuits.

They just got rejected by algorithm bias.

Jonathan Duarte: Yeah.

Deb Paul: That's what it

Jonathan Duarte: Give me your framework and tailor it in this sense. There's so many different AI tools out there, from AI job description writing to matching candidates to then pre-screening candidates and then AI interviewers

like the tech is changing incredibly quickly, but the underlying [00:11:00] is it right for these roles? Not just is it the fastest way to get candidates through a pipeline. So how does your framework work in that?

Safe, Legal, Auditable: Your AI Agent Is a New Hire (11:00)

Deb Paul: My framework is actually pretty simple. It's not like a complicated thing to think about.

It's like AI in hiring should be safe, legal, and auditable. That's all. Now the new AI laws that's coming in, each states are coming up with their own, that is all they're asking. "Hey, whatever you're doing, is it legal? Are you able to explain it? Are you able to audit it six months later and say, 'Yes, this is why we took a decision.'?"

If somebody comes and asks you, "Why did your AI tool reject me?" Are you able to say that? That's the simple framework I work with. And another simple language that I say is your AI tool is a new hire in your team. You don't hire the person and say, "Go do your thing and, show me the results."

You hire the person, you onboard it, [00:12:00] you give them goals, you monitor them day to day, you have a one-on-one as a leader, you mentor them, you do their performance reviews, you take responsibility for the work that they have done, and you also take responsibility for the mistakes that they have done. That is your AI agent.

That is important for every leader to understand.

Jonathan Duarte: Oh my God, we have to write that down bec- that is it seems so strange because everyone's like I have a AI agent that does this, that, and the other." I go how do you manage them? Are you doing one-on-ones?"

Which is so simple if you think because we already have the structure. We all know anyone in TA, anyone in HR has a structure. You have a one-on-one because you want to see where they are, and if they were an employee, they have questions about their career. They have questions, "What do you want me to specifically do?"

They have questions about, "Am I doing it correctly? And what else do you want me to do?" That's, the human things. But as the manager, it's the same thing. "This is what I want you to do." You cannot have an agent or you can't [00:13:00] automate anything if you don't know what the outcome is

Deb Paul: That is

Jonathan Duarte: so true.

And yeah, so I think that's a really interesting perspective, and I don't hear many people talking about it that way. I would love to see you interview one of your agents. Or you sit there with one of your clients and they go what's the first thing you want to do?"

I want to screen resumes." "Okay. What do you want to screen it for? Okay, now let's... it's been working for, a week. How'd it do?" What did it do?" Yeah. So have you walked through that with clients?

Deb Paul: And, trust me, by the second question, it is that face that you just did. Like-

Jonathan Duarte: Yeah ...

Deb Paul: I have a very simple question for it. My question is don't you do this with your own recruiter when you open a new requisition, and the recruiter is working on it for a week with the hiring manager, and you have a one-on-one with your recruiter?

You go back and say, "Hey, what did you do with that particular [00:14:00] requisition?" He's like, "Oh, I looked at the job description, I found five resumes." Did it click? No. Were you looking for the right thing? Do you w- want to go back to the hiring manager and recalibrate? We do that with our recruiters. Why don't we do that with our agents?

We should be doing that with our agents as well. And asking that question, by the second question what is your agent trying to find, it is oh I didn't know if they were finding the right candidates. I just looked at the percentage match. Yeah ... even the hiring manager didn't know the percentage match because we did not ask the follow-up questions to the hiring manager, and that's where the human aspect comes to it.

Jonathan Duarte: That's interesting 'cause I haven't talked to that many clients that have been implementing matching and then surveying the hiring manager based on the match. I have teams that are, using it to match, then still walking through the [00:15:00] match to top list it, to short list, so they're still involved.

But I don't recall that I've talked to anyone who is just using one of these tools and sending it to hiring manager, and hiring manager is then... there has to be some sort of feedback loop

Feedback Loops and the Human in the Loop (15:00)

Deb Paul: And that's where it is sometimes, and this is where the 2026 what we are seeing the trend is what was expected out of these recruiting tech agents or tools to be doing, it's not.

It's not that we have moved faster in recruiting. It's not that we are doing it way more efficiently. That's not showing up, and that's the question. And there was all of these predictions about recruiting teams are going to go away by 2026. And all of those predictions, no, there are actually more recruiter roles opening up right now- versus couple of years ago when the predictions happened. Why? Because maybe that's what you need in the process. It's like you said, it's not about the speed. It is about [00:16:00] what the outcome is trying to be, what we are trying to go get to.

Jonathan Duarte: Yeah. And I think it... The speed might be Good to be able to get your feedback loop faster.

For instance, we do the intake. Then we say, "Okay, we want these sourcing things." Normally that would be create the Boolean search, throw it against LinkedIn, whatever our sourcing tools, come back, bring it to the manager and say, "Are we close?" Like, where are we? What am I missing?

Et cetera. And then re-correct, go out again. And now we just have a tool that just goes out and finds everything, lots of these tools out there comes back and has... But where's that feedback loop? Is that feedback loop going to the software and not to a recruiter? Or, and is that software actually doing it correctly?

I think that's that part that I haven't heard much of. Maybe this would be the question even, [00:17:00] for our listeners is-

Jonathan Duarte: How are you doing that? Because if you're just source- outsourcing it to the software, are you guys actually learning anything? Maybe the software is learning, but your team's not.

Deb Paul: Team's not. And I think that's another piece to it. When these... So there's a lot of... I'm sure you heard a lot of it about human in the loop. Oh, a tool doesn't- Yeah ... make a decision, it's a human in the loop. Yes, but the goals for that human in the loop, earlier it was like three hires in a tough tech role in a month, it's probably become six hires because you have AI agents you should be doing, using and all of that.

But then is that human in the loop just rubber-stamping or actually looking? Because they're still, looking at their goals. Every recruiter is like, "Oh, I just need to do six hires that I'm supposed to do. Let my software take the decision. Let my software learn so that I can move fast and deliver to my goal."

But am I actually the human in the loop or I'm just rubber-stamping and moving it forward?

Jonathan Duarte: Yeah, 'cause if the recruiter's not asking [00:18:00] the right questions back to the hiring manager, the chances are the software's certainly not doing it. The whole th- like we all know this, the whole thing is help the manager define the role.

Yeah. Help the manager define the skill set. That's usually the number one problem. And we kn- we've known this for years. Job descriptions usually don't actually describe the actual role, describe how much JavaScript the person's gonna do, or they just need to spell JavaScript because it's something that, they should probably know.

But it's not a requirement, right? Yeah. So I think it almost leads me to believe that a lot of these tools, while people are doing a lot of investment in them there's The fallacy that you could take a job description and run AI on a poorly defined job description is somehow going to match, I think we all intellectually think that's not going to work.[00:19:00]

But yet the CFO doesn't care, and it's cheaper, and that means recruiters can have 400, placements on their head rather than 100.

Deb Paul: Yes. That's the key piece, and that's- ... where the thing is are we solving for the speed, accuracy, and efficiency, or just the, m- the money part of it?

And we don't know what it is, and that's where I say that let's be accountable for all the decisions we take, whether it is who interviews first, how we schedule the interviews versus who we give the offer out, and is the candidate matching up to the company culture, organization culture so we are able to retain the talent longer.

So it's not just about getting somebody into the interview loop, it's also about keeping the talent longer so that we are justifying the hire that we made.

Human-Led vs. Tech-Led AI in Hiring ROI (19:00)

Jonathan Duarte: Here's an interesting question that just came to me. The companies who are trying to implement a lot of this stuff at scale probably run predominantly on cost.

Like they're saying, "Let's... We need to simplify this thing, and we have to cut our [00:20:00] costs." Those are probably not the same companies that are calling you, because you're expensive to set up strategy. They're just looking at implementing a tactic. So my guess is the people who are looking to reduce costs aren't getting the strategy, and they're probably going to start calling staffing a recruiting team, outsource recruiting again, and not do it internally, and not solve any of their actual cost problems.

'Cause retention and everything else is gonna go down.

Deb Paul: That- that's true, and that's where it is. There was, I think it was a study done by Deloitte and it was a recent, September 2026. They said the companies which are leading the AI transformation and hiring, which is human-led, which is- human first versus a tech-led.

There is a higher chances, 1.6x ROI on the investment that's gone on, going on in AI hiring. [00:21:00] Results for human-led is way higher, 1.6 times more higher than the tech-led, and that's important piece like how are we getting the tools which work with humans versus- how are we getting the tools which is cheapest and gives us the best ROI? It will not give you the best ROI. You are going to spend money somewhere else, which is maybe hiring, getting external agencies to work with, maybe getting having a hire leave vi- earlier because the culture didn't match to what the tools said to the new hire.

The cost anyway goes higher. It might not be a hiring tool, but something else.

Jonathan Duarte: Yeah. Yeah, it's interesting. I had a recent guest on, Master Burnett, and Master mentioned a company... Very clear we didn't wanna mention names, but the company got rid of 2 million in their recruiting budget for recruiters.

Ended up spending 12.8 million on recruiting [00:22:00] fees-

Deb Paul: There

Jonathan Duarte: you go ... because they thought they could just automate through the process, and it cost them, six times? And just because they were like penny-pinching, they wanted to go do it their own way, and that whole team got fired.

Deb Paul: There you go.

Jonathan Duarte: Yeah. There

Deb Paul: And this is where the thing is It has to be with human. It's great to drive those technology. I'm all for it, but it has to be with human along with.

Where AI in Hiring Works Today (22:00)

Jonathan Duarte: I think that one of the areas that people ask me, "What, where's AI working the best now?"

Let me ask you that question, Where do you think AI in hiring anywhere in the process where do you think it's most effective right now? And, most effective depends on, of course, the kind of hiring, everything else too, but where do you see most companies getting the earliest gains?

Deb Paul: I think again, in recruiting there is so much of administrative and repetitive processes which recruiters did put in a lot of their- ... efforts and time. That's actually [00:23:00] great with AI, like the documentation the notes, the keeping track of what your deliverables are prioritizing all of that, making sure that, things are being done on time, repetitive work- has been taken care of, the administrative work's been taken care of, and it frees up a recruiter to do the work that they actually signed up for. No recruiter comes into a recruiting talent role saying, "Hey, I love making spreadsheets, and I like putting in I spoke to five candidates" in a spreadsheet.

No. They come in because they like matching the skills to the resumes, to the candidate, and connecting them with the goals. So as long as they're able to do that and not putting in efforts in repetitive work, and that is taking care of a lot of the AI, that increases their efficiency, which in results increases the business outcome as well.

Jonathan Duarte: And not pitching names and things, but are there software systems that you've seen in certain areas? Like ATSs are coming along. Some [00:24:00] are more advanced on this stuff. Workday's may... I think Workday makes more press release announcements than actual software changes. And I understand it's a big company, and there's a lot of reasons for that too.

But have you seen a lot of AI just tactically where the recruiters are gaining that time back?

Deb Paul: I think it's not about the tooling and the companies, it is about- ... how it's set up internally to use those tools. And I think that is where I always I'm harping on. It's not just on the tech vendors, HR tech or recruiting tech vendors to solve your recruiting efficiency.

It is up to you how you're using it and how you're defining the scope of how your recruiter is using it. That is the key piece,

they're coming up with technology. But how we use them is the key piece. I think everybody's trying to do the right thing.

Jonathan Duarte: Yeah, I like to say any time you change a screen on a software [00:25:00] program- It's process automation. It's process change.

Deb Paul: Yes. Yes.

Jonathan Duarte: And there is nothing... I wrote an article, I think it's now two years old called You Can't Implement AI Like It's Workday.

'Cause it, it's not a set of rules. It's not a just take the manual take a class, and then you have it done. And it's, I th- it's the same thing as any kind of upgrade on a software program that moves menus around is process change. And for some people who are non-technical, they learn the process, they learn their CRM, they learn the buttons they've gotta push and where they have to enter the stuff.

And then after that it's like, "Oh my gosh, like I can't... We're changing again." "Oh, where do we do that?" Now, you come from a sourcer background so obviously tech is not a problem because good sourcers are always hackers of some sort of another, right? And so I think those, being a sourcer almost ki- if you were [00:26:00] a sourcer of a technical side you're always digging in.

"Okay, how do I do this a little bit better? Where do I find these email addresses?" All these different tools. That comes from a different s- set that an HR business partner or an HR regional manager who has to do recruiting, sitting behind the computer and learning their ATS is not their primary objective in life.

Absolutely. So they have to keep learning. Yeah.

Deb Paul: Yeah.

Jonathan Duarte: Yeah.

Deb Paul: And again, I say that, people in talent are truly those folks who are actually always trying to figure out the best possible way to help business goals. A successful recruiter is a recruiter who is able to define how they find talent to the business goals, and that's why it's really important for us to define that aspect of it, AI or no AI.

I think the recruiters which did that before really [00:27:00] well are probably again still continuing to do that today itself. Recruiters who thrived on administrative work because they were really good note takers are probably not. They are struggling because that's, work, that's taken away from that.

Jonathan Duarte: Yeah. I think that's a great way to cap the conversation because that's so true is that human part of connecting, finding the right people, not just who can, schedule interviews faster. That's not where we get the most value out of recruiting.

'Cause you can recruit a bunch of people faster through the system, doesn't mean they're the right people in the right seat at the right time.

Deb Paul: Yep.

That's true.

How to Connect With Deb Paul (27:00)

Jonathan Duarte: Thank you so much. How does someone find you, Deb?

Deb Paul: Please connect with me on LinkedIn for sure. I write quite a bit on LinkedIn as well as Substack.

I have a newsletter coming out every other week. I write a lot of posts. My main key piece is to talk about AI accountable hiring. I talk about the laws, the lawsuits, what good looks like coming [00:28:00] in today, which would be super helpful for anybody to read and reach out to me. What I do is I can come in, look at your concurrent process in your company, give you the guardrails, help you with making your hiring and talent hiring process more AI accountable.

Reach out to me on LinkedIn, there is links to connect with me on Calendly and set up some time with me to talk to me. I'm happy to do that, and I also throw in some extra ideas on advice on how you want to improve the process if that is what you want to do. I'll definitely do that for the first 30 minutes for free.

Jonathan Duarte: Thank you so much, Deb, and I hope everyone got a lot of information about this AI versus human, being still the master of the whole process because ultimately we're trying to do it to speed up so we can get the right people in the right seats at the right time, but we can't outsource the thinking of the company culture, 'cause I've never heard [00:29:00] of an AI that can define your culture and the hiring manager team for you from a bad job description.

Deb Paul: No.

Jonathan Duarte: All right. Thank you so much.

Deb Paul: Thanks, Jonathan.

Outro: Thanks for spending your time here with GoHire Talks and listening to our guests. My intention has always been to create incredible content to help you in your career in recruiting, marketing, business, and leadership. We appreciate all of the insights and feedback you provide. If you've got guests you'd like us to have on the show, please let us know.

Again, thank you for all your support