Is your TA transformation on track?
What’s the Right AI-Human Ratio in Talent Acquisition?
AI is accelerating, but hiring is emotional, high-risk and human. As we navigate through economic challenges and a modern skill shortage, the questions around using AI in the talent acquisition process have never demanded clearer answers. Does your TA team have the right combination of AI and the people who govern, guide and improve those tools? What parts of the hiring process are a perfect fit for AI and which parts require a human being?
Ultimately, the real question for TA leaders is what is the perfect blend of machine learning tools and the people who will operate them and use them in the hiring lifecycle?
In this AMS Catalyst Interview, we asked two of our AMS thought leaders to debate the use of AI in the hiring process and what TA leaders need to remember when deploying these tools. As AMS’ AI Strategy, Governance & Engineering Director, Matt Poole will share his advocacy of AI in the TA office while AMS’ Senior Director of Retail, Hospitality, Leisure and Consumer Annie Hammer urges a more cautious approach in adopting AI for hiring talent. They also share their thoughts on why humans will have a role in working alongside AI, how it can find hidden skills, and the perfect AI-Human ratio.
Are TA leaders aware of the challenges of using AI in the hiring process?
Matt Poole: Long predating my recruitment 20-year career, people have struggled to give candidates the right experience and to give customers the outcomes that matter to them. That’s been challenging because industrializing recruitment at scale is hard work. TA teams typically get bogged down in the less valuable parts of the work that needs to be done. Whether it’s administration, highly repetitive tasks or clicking things through a queue, the human is often not being stretched and used to the best of their capabilities. AI is a great enabler of moving humans up the value chain into more impactful work. If we can do that, then we should create better experiences for everyone.
Annie, why is Matthew completely wrong?
Annie Hammer: [Laughs] Well, he’s not completely wrong. AI in general is incredibly valuable for TA teams because it removes friction from the system and protects recruiters’ time and energy so they can do work that only humans could do well.
While AI can help with repetitive tasks, there are certain aspects of TA where AI won’t understand the full potential and the context where candidates might have non-linear paths. Candidates might not present perfectly on paper, resumes or applications. You might have people who have moved careers, taken a leave of absence, or they’re frontline workers who don’t necessarily know how to set up their application.
In those cases, a human element might be able to help. You can still use AI but humans overseeing the hiring can spot transferable skills, an individual’s motivation or their learning agility to be able to assess a person from those bases. That’s why it’s important to have a human review the AI’s work, especially when some of these decisions affect an individual’s livelihood.
Does TA want AI in the background or performing the initial scanning of the application and resume? Would TAs want a candidate to interact with AI in the entire hiring process?
Matt Poole: Since Gen AI has become mainstream in the past three years, it continues to grow exponentially. You don’t have to zoom very far into the future to think it’s not going to be a capability reason that we don’t use AI models for other tasks. Maybe we want people to have a particular type of experience, and that experience might be people-focused, but as a proponent of AI’s capabilities, we are on a fast train and there are not many stations to get off right now. What we have to decide is how to use the technology sensibly, safely and ethically and constrain ourselves for other reasons than AI can’t do it. That can’t be the reason not to do something.
If TA leaders use AI in initial screening, when in the process do you hand that over to a human being?
Annie Hammer: AI can provide fast responses, clear instructions, easy scheduling, and consistent communication. Where you should think about either handing it over to a human or a human getting involved is points in the process where candidates really want to feel seen, heard, and valued. If there’s a point in the process where it could be confusing to navigate or seems cold, that’s where you need to have a human involved and make sure that that trust is built.
Giving a person feedback, especially if the feedback is maybe not just a sort of yes or no, but there’s more context to give, that should come from a person. It’s better for a human to deliver that when explaining a decision that they might not be moving forward in the interview process, but they might be a fit for a future role, that would be best coming from a person. At that point I would say that’s where the human touch needs to be involved.
Matt Poole: We have to remember that recruitment is about selling people to people. When you look at it through that lens, can you automate all of that process? AI is very powerful, but it is not a one-size-fits-all tool.
What about things AI cannot do at the moment, such as offer a sense of corporate culture? A candidate may be impressed that the TA and HR teams have impressive tools but they won’t be able to answer why there’s a job opening in the first place.
Matt Poole: I don’t think AI actually is the best delivery mechanism for what you’re describing, that kind of authentic culture of the employer but you could use AI in a different way. You could use AI to locate talent across your business. Where are the culture carriers? How do I take those culture carriers and bottle something about their experience in a way that isn’t very scalable? Because I can’t have everyone do a one-to-one interview with a prospective joiner.
What about finding skills? AI can scour an application to find those skills quickly but most candidates are often unaware of the total number they possess. Can AI help candidates recognize the skills even if they don’t know on their own?
Matt Poole: Absolutely. AI excels in the relationship between things that are similar but not the same. A candidate may have experience in customer service management, but that skill might be categorized under ‘stakeholder engagement.’ Are they coming from the same place with the same core competencies? Probably.
Annie, is the search for skills something that a human manager might be a bit better than AI?
Annie Hammer: Candidates don’t always know their skills. They’re sometimes under- or misrepresenting themselves in terms of what they’re putting down on paper. This is where humans could come in, and actually AI is capable of doing this, but I don’t think all humans are yet capable of using AI to do this to help them. Humans can see the capability beyond the labels, and I think AI is capable of doing that, but it requires a human to prompt AI to help do that. That’s probably where the gap is more so than AI not being able to do it. It’s actually the human not knowing that I should use AI to help draw out those hidden strengths or skills, or help use AI to help me better represent myself.
What is the ideal AI human blend? Is it 80-20? When will we reach 50- 50?
Matt Poole: What a question. I could take the cop-out answer and say it is really what’s effective in any given situation depending on the job’s tasks. If we take a typical TA process, I’ve not seen anything that’s particularly effective that is 70, 80% AI led today.
As a proponent of AI, I would always advocate for its capabilities, but it doesn’t get you past the compliance hurdles and the governance guardrails. I would say most models today are like 30 to 40% AI up or down.
Do I feel that’s right for 2026? I don’t think we’ll stay here for long honestly. You only have to look at the progression of other industries where there’s less sensitive commodities than people at play. Those are moving at a much faster clip.
Annie, what’s your ideal AI human blend?
Annie Hammer: The delineation is probably a bit around the hiring type and the geography. If you take something like frontline hiring, you’ll have processes that today are already at 60% AI and automation led. I could think of one company in particular that’s probably at about 90 to 95% in AI usage, but the problem that they have is they can’t get the process really smooth and frictionless. People get their offer, they get the job, but then they have problems with people actually showing up on day one because they never actually built trust with a human. I’m showing up at a job in the middle of the night in a parking lot that’s dark. Candidates think, “This is scary, I’m going home.”
To take that 60% in frontline hiring and put it to 80 or 90%, you’ll have downstream negative impacts on your business because of things like that. On the professional side that requires maybe more experienced hiring. Now it’s probably 30% closer to where Matt was thinking.
What’s the most interesting question that you get from potential clients? Are they aware of this human-AI blend?
Annie Hammer: Most frequently clients are asking, what tool should I buy? The other question is more around what level of risk will I be introducing to the business if I do this or if I don’t do this correctly? They are weighing the risks and the benefits to decide which way they should go.
Matt Poole: The other day I was meeting with a risk modeling team at a financial services company, and when they assess AI, they assess it through the lens of all the other AI in their organization. For them, AI is doing balance-sheet calculations of customers, predictive financials of billing models and more, and this kind of TA thing comes along and they’re asking, how do you make sure that the transcript from the interview is accurate?
It’s such a simple question. You are deploying this technology in a predominantly human commodity process where you are trying to hire people, not objects. When you talk to people, they have feelings, good days and bad days, and unique experiences. Trying to answer a question about something as simple as how do you prove what they’re saying to you is true is really hard, even if the AI model is the most capable AI on the planet.
It sounds like AI FOMO or fear-of-missing-out is real?
Matt Poole: There is still a lack of understanding of just how hard it is to get this right in TA. People see the headlines around AI in the workplace and think, “I should be doing more.”
Matt Poole: Map your process honestly. Look at your hiring workflow end to end and be honest about each step. Is it AI-led, human-led, or a grey area that nobody really owns? In my experience, the grey areas are where candidate experience falls apart. Not because the technology let you down, but because nobody decided who was responsible.”
Find your trust moments. Think about the two or three points in your process where a candidate is making an emotional decision. Whether to apply, whether to stay engaged, whether to actually show up on day one. Those are the moments where human involvement isn't optional and we should be building on those, scaling elsewhere.”
Test your funnel and metrics before you scale them. Before you push AI further into your process, look at what's actually happening downstream. Are your offer-to-start rates holding? Is early attrition stable? A blend that looks efficient on a dashboard can be losing you value in places you're not measuring yet, once you look under the surface.
More than anything, identify the problems and be led that way, don’t look for places and ways to apply AI.
Annie Hammer: Define what ‘great’ looks like before you try to optimize it.
Be clear on the experience you want to create for candidates and hiring managers in an AI-enabled process. If you don’t have a clear view of the end state, it becomes very easy to automate inefficiencies instead of removing them.
Assign accountability, not just ownership of steps.
AI-enabled hiring doesn’t sit neatly within TA, HR, or technology. Someone needs to be responsible for how the whole system performs, not just individual components. Without that, even well-designed processes drift over time.
Assess whether your team is equipped for the model you are building.
Blending AI and human decision-making requires different skills, from data literacy to judgment and stakeholder management. The gap is often not in the technology itself, but in how confidently teams can use it.
If you believe that your TA team would be doing more with AI, AMS can help. We work with clients who are looking to bring artificial intelligence and more automation to the recruiting and hiring process. Our experts work with leading global enterprises to redesign AI workflows, not just buy tools. AMS helps forward thinking firms combine AI and HR together to hire the next set of world-class employees thanks to our next-generation talent acquisition solution. Contact us today for an appointment or a demo.
© 2025 Alexander Mann Solutions (AMS)
