Human judgment remains critical as AI reshapes hiring, onboarding

June 10, 2025

Todd Humber

Organizations using AI in human resources should view artificial intelligence as “a team member” that can handle specific functions while recognizing its limitations, says Lisa Haydon, founder and CEO of Halifax-based Pivotal Growth.
“We know that AI can get to about 80 per cent accuracy if you’ve done a decent job with your prompts,” she notes.
But the human touch is needed to close the remaining 20 per cent, she added — a critical point that highlights the potential, and the constraints, of the rapidly evolving technology.

HCM Podcast

Produced with Google Notebook LM Using AI Narration

Recruitment and talent screening

Haydon has seen firsthand how AI is transforming recruitment processes. For example, she helped a client build a custom talent screening tool that used the technology to go beyond basic resumé analysis.

“The resumé gets at the hard skills — what’s your education, what are your credentials, what’s your experience,” Haydon says. “But there were these aspects and attributes and things about the leadership that they really wanted to bring out faster from the resumés.”

This approach delivered significant benefits in both efficiency and effectiveness, she notes.

“It allows recruiters to identify candidates’ strengths and potential limitations much faster, without investing countless hours in the screening process.”

Using AI, Haydon’s team also created customized interview question banks for clients.

“Interview questions are so hard to get right, we don’t do it well,” she says, adding that traditional queries are “are not specific at really drawing out candidates” and “may bring forward biases.”

But what AI absolutely cannot do is assess people for roles that require interpersonal skills or gauge their leadership potential — that is still a uniquely human call.

“If there is that ‘How are they going to align with the team?’ ‘What are their interpersonal capabilities?’ ‘Are they going to lead a team?’ — that’s where I think that you wouldn’t want to use an AI bot. You need that human judgment.”

AI’s broader benefits in HR

Julie Develin, senior partner for human insights at UKG in Lowell, Mass., sees benefits in using it throughout the entire employee experience.

“AI can be leveraged throughout that entire employee journey,” Develin says. “You can use AI to streamline recruiting, to identify candidates who might have more potential than other candidates. You can use it to do things to provide personalized experiences for employees through the onboarding process.”

This personalization aligns with shifting employee expectations.

“When we talk about what employee expectations are today, employees expect to have that unique and personalized experience, and they expect the organization to see them as more than a worker — they expect to see it as a human.”

Develin notes that AI tools can help with specific HR challenges, such as creating accurate job descriptions and developing effective interview questions.

“Sometimes that’s one of the hardest things for organizations to figure out. Are people interviewers? Like, what do I even ask these people?” she says.

” As organizations increasingly rely on AI for HR functions, establishing ethical frameworks becomes critical ”

The need for ethical frameworks

As organizations increasingly rely on AI for HR functions, establishing ethical frameworks becomes critical — particularly when it comes to handling sensitive employee data.

“Organizations would be well suited to have an AI framework in place, ethical AI framework in place,” Develin says. “It should include things like transparency, [safety, and fairness].”

Yet many organizations are still playing catch-up in this area. “The only blind spot that concerns me is that when I ask leaders whether or not they have an ethical AI framework, most of them say, ‘No, we don’t have that yet,’” Develin adds.

Haydon has seen this gap as well, noting that her enterprise clients are only beginning to develop rigorous data governance requirements for vendors.

“There’s limited rigor around our security, data management practices,” Haydon says. “We’re trying to make sure that we’re on top of it, because we know as they start to mature and catch up with the technology, that will be increasingly one of the vendor screening requirements.”

Addressing bias concerns

A significant concern with AI in recruitment is algorithmic bias, which can perpetuate existing biases in hiring.

Develin said it’s important to understand how AI models are trained. “The AI tool that you’re utilizing is only as good as the data that it’s been trained on,” she says. “There’s always going to be human intervention needed to ensure that the AI tools are doing what we need them to do. Just relying on AI alone to make decisions on hiring, especially, I don’t think is the best practice.”

Reimagining onboarding through AI

Beyond recruitment, Haydon is exploring how AI can enhance the onboarding experience through what she calls “relational onboarding.”

It goes beyond the typical onboarding process that focuses on paperwork, policies, and procedures and prioritizes creating a supportive and positive environment where new hires can build connections with their team and other employees.

This approach uses AI to create personalized conversation starters based on a new hire’s profile for managers and team members so everyone can build more meaningful connections from the outset.

Some conversation starters, Haydon suggests, include: “This is how I make decisions. This is my appetite with risk. This is how I like to control. This is how I delegate. Here’s what we’re going to work well on, here’s where our tensions [may] arise.”

Develin says achieving the right balance with AI in HR requires one key thing: Thoughtful implementation.

“When we deal only in the black and white, we forget that employees aren’t going to remember what you say. They’re going to remember how you made them feel,” Develin says. “AI is transactional, not relational. There’s no replacing that human interaction, which is why AI tools should be intentionally implemented to enhance the human experience, not replace it.”

What are your thoughts on