AI in recruitment, especially generative AI, is a curiously polarising concept. Talent acquisition practitioners either embrace it enthusiastically, or retreat from the idea with a noticeable lack of excitement. Some talent acquisition leaders have told People Matters that they would rather spend their budget on hiring more human recruiters to vet resumes and assess candidates, than invest in an AI-based system to do the same thing.

But why such a drastic difference in adoption?

It comes down to the original data analytics principle: garbage in, garbage out.

"If you're training AI models on data that is not complete, or in some cases not accurate, or in some cases biased, then you might get predictions out of that that are a complete mess," says Raghu Prasad, Regional Vice President at Workday. Speaking on the sidelines at Workday's Elevate 2025 last month, he told People Matters that recruiters who want to utilise AI need to give the model the specific context of the organisation: the role and skills, the function where the role is open, the business needs and how the function fits into the business.

This, of course, is where many organisations will hit a stumbling block. They may not have their own internal tool that can be trained in such a way. They may have policies in place around security and confidentiality that restrict what kind of and how much data can be used to train the model. Not all of their data may be available for the people working on the model, or in a form suitable for use.

How do you actually use AI, then?

The answer isn't to reject the tool completely, of course. Just as with any other function that could potentially be augmented with AI, the competition is other companies' recruitment teams that have figured out how to use AI.

"If an organisation can do this accurately, the value realisation can be very high," Prasad explains. He puts it in numbers: to hire five persons, you have to look through 100 resumes, and then you have to shortlist the best fits, move them through the evaluation process, negotiate a suitable compensation, and finally come to the actual hiring.

"There are multiple steps, and at each step there could be delays. If you could get a 20% to 30% reduction in the total time spent getting the right candidate, that's a huge value add."

He shared Workday's own approach to hiring, which he describes as skills-based and three-dimensional: "We look at it from the point of view of the core skills that we need for that particular function. Let's say you're in a sales function, maybe pre-sales. You will need product skills - learning about the product. Second, you will need domain skills, in terms of architecting solutions and learning about the industry. Third, you will need soft skills."

When a Workday team goes through the hiring process, he said, they look for candidates' ability to demonstrate the skills along all these different domains. Soft skills - now called power skills, and referring to influencing, credibility, growth, mindset, and agility in learning - are a major area of emphasis, and are thoroughly tested. 

That is where AI comes in, he explained: "When you have multiple points of checks, some of them AI-driven, some of them human-driven, you can get a reasonably good idea of whether a person has those. Whether mindset skills, influencing skills, or credibility skills and so on."

Clean data, judicious implementation

Recruiters who are enthusiastic about AI have most likely been able to take advantage of well-curated organisational data and a clear implementation strategy. There is also an element of embedding AI capabilities in existing tools, allowing users to self-service - essentially bringing information to their attention without them having to actively search for it.

Successfully using AI in recruitment, in other words, is not about replacing talent acquisition professionals, nor is it about relying on the tool. It may not even be about offloading the manual work such as screening resumes - that may be the most common application, but it is also the one that recruiters say tends to produce the most errors.

Instead, a good application of AI in recruitment will most likely involve using it to back up and substantiate human judgement, or occasionally to help humans rethink their judgement. Very much like other recommended ways of using the technology.