UK 
United States
It’s probably no exaggeration to say that agentic AI has hit an all-important tipping point in CPOs’ consciousness. For many it’s seen as offering lots of real benefits (data processing, analytics, Q&As – all the stuff that slows HR down).
But for plenty more (and thankfully so, some HRDs might privately think), there’s also the view is presents lots of drawbacks – typically that it shouldn’t be allowed to perform full start-to-finish tasks that they still believe require ‘human’ oversight.
But let us introduce you to Brian Jabarian. He wants to blow this theory apart.
Jabarian, Principal Research and Affiliate at the Center for Applied AI Research at The University of Chicago Booth School of Business, could be a name that in the years to come, could be referred to as amongst the first of a new breed of academics who really did start to prove that ‘AI Agents’ could do HR tasks ‘better’ than humans, paving the way for more mass deployment.
He’s recently finished a paper, based on his own research, analysing the hired outcomes from more than 70,000 real, telephone-based job interviews for customer service roles – some of which were interactions with real human recruiters, while others were with applicants ‘speaking’ down the phone to an AI agent (‘Anna’).
‘Anna’ was a conversationally-trained AI agent, capable of responding (in real-time), and had the same required questions ‘she’ needed to get through to determine whether a candidates’ suitability to the role. Candidates could choose who they spoke to for their job interview, and if they chose ‘Anna’ they knew upfront it was an AI voice agent. Crucially, Anna would then be able to determine if the person she interviewed should be hired or not. And get this: Not only did a staggering 78% of applicants chose to be interviewed by ‘Anna’ rather than a human (destroying the myth that people distrust AI), but the additional results of the experiment were – not to put too fine a point on it – monumental.
“My own bet was that the AI would fail,” says Jabarian honestly, speaking exclusively to HR Grapevine. “I felt decisions like this couldn’t be delegated to a machine. But what I was surprised to find was that Anna outperformed traditional hiring across nearly every hiring metric. Interviews conducted with the AI agent led to 12% more job offers, 18% more job starters, and 16% higher retention rates after 30 days of employment – the metric that the test employer we partnered with had as their definition of a successful hire.”
At this point, it’s probably worth pausing to let the magnitude of this actually sink in. You read correctly. The AI agent – who, by the way, was analysing responses based on things like the complexity of vocabulary used; conversational style, openness, and general engagement (all things the hiring company value) – did a better job at assessing people than humans did. If this isn’t validation of the direction AI Agents could take, it’s hard to know what is.
“What was interesting to observe from the AI-agent interactions was that the composition of the calls themselves were different,” says Jabarian. “They were usually longer, and people were more expressive – perhaps because they knew they weren’t getting judged by any human biases.”
According to Jabarian, it took him four years just to find an employer willing to subject itself to this experiment. But he argues that with results like this, it’s now just a matter of time before more widespread adoption could become mainstream – and when this comes, he thinks the moment will be pivotal.
“Oh, it’s going to get bloody!” he says. “In recruiters’ minds they see their jobs as being interviewers and evaluators, and they don’t want to be replaced. This clearly shows boundaries are going to need to be redefined. AI may not be perfect, but neither are humans. We’re proving the point AI can do this stuff. It’s going to be a bumpy road ahead.”
